Or...wireless peripherals have always been a stupid idea. Battery powered devices are a compromise and the whole point thing about them is that you can take it with you and use it where there is no external power available. Using battery powered peripherals with a desktop computer defeats the purpose and you are stuck with all of the downsides of battery power.
Even more impressive than Linus's dedication to overcomplicating things until they stop working is his ability to turn every home DIY project into a business expense.
@@billmiller4800 The problem wasn't voltage sag,* it's the fact that the voltage drop across a resistor varies according to how much current is being drawn - these little sensors vary a lot in their current draw, so the voltage keeps going up and down. Dan didn't really fully explain what he did at the end but they just replaced the regulator followed by resistor setup with 3 power diodes in series - the voltage drop across power diodes is much more stable and almost independent of how much current is being drawn in most real world circumstances, so there is no longer any voltage sag. *Their 500mA USB supply can very easily supply the 30mA peak current the device is asking for
This sounds like it'd be a good PCB way project. Get a button cell sized PCB with exposed contacts so you can just drop the board in like the button cell, use the same surface mount components as before, and add a female USB-C connector. Dremel out a slot in the back cover for the connector, and you're golden.
How hard is it to design a PCB like that? Is there any straightforward way of doing that, or is it basically the electronic hellscape version of learning CAD?
@@TrolloTV As you already have the component lists and how to wire them, so almost nothing to think about, it should be pretty easy. Honnestly, if you run KiCAD, with a good tutorial about it on the side, along with the PCB manufcaturers requirement for them to make your PCB, it shouldn't take more than 2 hours.
Love that Dan used the Diode volt drop trick! For anybody wondering why that works; diodes are typically used to stop current flow in one direction, but do incur a small voltage drop on the "open side" as well. Because the difference between 5v to 3.2 is relatively small, Dan was able to exploit the inherent voltage drop of the diodes by putting a few in series to get them where they needed to be. So cool!
I am not an electrical engineer, but wouldn't resistors be just the same except you could do it with more control since the resistors are rated for a specific resistance?
@@jacobmalloy8384 problem with voltage divider is, that it changes voltage depending on load - unless you burn several times the load current on them. Voltage drop on the diode is more on less stable 0.7V.
@@armLocalhost Wouldn't resistors limit the current severely? A 5 Volt source could only ever provide up to 2.5 milliamp over a 2k resistor, right? That would not be enough for the 7 milliamp spikes
Gotta love Dan! Has really come into his own with his presenter energy. Woulda loved to have seen this done as a collab with him at the House, like the Linus&Jake videos 😄
You'd almost certainly be fine to just throw 3.3v strait into it. Depending on the linear regulator IC you're using, you may also need a load capacitor on the output for it to deliver a stable voltage - the datasheet will specify what value that should be if it's required. I'd rate the diode solution as an acceptable solution for this low current application too, although as shown the output isn't very stable across different loads because diodes are non-linear devices.
@@null6343 Given the device needs to use such low current, I suspect it won't have a linear regulator in it as they're inherently not very efficient at all. It's quite likely just got an STM32 processor of some type, powered directly from the battery.
Considering batteries don't hold the voltage for all their lifetime, I guess anything from maybe 2.8 to 3.5V should be fine on the sensors. The sensitive parts will have a regulator anyway.
@@Mulakulu Buck converter is way overkill. Linear regulator is fine for this. At these current draws, any sort of switching regulator will use more power just to run than the load uses.
I mean, It's not hard to wire up a wall wort to a battery lug(point of the video in fact). You can even find 3.3v ones and not need a voltage regulator.
Dan is great and the attitude of using what you have is also very positive. I often postpone a project, thinking I'll pick up the right parts next weeks, then 5 months later, I forget what the complications were and have to start from scratch.
I realise you guys were trying to do this with what was in-house, but you could totally make a tiny pcb the size of the button cell and put something like a SMT LT3470 on it with its associated components. Drop the voltage down to around 3v and still be able to compensate for draw spikes. I imagine the device goes to sleep for twenty minutes and then wakes up to measure the temperature and report in. That'll be what the sudden current spikes are and how the unit lasts so long on a button cell.
Yeah, what crazy prime said. Design it and maybe send one to LTT maybe they would provide the specs. Maybe with the appropriate disclaimers. Then the rest of us could make them or have them made as needed.
Yeah, I'm thinking you could fit a regulator, a cap or two, and a USB socket easily on a coin cell sized PCB. Add a 3d printed cover and you're featured on hackaday.
This whole smart home series just made me more sure that right now having a dumb home is many times more painless Edit: I do of course understand that making a smart home is unspeakably less difficult if you're not Linus Sebastian and that Linus really loves making his life hell to have everything his way
No just don’t do it the way Linus is doing it. He’s trying to get everything the EXACT way he wants it rather than going with the closest available product with his desired features and accepting the downsides
You can do smart home without all that pain. Linus has a way of over-complicating things... The automated garage door, smart light switches, and now this heating stuff. My home also has all those things and they work pretty much all the time. And the few times there are hiccups, I've set things up in such a way that everything keeps working like it would in a dumb home. It takes some more planning and research ahead of time, but that's better than buying a system only to realize it doesn't satisfy your needs and trying to solder your way out of it.
Linus is definitely overcomplicating a lot of it. Their even a way to achieve this using Home Assistance with off the shelf components that are wired. Certain things like going ONLY Zwave cut off a ton of products. Then their likely definitely a better way than goin Ecobee. As some1 with Home Assistant showing over 50 devices, I can attest that it can be elaborate w/o the complication.
@@Niosus I'd say he isn't really overcomplicating it, rather he's gone down the patchwork route. I love my Home Assistant. It's reliable, works well and is easy to use by everyone in my house. Linus has done the same thing many of us starting out with home automation have done. Buy something. Try to make it work. Buy another thing to try and make the first thing work. Along the way you find out some new cool functionality that is now possible because you got the two first things, but you just need to buy this 3rd extra thing etc. That's how I got started too, but now I'm real boring so I if I'm adding something new, I first make a list of requirements, do some research to see what the most efficient way is to get those requirements met and only then do I click the "buy" button.
@Bingus That's what makes it fun though right? It's not like there's a school for home automation. You start out with an interest and just learn the required skills as you need them. Obviously in this case they take some shortcuts because they have to make content, but the process and the frustration is pretty similar to what most of us have experienced when working on these things and trying to get it to do what we want.
Dropping what is arguably Adam Savage's most recognisable quote right before referencing him directly is genius level writing and I love whoever wrote it.
The intro perfectly resonates with me. Not everything needs to be wireless. Some things are better with wire. I did a mod myself at some point. Turned a battery-powered bluetooth speaker into a wired one. It now runs at 5V USB. Works just as well and the battery still exists just in case. Highly recommended for everyone to look into the classic LM317 voltage regulator.
Someone needs to make a USB to fake button cell adapter with adjustable arms to simulate the different sizes of buttons. $5-10 each would be well worth less jank.
@@ccibinel the issue with that is that USB runs on 5V and most common button cells on either 1.5v or 3v, so some voltage adapter would need to be added as well.
@@HappyBeezerStudios AMS1117-3.3 (or others with different outputs) will do it and costs under $1 us. Messing with resistors to save 30 cents is dumb. If someone made a product combining this with a housing, arms and usb port it should be viable for $5-8
@@ccibinel Right, fixed voltage is all that is needed, since the batteries that are to be replaced are also fixed. And much more comfortable than a bunch of LM317
Better to use both I think? drop the voltage a little with a schottky or two and then a zener with a breakdown of 3.2v to lock it in. That should limit the wasted power and still be able to provide the spikes. Add a capacitor if it the voltage droops too much during the high load.
Or just 3 normal Diodes in series ,would be 5-(3×0,7)=2,9V that shouldbe find if not a basic 3.3 regulator with a little cap is the best option in my opinion
@@leobla744 Obviously 3.3v regulator would be better lol, not the point of this thought experiment at all. Issue with diodes in series is your peak voltage tends to be inversely proportional to the load, it's only marginally better than using a resistor as a voltage limiter. Using a zener isn't a great fix, but from my limited experience making power supply circuits, it's the best you can do without a timer + MOSFET circuit, or linear VRM (or either of these in packages).
I’ve probably been to the shop about 10 times last year just to buy a random battery because I didnt have a specific battery some random thing needed.. I feel u, Linus.
If you can wait a few days, ebay is a good source. Buy a 10 (or more) pack, because with the shipping it works out about the same as buying one. Then chuck the extras into a drawer, completely unlabelled, and next time you need one you get to spend ages searching through your battery drawer. That's my system anyway ;)
@@phuzz00 it is very obvious thing Linus pointed out the batteries get old batteries leak lithium batteries do good old and leak Oracle of old expired leaking batteries
@@phuzz00 but I understand the logic behind a large quantity from eBay or Amazon for cheaper also AliExpress or Alibaba all those the shipping times very
In all the cases, resister, diodes, Zener, and regulator, you need a load resistor (and a small capacitor across the load - 1uf) to get the minimum load current to get the diodes to be forward bias, Zener or regulator to regulate. With out the load resister, at very low current draw, none of the methods above will turn on or act like you would expect them to do. For example, the MIC5504 needs a minimum of 100uA to regulate (3v @ 100uA would need 30k load resister - do not forget the load capacitor).
@@Stanton_High what makes you think he doesn't respect Li-Ion battery safety? Could it be the soldering of 18650 cells to make battery packs? for a RED Camera i believe. Lol, I cant remember but something went wrong in that video, he either vented a cell or just never got it working cause of a circuit board regulator or something.
@@Frostbite1090 nah. the circuit board only had volatile memory. cause hey it's attached to a battery so if it ever doesn't have power the pack is dead and unrepairable anyway
@@Frostbite1090 that's one of those "you really shouldnt do it ever but if you genuinely don't have another option, you gotta" kinda things bad example but realistic execution imo
For those bumps on power draw add a capacitor in parallel with your device, it'll smooth out the power draw issue and probably make it a little better.
@@bosstowndynamics5488 I'd still put a cap on the output of that circuit as well. It's replacing a battery, you want that output voltage to be smooth with as little ripple as possible and come as close to a battery characteristics as possible.
