4:45 - With no diode the inductor will be producing a high voltage at its input and thus the output of the FET. The MOSFET won't be having a good time and is likely to fail.
Hardly a surprise the output gets nothing until the PWM is almost DC... Feeding a square wave from the FET through an inductor... Remind me, what does an inductor do again ;-) As you say, before it reaches this phase, the inductor is going to be throwing uncontrolled back EMF onto the FET output. Poor thing.
How long do we have to wait for an output capacitor? Muppet 3? The *power* ripple on the output is potentially affecting the accuracy of your readings.
KX36 I did a similar comment months ago. Proper fet drivers so they don't spend too much time swithing, schottky diode and low esr capacitor... Bit I guess he wants to do 10 videos and the corresponding views doing it one step at a time.
I like the way he is doing it because I am learning a ton and not missing anything. You have to remember these videos are not just for experienced people they are for beginners too.
chopping it up is creating ripple. The current output of the inductor will be a triangular(ish) wave and with only a resistive load (lamp) and no capacitor, both the voltage and current through the lamp have this triangular wave in phase, meaning the power in the load is a triangle wave. The devices he's using to measure power are designed for DC. they sample at a low frequency and have low bandwidth making their behaviour when faced with a high frequency triangle wave undefined. Adding an output capacitor will smooth out the output voltage and therefore the current and power through the lamp and so the DC power meters are likely to give more meaningful results. It may or may not make a big difference to the numbers but it gives more confidence that they actually mean something. As for the input side, the voltage is DC but the current is highly pulsatile, so there is a similar problem. An input capacitor in the circuit would smooth out the input current. But it depends on where the source is measuring the current. It may be that it measures it before its own output capacitor anyway.
This is the type of research I really enjoy. There aren't many synchronous buck/boost converter modules coming from the chinesium on ebay yet. Perhaps your videos will inspire them to start producing some higher power, all-in-one DC-DC modules, so that we don't need half a dozen different converter modules for our varous project needs. :)
there is at least one that I know of, the LTC3780 is exactly what he is building in one chip it is just 11-15$ per board. I will be comparing it to some SEPIC modules on my channel when they all arrive.
you really need some capacitance on it to smooth the voltages and average the power, your power meter is probably giving inaccurate results. anyway take another look at the LTC3780, it basically does what you are building, you would just need to control it from the arduino. anyway thanks for these fundamental (with experimentation) videos!!
Julian, for the sake of clarity would you add some black, Sharpie, lines on your breadboard between the connected, binding posts. Maybe it's just me, but I had to repeatedly think about what binding posts were connected during your explanation. Otherwise, I am enjoying your explanation of buck / boost conversion. Cheers.
Another thing to consider with diode choice is the reverse recovery time. If you're switching at 100kHz+ then some jellybean diodes may end up conducting the wrong way for a significant part of the switching cycle. Schottky diodes are normally quite fast in addition to the lower forward voltage. They're not perfect though, schottkies have a lower max blocking voltage (normally ~40V) and can have a few mA of reverse current when blocking.
Not today. After the diode comparisons, I tried the 2-MOSFET synchronous buck converter. That's when things started to go a bit wrong. So I'll probably film an update tomorrow - with the emphasis not on efficiency!
Oh tell me there was *SMOKE* !! And that you have it on video! ;-) I'm guessing your synchronization wasn't exactly in sync? Did you fry the PS? Another *_Julian suspenseful cliffhanger!_* Arrrggghhhh!
Excellent video, thank you! Datasheets for converter chips often refer to the speed of the diode, not the voltage. Which one is the most relevant regarding the efficiency ? Could you measure the forward voltage of the mosfet?
Julian Ilett - When I used to service TVs back in the day, there was one model notorious for a selenium diode failing, and it was described as one repair to avoid. I couldn’t see what the fuss was about, until a customer asked us to pick his TV up because it smoked and stunk his house out. The drive back to the workshop with the TV was unforgettable, until you’ve experienced the odour of selenium compounds, it’s impossible to grasp how vile it is! It stunk the car out for ages, and the workshop was little better. If you ever use selenium devices, be very careful not to fry them!