@@Mark-pm8nw But I would do all the voltage regulation outside of the sensor and near the power brick since they might get hot thus making the effort null
Hello LTT. I know you never read this but here it goes. A regulator of a normal size (for soldering) like a HT7533-1 that you can buy for 30 cents will serve you fine GND goes to ground IN goes to 5v OUT goes to 3.3v It is recommended to add a couple of capacitor in the IN and OUT for stability but are "optional" 10uF is more than ok (each would cost 10 cents) And you are done. You can pull up to 100mA and if you want more the LM1117T has you covered with up to 1.5A or if you feel like an engineer a transistor would also give you extra power a 2N3904 would provide you with extra 200mA, those are 15 cents and you need an extra resistor. Anyway...on a buget you can do just fine with 50 cents (gets cheaper by the 10s and 100s) For 2$ you can have lots of power and even make yourself a nice small PCB and make 20 of them (which will reduce cost) and power EVERYTHING of a cable. If you think it over a bit there are version of those regulators that if you have even more powerfull power supplies (over 5v) you can even replace devices that use two AA or two AAA batteries. Then again.. (and sorry for this since I live in europe and the plugs are specially painful here). having 20 more cables and european plugs around the house is a nightmare
I think it would be cool to make a little circular pcb that can just hold the voltage regulator and make direct contact with the contacts on the sensor. That seems a lot less jank than this lol
lol. I just came to comment the same thing. if you just design a pcb with detachable cable, and order it from china assembled, you can just replace any button cell device and call it a day. durable as well. but then no content. :3
@@minolhewage Wouldn't be surprised if there is (or will be soon) an open-source (set of) design(s) to replace the battery for various button cell devices
I love seeing you guys dive into electronics. If you do a project like this again - a 3.3v regular and an electrolytic capacitor in parallel across the output should run any device like this. The capacitor absorbs those spikes. Most circuits with linear regulators follow that approach
This, plus don't even bother with the potentiometer. The circuit is most likely 3.3v which most microcontrollers are today, so leaving it at 3.3v is just fine.
@@billmiller4800 By my guesstimation the linear regulator will waste $0.03 in electricity to save $3.00 for a battery. Not counting all the other costs, losses, and benefits.
I don't know why everyone is saying to add a capacitor for those "spikes", those are 7mA spikes on a device being powered through what in practice would almost certainly be a half to one amp regulator on a half amp or more USB supply, you aren't going to get voltage drops in that configuration (after all, the 3 diode solution they went with is functionally a dirty voltage regulator and didn't cause any voltage drops during the current spikes)
In fairness, I have 4 of those ecobee sensors and they’re well into 3+ years without ever changing the batteries. I’ve been extremely impressed with how efficient they are. I upgraded to an Ecobee in November of 2019 and bought the sensors that December. I’d gladly hardwire any of my other wireless smart devices, as Ecobee aside, everything else absolutely chews through batteries
Same with my Philips Hue switches and sensors. Three and a half years in and I’m just now seeing ‘Battery low’ for the first time. This problem does not exist lol.
@@heisenbaj Now take those 3 years, divide them by 100+ devices, switches, sensors like Linus has in his giant house and given enaugh time it would average out to 1 devices every ~10 days because not every battery lasts the same time.
@@gamingmarcus Usually it doesn't really average out all that well because similar devices (say 10 temp sensors of the same model) use the same amount of battery and batteries where put in there roughly at the same time. This means you gotta change a whole lot of batteries at once. And I would assume he does not have 100+ wireless room temperature sensors, the house is not that big. A more reasonable solution would be to make or buy some sort of energy harvesting solution that maybe uses a solar cell. These devices pull miniscule amounts of power if you calculate it and it should be doable. And you could keep them wireless.
@@gamingmarcus Sure, and so I get wiring everything possible. But these things are bought in waves or all at once, so it wouldn’t be nearly as sporadic as 1 every 10 days. This also introduces a failure point and I’m guessing voids warranty so solution seems worse than the problem.
It's probably fine at 3.3v. you could use a 3.3v zener diode in parallel with the device and then a single resistor in series beforehand to make a simple diy regulator. You just gotta make sure you spec your components right so you don't fry anything.
Yeah~. A potentiometer is not the way to regulate voltage to an active circuit. Much less anything radio. I'd even bet the first thing inside that sensor is a regulator boosting battery voltage to a standard 3.3V.
Yeah, that's the proper solution, but they probably didn't have any Zeners lying around. If you're ordering components anyways, that's way better than making a giant chain of diodes.
@@anlumo1 the series diode one isn't a bad solution. They have a generally stable voltage drop especially if your doesnt suck much current. With a general voltage drop of about 0.6 volts you could put 6 in series and maybe an extra shotky diode to make the 2 volt drop
@@RandornCanis Good point. Examining the actual power circuit can take a lot of guess work out of it. Sometimes electronics accept wider voltage ranges than you'd expect.
The funniest thing about this whole endeavour is when Linus at 15:00 implies that the sensor would likely run for years on its button battery lol. The truest definition of over-engineering right there! Also I think I'd take the battery over using up a plug socket :P
Agreed, I saw this being a problem from a mile away. The advantage these cells (and batteries) have is that they're effectively really large capacitors. They will refuse to drop the voltage significantly unless you draw some high loads (>100mA). If you're replacing batteries this is one of the first things you address lest you get "funnied" by dropping voltages. Wall chargers are NOT designed to handle sudden spikes.
I’ve been running one of those sensors for three years and have yet to need a new battery. While it would be nice to plug it in, ecobee has done a really good job keeping battery changes to a minimum.
glad im not the only one who noticed how good Dan has gotten on camera. not that he was bad before, but you can see the improvement and how comfortable he's gotten. i love anything involving Dan!
@@armLocalhost glad im not crazy. for a beginner like me, I have no idea why using diodes as resistors works. I know diodes typically have some resistance, but I don't know whether they have to be special diodes in this case?
The lack of capacitors is probably why the LDO / MIC5504 didn't work. For further reference, if interested in electronics... The MIC5504 (or any other LDO) requires input and output capacitors to function properly, these are called bypass capacitors and aid with stabilizing the voltages during random current spikes. The datasheet for the MIC5504 recommends 1µF at both input and output side, but it wouldn't hurt to oversize it with a 2.2µF capacitor. An LDO lowers voltage by dissipating the surplus energy as heat.Thus the heat dissipation can easily be calculated by subtracting the output wattage from te input wattage. Lets for example estimate the average output current to be 10mA. - The output wattage at 3 volts would be 3V*0.010A = 0.030 Watts (Or 30 mW) - The input wattage at 5 volts would be 5V*0.010A = 0.050 Watts (Or 50 mW) - The heat dissipation would be 50mW-30mW = 20mW ... The other 30mW is used to power the device. - Next we need to check the LDO's datasheet to get the temperature increase per dissipated watt of energy, usually labeled as °C/Watt. - Our MIC5504 has a temperature increase of about 250°C per watt. We only dissipate 0.02 watts of heat, thus we multiply 250°C by 0.02mW = 5°C temperature increase. - Thus at an ambient room temperature of 20°C we can expect the LDO to reach a max temperature of around 25°C. Which is below the maximum rated temperature of our LDO. * Note: - The lack of a pcb/circuitboard as cooling plate could cause slightly higher temperatures in reality. - I always try to keep an LDO below 70°C tops, even when the datasheet states 150°C. - Look into a buck converter when larger voltage drops are required! These won't dissipate as much heat, but are more complex and bit more expensive.
I have to admit, that SMD soldering was pretty impressive considering the fact that Linus was in such a sub-optimal environment for doing it. I've done some amount of SMD soldering though I tend to avoid it if I can because it can be fairly annoying (especially if you're talking 0402s).
Agreed, I think a better solution would be to use all through holes and get some very small breadboards. Using SMD is asking for trouble lol. If breadboards cost too much you can get these prototyping breadboards as well (these do require soldering though).
Two points to make. The battery voltage will vary above 3.3V when full so there should be no issue with a 3.3v regulator. Also for those larger current spikes a simple electrolytic capacitor for a few cents would help greatly.
I don't understand why they didn't ask a proper engineer which they have in the team and just over engineered it like hell. Like they could have used a voltage divider with a cap and call it a day.
@@armLocalhost could have been way over engineered but their point was to do it on the cheap. But I would say stay clear of voltage divider with no protections at all.
@@nadie9058 just to be bulletproof, I wouldn't use a shunt regulator here, they don't work well for bursty loads like this one, and are super inefficient. This was because he wanted 3.2v rather than 3.3. With jlcpcb being as cheap as it is, I probably would have just made up a pcb in kicad quickly with all the components and had them solder it together. At the same time, I am suuuuper surprised they don't have any 317s in a to92 there. That and a couple caps would solve the problem imediately.
I love the new format!! There is a commentary from Dan and explaining how it works and alternatives. Keep this up. And include the beep part on the edit. It feels more natural to me ateast
This is more than just "First world problems" this is another league: "Linus' house problems". I swear this guy has these super particular problems he needs to solve and make content of.
He made a problem for himself. At least with those sensors. They way underestimate with that 2 year life span. I've had some going for 3 or 4 years now without needing to change the battery.
I am an electronics engineering student, never have I though about using diodes voltage drop to drop down regulated voltages, this is really smart and I will use it whenever i need to do something the "fast and junky way". BTW, you could used a rezistor parallel to the ecobee terminals to keep the diodes in the 0.6 voltage drop range . this would increase current draw (but the powersuply still draws some amount of power so it does not matter) but it will allow you to use this trick on devices with overvoltage protection
It's an ancient hack, but relies too much on current draw. I usually can't use it because if it's that low voltage, it's also low-power and thus usually very "peaky" in current draw, and I can't justify pissing away milliamps in a device designed to take microamps on average. So I never design it into a device. And when hacking an initial prototype together, I just use an adjustable buck converter module or an adjustable LDO regulator. Or feed it off a power supply. So the utility is limited. The main issue with hacks like that is that you introduce unknowns, so there's that much more to exclude as root cause if something doesn't work.
I'm a EE who mainly deals in railway signalling, but I did a lot of power engineering too. Be it small or big. You can use a zener diode to pull a voltage to a known value, you'll learn about it later when doing courses about designing power supplies. A diode and a transistor are the basis for a voltage regulator. What happened here is he used a resistor devider but failed to consider that the ecobee pulls power too, thus acting like an impedance and influencing the voltage drop. Could have been solved with an op-amp in VF config. THB though i'd gone the simple route and grabbed a mini buck converter from ali express, they are tiny now and mitigate the hassle. No point re inventing the wheel when an absoulty perfect solution exsists.
@@robmckennie4203 Ah but you are then dependant on the temperature. I can remember while doing my EE courses, the actual math behind SC. An absoult pain, was the only bit I didn't enjoy.
I recommend to get some prototype PCB boards. They are available with the Through hole pitch and the SMD too. You can easily cut them to smaller pieces, and they provide a nice connection to your prototype device.
@@nezu_cc Because the amount of people that saw that section of a stream is a whole lower than the amount of people watching video's. Especially considering there's going to be repeated videos in his house.
@@SolutionJW yeah, but information spreads fast. If someone wants to know his location all they need to do is google it. Imo there is no point hiding it if you can just google it. Nobody is going to play geo guesser using his windows if the info is already out there in the wild and confirmed to be correct in multiple ways.
@@nezu_cc My point is more along the lines off 1) there's no point in NOT hiding it and 2) people that aren't already looking for the information might get interested because they see something interesting outside his windows. Why give them that opportunity, right? Besides, hiding his location might not be the sole reason for the blur, who knows
@@SolutionJW as others pointed out, sometimes is batter to hide in plain sight. Pixelation only draws unnecessary attention to it. It's especially pointless once the original purpose behind it is already gone. By this point, it does nothing else expect waisting editors time and drawing unnecessary attention.
Really love the home automation/improvement videos and glad you guys let Dan have a segment in to help understand things better. I would love to see more detailed content on this stuff in the future or something as a follow up to this video with some of the viewers suggested comments. I too hate batteries and wish to keep everything hardlined so seeing that it’s something that is doable has at least opened a door into further research for me on this topic
These videos just keep getting better and better... I've been watching for like 5 years and because of LTT I've built 3 computers and now solder my joysticks and retro systems. I would never even try something like that before. Ty Linus
Rather than soldering to the contacts in the device, cut a piece of plastic (e.g. bottle top or two, or even a bit of wood) to the right size and then stick the wires to it. This gives the added benefit of being able to convert back to a battery if needed.
Basically a plug the size of a CR2032 (THe name itself designates it as a lithium cell with 3V nominal and 2.0V end point with 20 mm diameter and 3.2mm height. The number is the trick: 2032 -> 20x3.2 or the similar 2025 -> 20x2.5 or even the 18650 -> 18x65.0 that are used in many laptop batteries)
Dan is my hero. My batman. Someone I didnt know I needed in my life but now everytime I see him in these videos it sparks me so much joy. He is so funny and has so much knowledge, that he even stands out next to so many other great people at LTT.
I'd advise against a DHT11 for something used to control the room heating as even a degree change can make the room super cozy vs uncomfortably cold, and the DHT11 is pretty inaccurate at ±2 degrees celsius. A DHT22 is a fraction pricier but much more accurate, and if you want to get even better readings( temp+humidity+air pressure), there's the BME280 I think it was called. But yes, the 2-pack of Linus sensors cost a whole 120 dollars, and the ESP is far far cheaper. I reckon if it's just a single sensor, they can also use the ESP01 as it is TINY.
@@namankohli1061 better would be the Sensirion (IIRC) SHT40, SHTC3, or SHT31 (in order of price), as they're I2C interfacing, not the weirdness of counting the pulses that a DHT11/22 uses. Less of a concern with a single purpose endpoint, but might as well avoid. Then again, I don't know how annoying those are to physically interface with if you don't get a dev board with them like Adafruit sells.
You can also use diodes in their forward direction to drop voltage. Like you could have used a schottky diode to drop ~0.2v off of the 3.3v supply. And that isn't dependent on current (ideally of course) like a resistor/pot is.
If you're doing a few, might have been worth drawing up a PCB and having JLCPCB or similar fab and SMT a tiny board the size of a battery to drop into the sensor that could have the voltage regulator and maybe the filter caps that are part of the reference implementation on the datasheet.
well... someone do it up and sent it to LTT to add in the description, there's bound to have lots of peeps wanting to do that... without the patience or tools to do it...
The fact that Linus is using surface mount components is hilarious, I am almost certain you could use a through hole LDO voltage regulator and have no soldering problems
Thank you to whoever edited in the fire hazard warning @ 6:04 regarding cheap USB A 5V 1/2A plugs. It’s become such a common problem now that even high street retailers like Poundland are selling these things and if you or anyone you know uses them, dispose of them straight away. It’s not a matter of if but WHEN they will either melt or explode which will start a fire if it’s unattended.
1:48 In my experience most things use the CR20 batteries which describe their diameter and thickness in their name. A CR2025 is 20mm in diameter and 2,5mm thick, depending on how the Battery is held in it can also serve as a replacement for a CR2030 battery.
Linear 3.3v regulator plus a small capacitor on input and output would be my approach to this - no need to faff about with a potentiometer, and the caps should soak up the power spikes reasonably well.
To simplify, just use the LM2596S DC-DC Buck Converter (2 wires in DC1.25V-35V - 2 tires out adjustable voltage) it only costs $1.96 free shipping from Aliexpress, simple and easy!
And here I thought this was actually a helpful tutorial on how to convert battery operated devices to hardwired, for people who aren't electronics experts, turns out they just needed to publish a video per the schedule. LTT is more of an entertainment channel than actual tech tips.
Most frustrating part is how needlessly rushed and botched it was. The amount of prep they did seemed appropriate for an emergency video they need to do _this afternoon_ to put out _something._ But from Dan's comment, it's clear it was in the pipeline for at least a week. They could have picked a proper 3V LDO and designed a CR2477-sized PCB around it (possibly even with a USB connector on the back) in 15 minutes (or a couple hours tops, if Dan doesn't have experience with it) and have it fabbed - including population - and delivered in ~3 days. They had the time and they certainly have the money.
I recommend getting some non-SMD sized (TO92) voltage regulators in various voltages for future "unplanned" projects. And get some transistors (BC547C, BC557C) as well while you're at it :) An inventory of various basic components is worth its weight in gold.
agree with that, I converted a 2x AAA battery wireless headset base to a Li-po powered and used a MCP1703 in TO92 package, never failled onced! it's a 3V fixed voltage so perfect for tha kind of application but if I remeber they offer other voltages
@@vappyreon1176 working on cars with a wireless headset is way better than wired or when I work it's plug in the computer but I can move without walking on the wire or have a 10m wire... the headset it self work for 3-4 days on 2x AAA but the base was like a day and half and I bought rechargeable when I got it, still the smae 4x batteries in 7 years. I broke the base case one day that was another reason for the switch and use a MAX1811 for charging the lipo
@@ItsJust2SXTs man, I would never trust putting a lipo that I had anything to do with that close to my head. I still remember the old olight lipo flashlight exploding in mouth thing.
@@marcusborderlands6177 me either that's why it the base not the headset it self, it's a 500mah battery so it's not that big. It transmit the sound in the FM band not bluetooth. When the lipo will die eventually(it's 5 years old now so it lost some charge but it's not buldge), I will change for a Li-ion at least they are in metal casing. Bluetooth Headset use li-po but still too close for my tase. Remember the samsung that were blowing in pants? that was scary to see..
Seems like a perfect use case for OSH Park. A perfect purple PCB would be awesome. It could just fit perfectly into the button cell holder. Anyway... Capacitors are great for current spikes too.
For this kind of projects I use MP1584EN or LM2596 buck down converters. They can deliver about 3A tops with 4-40V input. You can buy them in 5 packs for about $10, much easier to solder and way safer. They are adjustable, the LM2596 uses a 10 turn pot so you can adjust it at precisely 3V. They hold the output steady at variable amps and variable input voltage. The MP1548EN is the smallest which will fit nearly anywhere.
It would be really cool if you guys made a drop in replacement board that just took a micro-usb cable or something. A small, 2-layer PCB would be able to fit the voltage regulator as well as include contacts to the existing battery contacts. And maybe an inductor or something could help with the current spike?
Thanks for this amazing guide, Linus! I will literally have no uses for it, but it was incredible to watch you turn a wireless thing wired. I am more than impressed.
The regulator spec sheet would tell you how to calculate the resistance needed to get your desired voltage. Than you can use small exact resistors vs large pot. You could probably buy a pre-made regulator on PCB for ~$5-10 and just swap the resistors to get the voltage you want. Than you can easily solder the wires on.
Even more, most of these chips have a freely adjustable version. Then you have to attach the potentiometer between GND and the output, and the middle pin to the feedback (usually called FB) pin. then you can set almost any output voltage to the device, that is independent of the load on the output.
the way he used the resistor is not to adjust the voltage regulator output, but to drop some more voltage in series. However as Dan explained, the current draw from the sensor varies wildly, therefore the voltage drop across the resistor does awsell (remember Ohm's Law V = I*R)
@@gabrielenitti3243 You NEVER use a resistor to try and drop voltage as a power supply, especially for dynamic loads . V = I*R, meaning every time the current draw changes the voltage changes, those devices need constant voltage. I don't recall specifics but Dan showed diodes to drop voltage which is fine, just not efficient.
At that point, you may as well use the LM2596 board and just smear it with hot glue. Yes, it'll use more power on idle since it's made for supplying up to 3A, but the size and cost makes it very well worth it. Or use literally any $0.5-$1 buck converter off eBay
The voltage drop caused by the potentiometer depends on the current flowing through it, so the voltage you are measuring will vary depending on the current power consumption of the sensor, which explains the over 4 volts you saw
The only proper way to do this is to get an adjustable regulator that lets you connect a potentiometer (or fixed voltage divider) to an adjustment pin that will actually change the regulation voltage. Assuming 3.3V is really too high there are 3.0V fixed regulators easily available (well, they do manufacture them at least...)
@@ailivac yup, that's it. I think 3.3 V would easily be in spec assuming a ±10% tolerance. And I think the vanilla battery is probably over 3.2 V anyway when full.
I find that voltage regulators are best used to get a very exact voltage, for example for an ADC's reference voltage. For this use, I'd prefer a tiny QSKJ buck converter.
My smoke detectors and security system are hardwired but have battery backups for when the power is out. The smoke detectors start beeping once every couple of years and I just replaced my security system battery for the first time in 8 years. It was like a small car battery.