I would suspect that the body diode of the MOSFET is a little bit better than the discrete diode because of their difference in voltage threshold AND in their recovery time (the time for which the diode still conduct while freshly moved into reversed bias, which can reach the milli-second). A discrete silicon diode has a slower recovery time than the body diode and itself, lower than a Schottky. If the power is too high for a Schottky, it is recommended by the literature to use SiC (Silicon-Carbide) diode. You should check also a fast switching silicium diode (1N914, 1N4148) versus a rectifier silicium diode (1N400x), there should be a noticeable difference, with a better efficiency for the 1N914.
What frequency is this thing working at? Might be interesting to try some ultrafast pn diodes (e.g. UF4007) and see how they compare to the MOSFET body diode. Also, as it has been stated already, please add some smoothing capacitors at the output.
So much engineering paranoia around switching converters that I almost expected this setup not to work decently at macro scale. Turns out this kind of old school experimental demonstration setup is still a great way to visualize. Looking forward to going synchronous in part two!
There is one thing you missed. The schottky diode switches a lot faster. I guess your switching frequency is a lot lower than most SMPS. That's why schottky is not much different than normal rectifier. But once you get to a bit higher frequencies - let's say 50 kHz (which is still relatively low), a regular diode won't be able to do the job. I've tried it at 100 kHz - it just doesn't work.
I'd find it useful if you could briefly summarize what you've got set up and what you're trying to do with it at the beginning. Sometimes in the process of trying to work out what everything is, I miss relevant portions of the presentation.
Is this really accurate, or did i miss something in my RUclips electronic classes? I thought parallel diodes will always have less forward drop than the value of the smallest diode in the circuit. Like resistors do in parallel.
Your first case with a high voltage diode is not a good example of a silicon diode in this converter. Typically, such a converter would use a low voltage, fast, reverse recovery diode.
that would essentially be the same as synchronous but what I want to know is if there are any that will work with high frequencies for PSU switching applications.
I just remember Julian playing about with ideal diodes a long time ago, I think in a postbag vid and there was some issue with the ones he got, but they never reappeared for further investigation
"ideal diodes" are not real components. There are ICs that act like an ideal diode, but they´re basically just a small op-amp circuit that needs power and cannot offer any useful current capability, and even if they could they would be horribly inefficient.
I looked up the video I'd seen it in, it was the last item in postbag #74 way back in Nov 16. the one that was shown had 2 n-channel fets in a configuration that didnt seem to do anything. it was just a thing I'd never heard of before
Well...all he needs to know is if the battery is more than 12V(or whatever the maximum voltage is); which he could easily do with an optocoupler...That mosfet driver does not look very promising...
"Not easy by common positive side" birds-are-nice.me/ipfs/QmeHTKsAWhaKeygAqRBkAUkt5t2e9NPZcHT2ZcuyZcXcwy/solar_optimiser_schematic.png I made a Muppet of my own months ago. Common positive isn't that hard. Trick is to measure the inverted output voltage, then subtract it from the input to get true output. Common positive may be the way to go - it certainly simplifies drive. No messing around with opto-isolators.
Yea... That is a way of doing it but there will always be a current path through the voltage divider you used to measure the output voltage... Which basically will charge the battery even if the mosfet is closed... I am not that experienced in battery technology but that does not seem like a good thing
Now try putting 3 schottky diodes in parallel, not to greatly increase "safe" current capacity but instead, lower forward drop at any current that's a significant % of the diode rating, since ultimately the diode rating depends on the heat generated in that package size, which is based upon the forward drop at that voltage (and of course, heatsinking in a package that allows for it. Purists would hate the idea of 3 diodes in series but stop to consider how inexpensive a bag of diodes is these day and how precious efficiency can be on battery or solar power.
4:45 - With no diode the inductor will be producing a high voltage at its input and thus the output of the FET. The MOSFET won't be having a good time and is likely to fail.