I wired up a 2XAA battery holder which outputs 3 volts to my scale to replace it's original CR2032 and it's still going after a few years. If you don't care about the aesthetics I think it's a decent solution
Tip for next time: If you want to have a small DC power supply, you can use a Z diode and thus set the voltage to a fixed voltage. The structure looks like you have a resistor for the Z diode so that the current to the diode and the consumer is limited. The consumer is then connected parallel to the Z diode. The Z diode determines how high the constant voltage is that falls off via the diode. The Z diode is installed in the blocking direction. This makes it relatively easy to set up a small power supply.
That is what I was thinking. Why not get a zener diode and just be done with it? Or if that's too much hassle to get the right one, just use a good old LM317
I might be misunderstanding because I haven't used zeners myself before, but wouldn't shorting one across the supply output of a regulated voltage source (the 5V wall wart) cause the upstream active regulator to dump upwards of half an amp through it trying to push the voltage back up to 5V? That seems much more wasteful than an in line linear regulator.
@@bosstowndynamics5488 you’d add a resistor to the 5V output. That will limit the current. I’d guess the resistor would be in the range of 1k to 10k, but that can definitely change
@@bosstowndynamics5488 In this case you'd use a 3.3V zener diode. You'd put that in-line on the positive side. So you backfeed it with the +5V from the wall wart, into the +3.2V for the sensor. The +5V exceeds the zener diode's zener voltage. Which causes it to break down and start conducting at it's zener voltage of 3.3V. Add in a smoothing cap and a discharge resistor. Done. No need to fiddle with a pot to try and calibrate a regulator. As long as you don't try and draw more amps than the zener diode is rated for, it'll regulate the voltage.
10:13 something you could do for the future is find all of your QR codes for your various devices and export them as a large image file that is stored on your server. I have this done on my security cameras so that in the unilikely event i need to add them on a new phone or because i forgot how to rearrange them in the app UI, i can just import the QR code from a file rather than to go to each security camera. but this is only a solution if the software allows you to import a QR code. I guess another option is to open the QR code file on another device and point your app/camera at the other device
Battery eliminator are made to replace batteries by a power supply, it either comes with one integrated or you have to plug it on a corresponding power supply. It would have avoided the sketchy soldering with SMD.
I had a unit in my last two houses that would turn on the HVAC's central fan to circuit and average the air within the house. Some bedrooms would get super hot, or super cool, and this balanced the house's overall temp without cycling the heater or air conditioner as much.
Instead of the three regular diodes in series, you could've used a Zener Diode. For you application with a constant load, Zener Diode would be your best option
I want to see a video about battery tech. How good are they today? What types are there. Are there other kinds beside Lithium that could power a handheld?
outside of nimh batteries, not really. Like you have Lithium Ion which do 4.2 V, and the lithium cells that do 3 V but with super low discharge. NIMH do 1.2 V and Alkalines do 1.5 V. There are lithium ion AAs but they are pricy and need a special charger. Basically the best of both worlds
There are way to many battery types, to be honest. Most of them are/were either proprietary for cameras or for a specific size constrained applications.
Hey Dan, Linus - the voltage regulator solution with potentiometer is actually still almost right on the money. You've got a few realistic possibilities for keeping this cheap and fixing the problem with transient spikes in current draw ruining the voltage. The first, and probably easiest is just to toss a small capacitor in series with the voltage supply. This will smooth out those transient spikes and keep the device online. Since these devices are tiny and only pulling a few milliamps at peak, even a 1uf is probably sufficient, though I'd spring for 10uf. Also, anybody seeking to use the DIY approach should probably just buy a lot of 100. 100 caps is usually like $3, or $0.03ea, so if you make a mistake or need to do another device, no big deal. Also, there are ready-made solutions for this you can get straight from china for super cheap, so it might not be worth the effort. For example, with prime shipping you can get a pack of 10 3-12v input 3-12v output range 3A voltage regular PCBs. These are more robust - it's a pcb with surface mount components and nice through hole pads for input and output, and its a complete variable voltage regulator circuit with fuses, an output transistor, and capacitors for smoothing. A ten pack of them usually runs about eight US dollars, or about eighty cents to convert a device -- what Linus ended up with after accidentally breaking one of the regulator ICs.
Did Linus consider moving the thermostat to the other side of the wall? It would be significantly more reliable to move the thermostat instead of adding another device. If Linus just needs the thermostat on the back side of existing wall, he could drill another hole.
Yeah I don't get it, that house was just renovated and they added a lot of wiring. It was intentional to put the thermostat in the wrong spot... A lot of weird things have been done with that house.
@@cedricpomerleau5586 Don't forget Linus has been giving a crapton of instructions on what he wanted to contractors, and a lot of said instructions were different from what they would typically expect to do. Mistakes and/or confusion have happened a lot because of this.
he went from no wires setup to making wireless things wired. its either his villan arc or character development unmatched
what did the intel arc challenge do to him?
@Don't Read My Profile Picture ok
OR, hear me out
He doesnt care and he just wants to pump out content of varying qualities
Either you die a hero or live long enough to become the villain
Or...wireless peripherals have always been a stupid idea. Battery powered devices are a compromise and the whole point thing about them is that you can take it with you and use it where there is no external power available. Using battery powered peripherals with a desktop computer defeats the purpose and you are stuck with all of the downsides of battery power.
Even more impressive than Linus's dedication to overcomplicating things until they stop working is his ability to turn every home DIY project into a business expense.
I heard even cablemod melted by 4090 ugly 12 pin.
He's making money from the videos
He did start the channel from his home. He's just gone full circle.
@@psedog Just how many houses did Linus move in at this point?
It feels shameless at this point, and I don’t think I can kick it with this channel anymore. This hurts to watch
The highlight of this video is Dan's part. Now I wish there would be a library of videos of Dan explaining electeonics
Agreed, Dan's segment was really interesting and he's great on camera.
Same
He has great vibes. There's something really relaxing in just hearing him explain.
If you need more of a burst current, a smallish capacitor could help with voltage sag... but normally shouldn't be needed
@@billmiller4800 The problem wasn't voltage sag,* it's the fact that the voltage drop across a resistor varies according to how much current is being drawn - these little sensors vary a lot in their current draw, so the voltage keeps going up and down. Dan didn't really fully explain what he did at the end but they just replaced the regulator followed by resistor setup with 3 power diodes in series - the voltage drop across power diodes is much more stable and almost independent of how much current is being drawn in most real world circumstances, so there is no longer any voltage sag.
*Their 500mA USB supply can very easily supply the 30mA peak current the device is asking for
This sounds like it'd be a good PCB way project. Get a button cell sized PCB with exposed contacts so you can just drop the board in like the button cell, use the same surface mount components as before, and add a female USB-C connector. Dremel out a slot in the back cover for the connector, and you're golden.
yep - and it could be even sold as a conversion kit later on LTT store lol
How hard is it to design a PCB like that? Is there any straightforward way of doing that, or is it basically the electronic hellscape version of learning CAD?
Or just leave some open contacts for more flexibility. Drilling a round hole is easier than the rounded slot needed for usb c.
Also add a switch to choose the input voltage (5 volts, 12 volts etc).
@@TrolloTV As you already have the component lists and how to wire them, so almost nothing to think about, it should be pretty easy. Honnestly, if you run KiCAD, with a good tutorial about it on the side, along with the PCB manufcaturers requirement for them to make your PCB, it shouldn't take more than 2 hours.
God the editors are so impressive, and Dan needs to be a full time presenter, love seeing him in the videos!
Ok Dan has gotten SO much more comfortable on the camera. He's great. Awesome addition to the LTT team!
I think doing the wan-show for a while as a 3rd wheel has helped him
I love seeing Dan. He's so calm yet chaotic and explains things well.
Love that Dan used the Diode volt drop trick! For anybody wondering why that works; diodes are typically used to stop current flow in one direction, but do incur a small voltage drop on the "open side" as well. Because the difference between 5v to 3.2 is relatively small, Dan was able to exploit the inherent voltage drop of the diodes by putting a few in series to get them where they needed to be. So cool!
If you are going to do a hacky solution a voltage divider using resistors would be better.
@@backgammonbacon my thoughts as well. Instead of using two resistors (like 2k/4k) they over engineered with a linear voltage regulator.
I am not an electrical engineer, but wouldn't resistors be just the same except you could do it with more control since the resistors are rated for a specific resistance?
@@jacobmalloy8384 problem with voltage divider is, that it changes voltage depending on load - unless you burn several times the load current on them. Voltage drop on the diode is more on less stable 0.7V.
@@armLocalhost Wouldn't resistors limit the current severely? A 5 Volt source could only ever provide up to 2.5 milliamp over a 2k resistor, right? That would not be enough for the 7 milliamp spikes
" I love Dan because of how awesome he is, I hate Dan because how he makes the rest of us look" Classic love Dan keep up the great work guys
Gotta love Dan! Has really come into his own with his presenter energy. Woulda loved to have seen this done as a collab with him at the House, like the Linus&Jake videos 😄
Man, Dan explaining electronics was like Bob Ross painting. Actual ASMR for me. I want more of him in videos.
Have you discovered BigClive?
@@daveedee3626 ooohhhhohohoho YES
true, we need dat bpi more
@@daveedee3626 what's bigclive? Faketaxi?
@@daveedee3626 I love when he says 'scematics'
You'd almost certainly be fine to just throw 3.3v strait into it. Depending on the linear regulator IC you're using, you may also need a load capacitor on the output for it to deliver a stable voltage - the datasheet will specify what value that should be if it's required. I'd rate the diode solution as an acceptable solution for this low current application too, although as shown the output isn't very stable across different loads because diodes are non-linear devices.
A fully charged button cell is already 3.3v.
Yup. Pretty sure a simple 3.3v power brick would’ve done the job. Higher chance to have that on hand than the ICs required.
I was nearly screaming at the video. How did they not know nominal isn't the max voltage of a cell.
@@null6343 Given the device needs to use such low current, I suspect it won't have a linear regulator in it as they're inherently not very efficient at all. It's quite likely just got an STM32 processor of some type, powered directly from the battery.
Considering batteries don't hold the voltage for all their lifetime, I guess anything from maybe 2.8 to 3.5V should be fine on the sensors. The sensitive parts will have a regulator anyway.
Side note: You can get a converter supply module that will convert 5v or 12v to any smaller voltage for about a dollar without any extra steps.
Yepp. They're called Buck converters. Boost converter to turn the voltage up. They're also way more efficient than the voltage regulators he used.
Exactly but it is too big to fit into the device I think this is why decide to go in this particular way
I was thinking he was going to use one until he pulled out the chip ribbon
@@ChristianBelotti They are not. Polulu make 1x1cm versions. It's also way better to solder.
@@Mulakulu Buck converter is way overkill. Linear regulator is fine for this. At these current draws, any sort of switching regulator will use more power just to run than the load uses.
Credit where it’s due, Linus has managed to accrue an impressive team of personable and talented persons. Dan is an outstanding asset.
Your mom is an outstanding asset.
Not this time Dan! I know it's you!
@@flyingtentacle7631 bet your mom isn't 😏
I mean, It's not hard to wire up a wall wort to a battery lug(point of the video in fact). You can even find 3.3v ones and not need a voltage regulator.
@@i5uskoAh. Who.
Dan is great and the attitude of using what you have is also very positive. I often postpone a project, thinking I'll pick up the right parts next weeks, then 5 months later, I forget what the complications were and have to start from scratch.
Linus's commitment to turning his house into an IoT nightmare is commendable 😂
Scandinavian gang
I love ours, luckily though ADT has started doing home automation integration which is nice.
+1
I realise you guys were trying to do this with what was in-house, but you could totally make a tiny pcb the size of the button cell and put something like a SMT LT3470 on it with its associated components. Drop the voltage down to around 3v and still be able to compensate for draw spikes. I imagine the device goes to sleep for twenty minutes and then wakes up to measure the temperature and report in. That'll be what the sudden current spikes are and how the unit lasts so long on a button cell.
you should design the pcb and make them or what not, use pcbway.
Yeah, what crazy prime said. Design it and maybe send one to LTT maybe they would provide the specs. Maybe with the appropriate disclaimers. Then the rest of us could make them or have them made as needed.
Just use an lm317 in stead of that shitty little surface mount. What am i missing here :/
Yeah, I'm thinking you could fit a regulator, a cap or two, and a USB socket easily on a coin cell sized PCB. Add a 3d printed cover and you're featured on hackaday.
@@mmmhorsesteaks two diodes will do the same and will probably be cheaper
That's why you should use a capacitor as a buffer/filter.
buffer caps and slightly higher voltage. Batteries aren't stable over their lifetime, so pushing maybe 3.3-3.35V in should be plenty.
This whole smart home series just made me more sure that right now having a dumb home is many times more painless
Edit: I do of course understand that making a smart home is unspeakably less difficult if you're not Linus Sebastian and that Linus really loves making his life hell to have everything his way
No just don’t do it the way Linus is doing it. He’s trying to get everything the EXACT way he wants it rather than going with the closest available product with his desired features and accepting the downsides
You can do smart home without all that pain. Linus has a way of over-complicating things... The automated garage door, smart light switches, and now this heating stuff. My home also has all those things and they work pretty much all the time. And the few times there are hiccups, I've set things up in such a way that everything keeps working like it would in a dumb home. It takes some more planning and research ahead of time, but that's better than buying a system only to realize it doesn't satisfy your needs and trying to solder your way out of it.
Linus is definitely overcomplicating a lot of it. Their even a way to achieve this using Home Assistance with off the shelf components that are wired. Certain things like going ONLY Zwave cut off a ton of products. Then their likely definitely a better way than goin Ecobee. As some1 with Home Assistant showing over 50 devices, I can attest that it can be elaborate w/o the complication.
@@Niosus I'd say he isn't really overcomplicating it, rather he's gone down the patchwork route. I love my Home Assistant. It's reliable, works well and is easy to use by everyone in my house.
Linus has done the same thing many of us starting out with home automation have done. Buy something. Try to make it work. Buy another thing to try and make the first thing work. Along the way you find out some new cool functionality that is now possible because you got the two first things, but you just need to buy this 3rd extra thing etc.
That's how I got started too, but now I'm real boring so I if I'm adding something new, I first make a list of requirements, do some research to see what the most efficient way is to get those requirements met and only then do I click the "buy" button.
@Bingus That's what makes it fun though right? It's not like there's a school for home automation. You start out with an interest and just learn the required skills as you need them.
Obviously in this case they take some shortcuts because they have to make content, but the process and the frustration is pretty similar to what most of us have experienced when working on these things and trying to get it to do what we want.
Bravo Dan! When I saw the voltage regulator I was going to suggest simply using diodes and I was excited when you brought it up.
Dropping what is arguably Adam Savage's most recognisable quote right before referencing him directly is genius level writing and I love whoever wrote it.
Thing is, Adam Savage is quoting Dr Who (Tom Baker in "The Deadly Assassin" 1977?)
Can we start a petition to have Dan do videos on explaining electronics?
Pls, I would love that. He's so relaxing when explain something, plus ltt media group production.
Just watch eevblog.
As long as he stops over annunciating "soldering"
@@lilstubthumb But then that will make a 15 minute video to less than 10 and then Linus would get angry and drop Dan.
@@motosk8er2 bigclive too
Dan's camera presence is stellar and I hope he gets more on-screen time
+1
+1
The intro perfectly resonates with me. Not everything needs to be wireless. Some things are better with wire.
I did a mod myself at some point. Turned a battery-powered bluetooth speaker into a wired one. It now runs at 5V USB. Works just as well and the battery still exists just in case.
Highly recommended for everyone to look into the classic LM317 voltage regulator.
Someone needs to make a USB to fake button cell adapter with adjustable arms to simulate the different sizes of buttons. $5-10 each would be well worth less jank.
@@ccibinel the issue with that is that USB runs on 5V and most common button cells on either 1.5v or 3v, so some voltage adapter would need to be added as well.
@@HappyBeezerStudios AMS1117-3.3 (or others with different outputs) will do it and costs under $1 us. Messing with resistors to save 30 cents is dumb. If someone made a product combining this with a housing, arms and usb port it should be viable for $5-8
@@ccibinel Right, fixed voltage is all that is needed, since the batteries that are to be replaced are also fixed. And much more comfortable than a bunch of LM317
12:45 You can use a zener diode instead of a schottky diode to create a voltage maximum, rather than using the voltage drop from the turn-on voltage.
I was going to write the same thing :D a zener and a resistor
Better to use both I think? drop the voltage a little with a schottky or two and then a zener with a breakdown of 3.2v to lock it in. That should limit the wasted power and still be able to provide the spikes. Add a capacitor if it the voltage droops too much during the high load.
Or just 3 normal Diodes in series ,would be 5-(3×0,7)=2,9V that shouldbe find if not a basic 3.3 regulator with a little cap is the best option in my opinion
@@leobla744 Obviously 3.3v regulator would be better lol, not the point of this thought experiment at all. Issue with diodes in series is your peak voltage tends to be inversely proportional to the load, it's only marginally better than using a resistor as a voltage limiter. Using a zener isn't a great fix, but from my limited experience making power supply circuits, it's the best you can do without a timer + MOSFET circuit, or linear VRM (or either of these in packages).
the real advice is always in the comments - love it & thanks for the deep dive!
I’ve probably been to the shop about 10 times last year just to buy a random battery because I didnt have a specific battery some random thing needed.. I feel u, Linus.
don't you go to the supermarket?
If you can wait a few days, ebay is a good source. Buy a 10 (or more) pack, because with the shipping it works out about the same as buying one. Then chuck the extras into a drawer, completely unlabelled, and next time you need one you get to spend ages searching through your battery drawer.
That's my system anyway ;)
@@phuzz00 it is very obvious thing Linus pointed out the batteries get old batteries leak lithium batteries do good old and leak Oracle of old expired leaking batteries
@@phuzz00 but I understand the logic behind a large quantity from eBay or Amazon for cheaper also AliExpress or Alibaba all those the shipping times very
For me CR2032, AA & AAA is all I need. Other battery-requiring devices just don't make the cut or have their battery integrated anyway.
In all the cases, resister, diodes, Zener, and regulator, you need a load resistor (and a small capacitor across the load - 1uf) to get the minimum load current to get the
diodes to be forward bias, Zener or regulator to regulate. With out the load resister, at very low current draw, none of the methods above will turn on or act like you would expect them to do.
For example, the MIC5504 needs a minimum of 100uA to regulate (3v @ 100uA would need 30k load resister - do not forget the load capacitor).
And to think we thought Linus would have avoided this problem since the new house was so old that everything was already wired anyway.
We would think he would understand battery safety too, but here we are
@@Stanton_High what makes you think he doesn't respect Li-Ion battery safety? Could it be the soldering of 18650 cells to make battery packs? for a RED Camera i believe. Lol, I cant remember but something went wrong in that video, he either vented a cell or just never got it working cause of a circuit board regulator or something.
@@Frostbite1090 it was the way the circuit worked, it would fail/ shut down if voltage was removed completely
@@Frostbite1090 nah. the circuit board only had volatile memory. cause hey it's attached to a battery so if it ever doesn't have power the pack is dead and unrepairable anyway
@@Frostbite1090 that's one of those "you really shouldnt do it ever but if you genuinely don't have another option, you gotta" kinda things
bad example but realistic execution imo
For those bumps on power draw add a capacitor in parallel with your device, it'll smooth out the power draw issue and probably make it a little better.
Or use a voltage based approach like they did in the end which will make it *a lot* better instead.
@@bosstowndynamics5488 I'd still put a cap on the output of that circuit as well. It's replacing a battery, you want that output voltage to be smooth with as little ripple as possible and come as close to a battery characteristics as possible.
@@Mark-pm8nw But I would do all the voltage regulation outside of the sensor and near the power brick since they might get hot thus making the effort null
@@tommisgr I don't think a device that is intended to run on battery for years will draw enough current to make heat on that circuit a big consern.
Hello LTT. I know you never read this but here it goes.
A regulator of a normal size (for soldering) like a HT7533-1 that you can buy for 30 cents will serve you fine
GND goes to ground
IN goes to 5v
OUT goes to 3.3v
It is recommended to add a couple of capacitor in the IN and OUT for stability but are "optional" 10uF is more than ok (each would cost 10 cents)
And you are done. You can pull up to 100mA and if you want more the LM1117T has you covered with up to 1.5A or if you feel like an engineer a transistor would also give you extra power a 2N3904 would provide you with extra 200mA, those are 15 cents and you need an extra resistor.