Correct, the mosfet would basically be conducting in breakdown, which is not healthy for it and will cause heating.
Hardly a surprise the output gets nothing until the PWM is almost DC... Feeding a square wave from the FET through an inductor... Remind me, what does an inductor do again ;-)
As you say, before it reaches this phase, the inductor is going to be throwing uncontrolled back EMF onto the FET output. Poor thing.
How long do we have to wait for an output capacitor? Muppet 3? The *power* ripple on the output is potentially affecting the accuracy of your readings.
KX36 I did a similar comment months ago. Proper fet drivers so they don't spend too much time swithing, schottky diode and low esr capacitor... Bit I guess he wants to do 10 videos and the corresponding views doing it one step at a time.
I like the way he is doing it because I am learning a ton and not missing anything. You have to remember these videos are not just for experienced people they are for beginners too.
Shoosh! Don't say that, otherwise people who "build this stuff" will say that there's no ripple even without the cap on the output. I've been there :)
Would his source of power make that moot? If he is using solar then he is just chopping up clean DC power so ripple would not be a factor right?
chopping it up is creating ripple. The current output of the inductor will be a triangular(ish) wave and with only a resistive load (lamp) and no capacitor, both the voltage and current through the lamp have this triangular wave in phase, meaning the power in the load is a triangle wave. The devices he's using to measure power are designed for DC. they sample at a low frequency and have low bandwidth making their behaviour when faced with a high frequency triangle wave undefined. Adding an output capacitor will smooth out the output voltage and therefore the current and power through the lamp and so the DC power meters are likely to give more meaningful results. It may or may not make a big difference to the numbers but it gives more confidence that they actually mean something.
As for the input side, the voltage is DC but the current is highly pulsatile, so there is a similar problem. An input capacitor in the circuit would smooth out the input current. But it depends on where the source is measuring the current. It may be that it measures it before its own output capacitor anyway.
This is the type of research I really enjoy. There aren't many synchronous buck/boost converter modules coming from the chinesium on ebay yet. Perhaps your videos will inspire them to start producing some higher power, all-in-one DC-DC modules, so that we don't need half a dozen different converter modules for our varous project needs. :)
there is at least one that I know of, the LTC3780 is exactly what he is building in one chip it is just 11-15$ per board. I will be comparing it to some SEPIC modules on my channel when they all arrive.
synchronous converters are more expensive and more complicated, so it´s not really worth it for low power low cost converters.
I've not seen a bidirectional LTC3780
Julian Ilett oh true it only goes one way, hrm something for me to look into I guess
you really need some capacitance on it to smooth the voltages and average the power, your power meter is probably giving inaccurate results. anyway take another look at the LTC3780, it basically does what you are building, you would just need to control it from the arduino. anyway thanks for these fundamental (with experimentation) videos!!
The ltc3780 is my inspiration for this project, but muppet2 will be bidirectional.
Oh forgot it is only one way, left to right, The goal is to have it be a power shifter from one side to the other and vise-versa
Julian, for the sake of clarity would you add some black, Sharpie, lines on your breadboard between the connected, binding posts. Maybe it's just me, but I had to repeatedly think about what binding posts were connected during your explanation. Otherwise, I am enjoying your explanation of buck / boost conversion. Cheers.
Another thing to consider with diode choice is the reverse recovery time. If you're switching at 100kHz+ then some jellybean diodes may end up conducting the wrong way for a significant part of the switching cycle. Schottky diodes are normally quite fast in addition to the lower forward voltage. They're not perfect though, schottkies have a lower max blocking voltage (normally ~40V) and can have a few mA of reverse current when blocking.
Are we going to get a 3rd video today? Waiting for part 2 of this one!
Not today. After the diode comparisons, I tried the 2-MOSFET synchronous buck converter. That's when things started to go a bit wrong. So I'll probably film an update tomorrow - with the emphasis not on efficiency!