Anyway...on a buget you can do just fine with 50 cents (gets cheaper by the 10s and 100s)
For 2$ you can have lots of power and even make yourself a nice small PCB and make 20 of them (which will reduce cost) and power EVERYTHING of a cable.
If you think it over a bit there are version of those regulators that if you have even more powerfull power supplies (over 5v) you can even replace devices that use two AA or two AAA batteries.
Then again.. (and sorry for this since I live in europe and the plugs are specially painful here). having 20 more cables and european plugs around the house is a nightmare
I think it would be cool to make a little circular pcb that can just hold the voltage regulator and make direct contact with the contacts on the sensor. That seems a lot less jank than this lol
lol. I just came to comment the same thing. if you just design a pcb with detachable cable, and order it from china assembled, you can just replace any button cell device and call it a day. durable as well. but then no content. :3
@@minolhewage Wouldn't be surprised if there is (or will be soon) an open-source (set of) design(s) to replace the battery for various button cell devices
You can get AAA and AA batteries that do this but I've not seen anything like that for coin cell batteries.
@@minolhewage flat flex pcb based usb cable no need to detach. battery lid fits. supercapacitor mitigates power fail. china not cheap
I love seeing you guys dive into electronics. If you do a project like this again - a 3.3v regular and an electrolytic capacitor in parallel across the output should run any device like this. The capacitor absorbs those spikes. Most circuits with linear regulators follow that approach
This, plus don't even bother with the potentiometer. The circuit is most likely 3.3v which most microcontrollers are today, so leaving it at 3.3v is just fine.
you need a LDO VR IC. The current spike is caused by the load. The cap would act to supply the current.
It just sucks that a linear regulator will waste some power, mot much at these voltages, but some (for the ultra power efficient folks)
@@billmiller4800 By my guesstimation the linear regulator will waste $0.03 in electricity to save $3.00 for a battery. Not counting all the other costs, losses, and benefits.
I don't know why everyone is saying to add a capacitor for those "spikes", those are 7mA spikes on a device being powered through what in practice would almost certainly be a half to one amp regulator on a half amp or more USB supply, you aren't going to get voltage drops in that configuration (after all, the 3 diode solution they went with is functionally a dirty voltage regulator and didn't cause any voltage drops during the current spikes)
11:06 Just casually soldering on a carpet.
In fairness, I have 4 of those ecobee sensors and they’re well into 3+ years without ever changing the batteries. I’ve been extremely impressed with how efficient they are. I upgraded to an Ecobee in November of 2019 and bought the sensors that December. I’d gladly hardwire any of my other wireless smart devices, as Ecobee aside, everything else absolutely chews through batteries
Yeah this is probably a dumb solution to a problem that doesn't exist.
Same with my Philips Hue switches and sensors. Three and a half years in and I’m just now seeing ‘Battery low’ for the first time. This problem does not exist lol.
@@heisenbaj Now take those 3 years, divide them by 100+ devices, switches, sensors like Linus has in his giant house and given enaugh time it would average out to 1 devices every ~10 days because not every battery lasts the same time.
@@gamingmarcus Usually it doesn't really average out all that well because similar devices (say 10 temp sensors of the same model) use the same amount of battery and batteries where put in there roughly at the same time. This means you gotta change a whole lot of batteries at once. And I would assume he does not have 100+ wireless room temperature sensors, the house is not that big.
A more reasonable solution would be to make or buy some sort of energy harvesting solution that maybe uses a solar cell. These devices pull miniscule amounts of power if you calculate it and it should be doable. And you could keep them wireless.
@@gamingmarcus Sure, and so I get wiring everything possible. But these things are bought in waves or all at once, so it wouldn’t be nearly as sporadic as 1 every 10 days. This also introduces a failure point and I’m guessing voids warranty so solution seems worse than the problem.
It's probably fine at 3.3v. you could use a 3.3v zener diode in parallel with the device and then a single resistor in series beforehand to make a simple diy regulator. You just gotta make sure you spec your components right so you don't fry anything.
Yeah~. A potentiometer is not the way to regulate voltage to an active circuit. Much less anything radio.
I'd even bet the first thing inside that sensor is a regulator boosting battery voltage to a standard 3.3V.
Yeah, that's the proper solution, but they probably didn't have any Zeners lying around. If you're ordering components anyways, that's way better than making a giant chain of diodes.
@@anlumo1 the series diode one isn't a bad solution. They have a generally stable voltage drop especially if your doesnt suck much current. With a general voltage drop of about 0.6 volts you could put 6 in series and maybe an extra shotky diode to make the 2 volt drop
@@RandornCanis Good point. Examining the actual power circuit can take a lot of guess work out of it. Sometimes electronics accept wider voltage ranges than you'd expect.
The funniest thing about this whole endeavour is when Linus at 15:00 implies that the sensor would likely run for years on its button battery lol. The truest definition of over-engineering right there! Also I think I'd take the battery over using up a plug socket :P
13:45 You could put capasitor parallel between plus and minus, so those spikes don't cause voltage drops
@@marcogenovesi8570 Don't they actually have an electrical engineer on staff?
@@marcogenovesi8570 Or maybe they didn't ask him?
Agreed, I saw this being a problem from a mile away. The advantage these cells (and batteries) have is that they're effectively really large capacitors. They will refuse to drop the voltage significantly unless you draw some high loads (>100mA). If you're replacing batteries this is one of the first things you address lest you get "funnied" by dropping voltages. Wall chargers are NOT designed to handle sudden spikes.
@@marcogenovesi8570 To get more content for a video? I don't know.
It's right there in the data sheet: VIN and VOUT are both supposed to be connected to GND with a 1uF ceramic capacitor.
I’ve been running one of those sensors for three years and have yet to need a new battery. While it would be nice to plug it in, ecobee has done a really good job keeping battery changes to a minimum.
glad im not the only one who noticed how good Dan has gotten on camera. not that he was bad before, but you can see the improvement and how comfortable he's gotten. i love anything involving Dan!
wow dan should host more, that was actually the best explanation of any electrical engineering i’ve ever seen
You probably don't watch any EE videos because he didn't even show what he built.
@@armLocalhost glad im not crazy. for a beginner like me, I have no idea why using diodes as resistors works. I know diodes typically have some resistance, but I don't know whether they have to be special diodes in this case?
@@purplelord8531 They drop the voltage by 0.6v each in forward mode, when there is no current there is no voltage drop
The lack of capacitors is probably why the LDO / MIC5504 didn't work. For further reference, if interested in electronics...
The MIC5504 (or any other LDO) requires input and output capacitors to function properly, these are called bypass capacitors and aid with stabilizing the voltages during random current spikes. The datasheet for the MIC5504 recommends 1µF at both input and output side, but it wouldn't hurt to oversize it with a 2.2µF capacitor.
An LDO lowers voltage by dissipating the surplus energy as heat.Thus the heat dissipation can easily be calculated by subtracting the output wattage from te input wattage.
Lets for example estimate the average output current to be 10mA.
- The output wattage at 3 volts would be 3V*0.010A = 0.030 Watts (Or 30 mW)
- The input wattage at 5 volts would be 5V*0.010A = 0.050 Watts (Or 50 mW)
- The heat dissipation would be 50mW-30mW = 20mW ... The other 30mW is used to power the device.
- Next we need to check the LDO's datasheet to get the temperature increase per dissipated watt of energy, usually labeled as °C/Watt.
- Our MIC5504 has a temperature increase of about 250°C per watt. We only dissipate 0.02 watts of heat, thus we multiply 250°C by 0.02mW = 5°C temperature increase.
- Thus at an ambient room temperature of 20°C we can expect the LDO to reach a max temperature of around 25°C. Which is below the maximum rated temperature of our LDO.
* Note:
- The lack of a pcb/circuitboard as cooling plate could cause slightly higher temperatures in reality.
- I always try to keep an LDO below 70°C tops, even when the datasheet states 150°C.
- Look into a buck converter when larger voltage drops are required! These won't dissipate as much heat, but are more complex and bit more expensive.
Do u have electric engineering degree? How much money do u make with that?
I have to admit, that SMD soldering was pretty impressive considering the fact that Linus was in such a sub-optimal environment for doing it. I've done some amount of SMD soldering though I tend to avoid it if I can because it can be fairly annoying (especially if you're talking 0402s).
Agreed, I think a better solution would be to use all through holes and get some very small breadboards. Using SMD is asking for trouble lol. If breadboards cost too much you can get these prototyping breadboards as well (these do require soldering though).
The SOT23-5 is plenty big enough. Though I probably would have used through hole passive components and soldered them point to point.
@@SnakebitSTI Eh, I wouldn't do that to myself or unironically recommend it to anyone else.
We need a Dan vs. Anthony P.C. build off, both of them are as wholesome as can be. We love seeing them both!
Two points to make. The battery voltage will vary above 3.3V when full so there should be no issue with a 3.3v regulator.
Also for those larger current spikes a simple electrolytic capacitor for a few cents would help greatly.
I don't understand why they didn't ask a proper engineer which they have in the team and just over engineered it like hell.
Like they could have used a voltage divider with a cap and call it a day.
@@armLocalhost I mean they could, but also why not just use an lm317.
@@varno Not even, 3.3 V Zener diodes are common, way smaller and simpler to solder than an lm317.
@@armLocalhost could have been way over engineered but their point was to do it on the cheap.
But I would say stay clear of voltage divider with no protections at all.
@@nadie9058 just to be bulletproof, I wouldn't use a shunt regulator here, they don't work well for bursty loads like this one, and are super inefficient. This was because he wanted 3.2v rather than 3.3.
With jlcpcb being as cheap as it is, I probably would have just made up a pcb in kicad quickly with all the components and had them solder it together.
At the same time, I am suuuuper surprised they don't have any 317s in a to92 there. That and a couple caps would solve the problem imediately.
I love Dan's quiet voice and demeanor. Makes difficult things seem not so stressful and could probably read the phone book and put me to sleep
I love the new format!!
There is a commentary from Dan and explaining how it works and alternatives. Keep this up.
And include the beep part on the edit. It feels more natural to me ateast
This is more than just "First world problems" this is another league: "Linus' house problems". I swear this guy has these super particular problems he needs to solve and make content of.
I agree, but this video is one of the more useful ones. Although most people won't have such specific voltages
That's because this is entertainment more than education
Yet I wouldn't want to say the arguments he states are invalid, they most certainly are.
To be honest I'm from a 3rd world country and i despise batteries my wirings are usually way jankier lol. My only entirely battery device is my phone
He made a problem for himself. At least with those sensors. They way underestimate with that 2 year life span. I've had some going for 3 or 4 years now without needing to change the battery.
I am an electronics engineering student, never have I though about using diodes voltage drop to drop down regulated voltages, this is really smart and I will use it whenever i need to do something the "fast and junky way".
BTW, you could used a rezistor parallel to the ecobee terminals to keep the diodes in the 0.6 voltage drop range . this would increase current draw (but the powersuply still draws some amount of power so it does not matter) but it will allow you to use this trick on devices with overvoltage protection
It's an ancient hack, but relies too much on current draw. I usually can't use it because if it's that low voltage, it's also low-power and thus usually very "peaky" in current draw, and I can't justify pissing away milliamps in a device designed to take microamps on average. So I never design it into a device. And when hacking an initial prototype together, I just use an adjustable buck converter module or an adjustable LDO regulator. Or feed it off a power supply. So the utility is limited.
The main issue with hacks like that is that you introduce unknowns, so there's that much more to exclude as root cause if something doesn't work.
Using the characteristic voltage drop of a semiconductor to get a reference voltage is a fundamental concept is designing voltage regulators
I'm a EE who mainly deals in railway signalling, but I did a lot of power engineering too. Be it small or big.
You can use a zener diode to pull a voltage to a known value, you'll learn about it later when doing courses about designing power supplies. A diode and a transistor are the basis for a voltage regulator.
What happened here is he used a resistor devider but failed to consider that the ecobee pulls power too, thus acting like an impedance and influencing the voltage drop. Could have been solved with an op-amp in VF config.
THB though i'd gone the simple route and grabbed a mini buck converter from ali express, they are tiny now and mitigate the hassle. No point re inventing the wheel when an absoulty perfect solution exsists.
@@robmckennie4203 Ah but you are then dependant on the temperature. I can remember while doing my EE courses, the actual math behind SC. An absoult pain, was the only bit I didn't enjoy.
We usually use diodes to combine low-power power supplies together without too much hassle.
I recommend to get some prototype PCB boards. They are available with the Through hole pitch and the SMD too.
You can easily cut them to smaller pieces, and they provide a nice connection to your prototype device.
You should get green-screen pull down window shades for the house, and then just drop in various windows screen backgrounds outside the windows.
he already leaked his location live on stream, idk why he is even trying to hide it by this point.
@@nezu_cc Because the amount of people that saw that section of a stream is a whole lower than the amount of people watching video's. Especially considering there's going to be repeated videos in his house.
@@SolutionJW yeah, but information spreads fast. If someone wants to know his location all they need to do is google it. Imo there is no point hiding it if you can just google it. Nobody is going to play geo guesser using his windows if the info is already out there in the wild and confirmed to be correct in multiple ways.
@@nezu_cc My point is more along the lines off 1) there's no point in NOT hiding it and 2) people that aren't already looking for the information might get interested because they see something interesting outside his windows. Why give them that opportunity, right?
Besides, hiding his location might not be the sole reason for the blur, who knows
@@SolutionJW as others pointed out, sometimes is batter to hide in plain sight. Pixelation only draws unnecessary attention to it. It's especially pointless once the original purpose behind it is already gone. By this point, it does nothing else expect waisting editors time and drawing unnecessary attention.
Really love the home automation/improvement videos and glad you guys let Dan have a segment in to help understand things better. I would love to see more detailed content on this stuff in the future or something as a follow up to this video with some of the viewers suggested comments. I too hate batteries and wish to keep everything hardlined so seeing that it’s something that is doable has at least opened a door into further research for me on this topic
Dan's screen presence is amazing.
These videos just keep getting better and better... I've been watching for like 5 years and because of LTT I've built 3 computers and now solder my joysticks and retro systems. I would never even try something like that before. Ty Linus
Rather than soldering to the contacts in the device, cut a piece of plastic (e.g. bottle top or two, or even a bit of wood) to the right size and then stick the wires to it. This gives the added benefit of being able to convert back to a battery if needed.
Basically a plug the size of a CR2032 (THe name itself designates it as a lithium cell with 3V nominal and 2.0V end point with 20 mm diameter and 3.2mm height. The number is the trick: 2032 -> 20x3.2 or the similar 2025 -> 20x2.5 or even the 18650 -> 18x65.0 that are used in many laptop batteries)
Dan is my hero. My batman. Someone I didnt know I needed in my life but now everytime I see him in these videos it sparks me so much joy. He is so funny and has so much knowledge, that he even stands out next to so many other great people at LTT.
Dan is getting so comfortable and confident on camera, I love it!
Anyone else had moments where you remember buying a lot of batteries recently, but you can't find any? lol
Do you have children?
Because if so it was definitely them.
Ong I buy 30 pack of AA and they’re gone the next week
every time I buy them
I think you left them in the drawer with random papers and wires
I recently orderd 30 rechargable AA batteries I am missing 1 of them already...
As someone who did something with the same outcome couple of weeks ago, doing this live on camera would be unbelievably hard!
Since you're on home assistant you just could assemble your own sensor based on ESPHome (esp8266+DHT) and put it in your wall socket
I'd advise against a DHT11 for something used to control the room heating as even a degree change can make the room super cozy vs uncomfortably cold, and the DHT11 is pretty inaccurate at ±2 degrees celsius. A DHT22 is a fraction pricier but much more accurate, and if you want to get even better readings( temp+humidity+air pressure), there's the BME280 I think it was called.
But yes, the 2-pack of Linus sensors cost a whole 120 dollars, and the ESP is far far cheaper. I reckon if it's just a single sensor, they can also use the ESP01 as it is TINY.
@@namankohli1061 better would be the Sensirion (IIRC) SHT40, SHTC3, or SHT31 (in order of price), as they're I2C interfacing, not the weirdness of counting the pulses that a DHT11/22 uses. Less of a concern with a single purpose endpoint, but might as well avoid.
Then again, I don't know how annoying those are to physically interface with if you don't get a dev board with them like Adafruit sells.
Big brain solutions. I doubt linus wants to dedicate that much time into this however.
You can also use diodes in their forward direction to drop voltage. Like you could have used a schottky diode to drop ~0.2v off of the 3.3v supply. And that isn't dependent on current (ideally of course) like a resistor/pot is.
Ah, or I could have just watched the rest of the video 😁
I thought about the same.
I thought of that but assumed it wouldn't work because linus said "exactly 3.2v"
@@munjee2 Exact? With batteries?
They do drop in voltage from fresh to when the device starts complaining...
@@michaeltempsch5282 that's what I was confused by myself
0:48 "shockingly inexpensive" I love that pun
If you're doing a few, might have been worth drawing up a PCB and having JLCPCB or similar fab and SMT a tiny board the size of a battery to drop into the sensor that could have the voltage regulator and maybe the filter caps that are part of the reference implementation on the datasheet.
well... someone do it up and sent it to LTT to add in the description, there's bound to have lots of peeps wanting to do that... without the patience or tools to do it...
You could do that, or just order an off the shelf 3.3 V voltage converter.
plus you could add a USB-C port 😏
The fact that Linus is using surface mount components is hilarious, I am almost certain you could use a through hole LDO voltage regulator and have no soldering problems
Thank you to whoever edited in the fire hazard warning @ 6:04 regarding cheap USB A 5V 1/2A plugs.
It’s become such a common problem now that even high street retailers like Poundland are selling these things and if you or anyone you know uses them, dispose of them straight away.
It’s not a matter of if but WHEN they will either melt or explode which will start a fire if it’s unattended.
That screen transition edit from Linus to Dan was absolute perfection. Hats off the editor.
1:48 In my experience most things use the CR20 batteries which describe their diameter and thickness in their name. A CR2025 is 20mm in diameter and 2,5mm thick, depending on how the Battery is held in it can also serve as a replacement for a CR2030 battery.
Haha editor I love the screen wipe where dan pushes away the screen 😂 more fun like this!
That cut in was a great format for Dan. He gets to bring the facts and knowledge, but we still get to watch linus suffer on his own for a bit.
Linear 3.3v regulator plus a small capacitor on input and output would be my approach to this - no need to faff about with a potentiometer, and the caps should soak up the power spikes reasonably well.
would have to be a LDO. you need a LDO VR IC. The current spike is caused by the load. The cap would act to supply the current.
To simplify, just use the LM2596S DC-DC Buck Converter (2 wires in DC1.25V-35V - 2 tires out adjustable voltage) it only costs $1.96 free shipping from Aliexpress, simple and easy!
I’ve been considering doing this to my Ecobee sensors for years, and I’m glad I talked my way out of it for the very same reasons you had trouble
And here I thought this was actually a helpful tutorial on how to convert battery operated devices to hardwired, for people who aren't electronics experts, turns out they just needed to publish a video per the schedule. LTT is more of an entertainment channel than actual tech tips.
Most frustrating part is how needlessly rushed and botched it was. The amount of prep they did seemed appropriate for an emergency video they need to do _this afternoon_ to put out _something._ But from Dan's comment, it's clear it was in the pipeline for at least a week. They could have picked a proper 3V LDO and designed a CR2477-sized PCB around it (possibly even with a USB connector on the back) in 15 minutes (or a couple hours tops, if Dan doesn't have experience with it) and have it fabbed - including population - and delivered in ~3 days. They had the time and they certainly have the money.
Love the more engineering-type content. I have 0 background in elektro or engineering but it's really interesting to follow these simpler things
I recommend getting some non-SMD sized (TO92) voltage regulators in various voltages for future "unplanned" projects. And get some transistors (BC547C, BC557C) as well while you're at it :) An inventory of various basic components is worth its weight in gold.
agree with that, I converted a 2x AAA battery wireless headset base to a Li-po powered and used a MCP1703 in TO92 package, never failled onced! it's a 3V fixed voltage so perfect for tha kind of application but if I remeber they offer other voltages
@@ItsJust2SXTs me using a wired headset like a pro
@@vappyreon1176 working on cars with a wireless headset is way better than wired or when I work it's plug in the computer but I can move without walking on the wire or have a 10m wire... the headset it self work for 3-4 days on 2x AAA but the base was like a day and half and I bought rechargeable when I got it, still the smae 4x batteries in 7 years. I broke the base case one day that was another reason for the switch and use a MAX1811 for charging the lipo
@@ItsJust2SXTs man, I would never trust putting a lipo that I had anything to do with that close to my head. I still remember the old olight lipo flashlight exploding in mouth thing.
@@marcusborderlands6177 me either that's why it the base not the headset it self, it's a 500mah battery so it's not that big. It transmit the sound in the FM band not bluetooth. When the lipo will die eventually(it's 5 years old now so it lost some charge but it's not buldge), I will change for a Li-ion at least they are in metal casing. Bluetooth Headset use li-po but still too close for my tase. Remember the samsung that were blowing in pants? that was scary to see..
Seems like a perfect use case for OSH Park. A perfect purple PCB would be awesome. It could just fit perfectly into the button cell holder. Anyway... Capacitors are great for current spikes too.
honestly considering they have EE(s?) on staff it seems a bit like no one has read that datasheet :/
For this kind of projects I use MP1584EN or LM2596 buck down converters. They can deliver about 3A tops with 4-40V input. You can buy them in 5 packs for about $10, much easier to solder and way safer. They are adjustable, the LM2596 uses a 10 turn pot so you can adjust it at precisely 3V. They hold the output steady at variable amps and variable input voltage. The MP1548EN is the smallest which will fit nearly anywhere.
It would be really cool if you guys made a drop in replacement board that just took a micro-usb cable or something. A small, 2-layer PCB would be able to fit the voltage regulator as well as include contacts to the existing battery contacts. And maybe an inductor or something could help with the current spike?
Thanks for this amazing guide, Linus! I will literally have no uses for it, but it was incredible to watch you turn a wireless thing wired. I am more than impressed.
I absolutely love the formate of Linus try's something somewhat jank - struggles - cut to how to get it to work/easier way - back to linus
The regulator spec sheet would tell you how to calculate the resistance needed to get your desired voltage. Than you can use small exact resistors vs large pot.
You could probably buy a pre-made regulator on PCB for ~$5-10 and just swap the resistors to get the voltage you want. Than you can easily solder the wires on.
Even more, most of these chips have a freely adjustable version. Then you have to attach the potentiometer between GND and the output, and the middle pin to the feedback (usually called FB) pin. then you can set almost any output voltage to the device, that is independent of the load on the output.
the way he used the resistor is not to adjust the voltage regulator output, but to drop some more voltage in series. However as Dan explained, the current draw from the sensor varies wildly, therefore the voltage drop across the resistor does awsell (remember Ohm's Law V = I*R)
@@gabrielenitti3243 You NEVER use a resistor to try and drop voltage as a power supply, especially for dynamic loads . V = I*R, meaning every time the current draw changes the voltage changes, those devices need constant voltage. I don't recall specifics but Dan showed diodes to drop voltage which is fine, just not efficient.
At that point, you may as well use the LM2596 board and just smear it with hot glue. Yes, it'll use more power on idle since it's made for supplying up to 3A, but the size and cost makes it very well worth it.
Or use literally any $0.5-$1 buck converter off eBay
Riley absolutely nailed the entire sponsor spot at the end.
To make a sponsor feel that natural and entertaining is some premium talent fr
I noticed that too I want those shoes now
The voltage drop caused by the potentiometer depends on the current flowing through it, so the voltage you are measuring will vary depending on the current power consumption of the sensor, which explains the over 4 volts you saw
The only proper way to do this is to get an adjustable regulator that lets you connect a potentiometer (or fixed voltage divider) to an adjustment pin that will actually change the regulation voltage. Assuming 3.3V is really too high there are 3.0V fixed regulators easily available (well, they do manufacture them at least...)
@@ailivac yup, that's it. I think 3.3 V would easily be in spec assuming a ±10% tolerance. And I think the vanilla battery is probably over 3.2 V anyway when full.
I find that voltage regulators are best used to get a very exact voltage, for example for an ADC's reference voltage. For this use, I'd prefer a tiny QSKJ buck converter.
It's amazing how Ecobee had those room sensors for awhile, Google Nest would implement their own version less than a decade later.
Love to see more tech creators doing this type of content! This exactly the type of content we need to expose the next generation of techies to!
My smoke detectors and security system are hardwired but have battery backups for when the power is out. The smoke detectors start beeping once every couple of years and I just replaced my security system battery for the first time in 8 years. It was like a small car battery.
I wired up a 2XAA battery holder which outputs 3 volts to my scale to replace it's original CR2032 and it's still going after a few years. If you don't care about the aesthetics I think it's a decent solution
Tip for next time: If you want to have a small DC power supply, you can use a Z diode and thus set the voltage to a fixed voltage. The structure looks like you have a resistor for the Z diode so that the current to the diode and the consumer is limited. The consumer is then connected parallel to the Z diode. The Z diode determines how high the constant voltage is that falls off via the diode. The Z diode is installed in the blocking direction. This makes it relatively easy to set up a small power supply.
That is what I was thinking. Why not get a zener diode and just be done with it? Or if that's too much hassle to get the right one, just use a good old LM317
@@fermitupoupon1754 Yeah, an LDO or a zener seems like the move here
I might be misunderstanding because I haven't used zeners myself before, but wouldn't shorting one across the supply output of a regulated voltage source (the 5V wall wart) cause the upstream active regulator to dump upwards of half an amp through it trying to push the voltage back up to 5V? That seems much more wasteful than an in line linear regulator.
@@bosstowndynamics5488 you’d add a resistor to the 5V output. That will limit the current.
I’d guess the resistor would be in the range of 1k to 10k, but that can definitely change
@@bosstowndynamics5488 In this case you'd use a 3.3V zener diode. You'd put that in-line on the positive side.
So you backfeed it with the +5V from the wall wart, into the +3.2V for the sensor.
The +5V exceeds the zener diode's zener voltage. Which causes it to break down and start conducting at it's zener voltage of 3.3V.
Add in a smoothing cap and a discharge resistor. Done.
No need to fiddle with a pot to try and calibrate a regulator. As long as you don't try and draw more amps than the zener diode is rated for, it'll regulate the voltage.
10:13 something you could do for the future is find all of your QR codes for your various devices and export them as a large image file that is stored on your server. I have this done on my security cameras so that in the unilikely event i need to add them on a new phone or because i forgot how to rearrange them in the app UI, i can just import the QR code from a file rather than to go to each security camera.
but this is only a solution if the software allows you to import a QR code. I guess another option is to open the QR code file on another device and point your app/camera at the other device
I love how much Linus loves to overcomplicate things, it makes for entertaining viewing.
His misery is really entertaining. 🤣😂👍
If not for the "cheapskate" part, you could have used a CR2032 battery eliminator and a bit of padding to make the contacts touch each others.
You still have to supply the battery with 3.2 volts
Battery eliminator are made to replace batteries by a power supply, it either comes with one integrated or you have to plug it on a corresponding power supply.
It would have avoided the sketchy soldering with SMD.
3:33 - man’s got the backrooms in his house
I had a unit in my last two houses that would turn on the HVAC's central fan to circuit and average the air within the house. Some bedrooms would get super hot, or super cool, and this balanced the house's overall temp without cycling the heater or air conditioner as much.
Instead of the three regular diodes in series, you could've used a Zener Diode. For you application with a constant load, Zener Diode would be your best option
Yepp, an 1.8V diode would have exactly resulted in the 3.2V needed.
Next time use a ready to use stepdown converter. It's a little bit bigger, but you can make it part of the cable
That was a wonderful insert Dan, I very much enjoyed hearing him explain.
That Adam Savage quote is my favorite one of his.
DUNGEONMASTER quote*
@@Jorgelhus what?
@@BorealBlizzard the quote came from the movie DUNGEONMASTER, but was popularized by Mythbusters
@@Jorgelhus Ah I had no idea
I want to see a video about battery tech. How good are they today? What types are there. Are there other kinds beside Lithium that could power a handheld?
That would probably be a Techquickie video.
outside of nimh batteries, not really. Like you have Lithium Ion which do 4.2 V, and the lithium cells that do 3 V but with super low discharge. NIMH do 1.2 V and Alkalines do 1.5 V. There are lithium ion AAs but they are pricy and need a special charger. Basically the best of both worlds
There are way to many battery types, to be honest. Most of them are/were either proprietary for cameras or for a specific size constrained applications.
@@abhimaanmayadam5713 Maybe solid state batteries will be viable for consumer products soon.
Hey Dan, Linus - the voltage regulator solution with potentiometer is actually still almost right on the money. You've got a few realistic possibilities for keeping this cheap and fixing the problem with transient spikes in current draw ruining the voltage.
The first, and probably easiest is just to toss a small capacitor in series with the voltage supply. This will smooth out those transient spikes and keep the device online. Since these devices are tiny and only pulling a few milliamps at peak, even a 1uf is probably sufficient, though I'd spring for 10uf. Also, anybody seeking to use the DIY approach should probably just buy a lot of 100. 100 caps is usually like $3, or $0.03ea, so if you make a mistake or need to do another device, no big deal.
Also, there are ready-made solutions for this you can get straight from china for super cheap, so it might not be worth the effort. For example, with prime shipping you can get a pack of 10 3-12v input 3-12v output range 3A voltage regular PCBs. These are more robust - it's a pcb with surface mount components and nice through hole pads for input and output, and its a complete variable voltage regulator circuit with fuses, an output transistor, and capacitors for smoothing. A ten pack of them usually runs about eight US dollars, or about eighty cents to convert a device -- what Linus ended up with after accidentally breaking one of the regulator ICs.
Did Linus consider moving the thermostat to the other side of the wall? It would be significantly more reliable to move the thermostat instead of adding another device. If Linus just needs the thermostat on the back side of existing wall, he could drill another hole.
Yeah I don't get it, that house was just renovated and they added a lot of wiring. It was intentional to put the thermostat in the wrong spot... A lot of weird things have been done with that house.
@@cedricpomerleau5586 Don't forget Linus has been giving a crapton of instructions on what he wanted to contractors, and a lot of said instructions were different from what they would typically expect to do. Mistakes and/or confusion have happened a lot because of this.
Yea, what's hilarious is that Linus needed to get these sensors in the first place because his smart thermostat system is borked.
@@cedricpomerleau5586 it's possible they weren't able to predict the issue of unintentional heating