Oh tell me there was *SMOKE* !! And that you have it on video! ;-)
I'm guessing your synchronization wasn't exactly in sync? Did you fry the PS? Another *_Julian suspenseful cliffhanger!_* Arrrggghhhh!
You left us hanging, Arg!
I really like this series of videos. I wonder if the full 4-FET design could be implemented with an h-bridge IC and an arduino.
Excellent video, thank you! Datasheets for converter chips often refer to the speed of the diode, not the voltage. Which one is the most relevant regarding the efficiency ? Could you measure the forward voltage of the mosfet?
Why don't you draw lines on the wood between the (closer) terminal posts that are interconnected?
How could you forget to test the beloved germanium diode? With a forward voltage of around 0,3 Volts, it should be even better
I don't remember high power germanium diodes - did they exist?
There's also the selenium rectifier :)
Julian Ilett I dont think there are high power ones. But it should be enough to test its efficiency before it melts. How about a vacuum tube diode?
Germanium diodes have insanely high reverse leakage, especially as they warm up. So they won´t be very efficient.
Julian Ilett - When I used to service TVs back in the day, there was one model notorious for a selenium diode failing, and it was described as one repair to avoid. I couldn’t see what the fuss was about, until a customer asked us to pick his TV up because it smoked and stunk his house out.
The drive back to the workshop with the TV was unforgettable, until you’ve experienced the odour of selenium compounds, it’s impossible to grasp how vile it is! It stunk the car out for ages, and the workshop was little better. If you ever use selenium devices, be very careful not to fry them!
I would suspect that the body diode of the MOSFET is a little bit better than the discrete diode because of their difference in voltage threshold AND in their recovery time (the time for which the diode still conduct while freshly moved into reversed bias, which can reach the milli-second). A discrete silicon diode has a slower recovery time than the body diode and itself, lower than a Schottky. If the power is too high for a Schottky, it is recommended by the literature to use SiC (Silicon-Carbide) diode.
You should check also a fast switching silicium diode (1N914, 1N4148) versus a rectifier silicium diode (1N400x), there should be a noticeable difference, with a better efficiency for the 1N914.
have u checked out "interleaved" buck/boost converters? that was the magic search term for me when i was building my solar mppt circuit
Looks interesting - I'll have to do some reading.
What frequency is this thing working at? Might be interesting to try some ultrafast pn diodes (e.g. UF4007) and see how they compare to the MOSFET body diode. Also, as it has been stated already, please add some smoothing capacitors at the output.
So much engineering paranoia around switching converters that I almost expected this setup not to work decently at macro scale. Turns out this kind of old school experimental demonstration setup is still a great way to visualize. Looking forward to going synchronous in part two!
There is one thing you missed. The schottky diode switches a lot faster. I guess your switching frequency is a lot lower than most SMPS. That's why schottky is not much different than normal rectifier. But once you get to a bit higher frequencies - let's say 50 kHz (which is still relatively low), a regular diode won't be able to do the job. I've tried it at 100 kHz - it just doesn't work.
Very good Julian. Watched at ‘eafrow on the way home 😀
Inbound or outbound?
Julian Ilett outbound to the warmer 😀
what is this? are you getting fancy with graphics in your videos now? are we soon to see realtime rendered schematics ?
I'd find it useful if you could briefly summarize what you've got set up and what you're trying to do with it at the beginning. Sometimes in the process of trying to work out what everything is, I miss relevant portions of the presentation.
I am guessing the body diode is quite fast turn off so the dissipation will be lower.
Is this really accurate, or did i miss something in my RUclips electronic classes? I thought parallel diodes will always have less forward drop than the value of the smallest diode in the circuit. Like resistors do in parallel.
Great Sir.
are you going to eventually sell this? maybe in kit form?
Why not use another MOSFET as a rectifier? Not the body diode, the actual MOSFET. You have a floating gate driver, it'll work. Great efficiency, too.
Your first case with a high voltage diode is not a good example of a silicon diode in this converter. Typically, such a converter would use a low voltage, fast, reverse recovery diode.
ideal diodes?
that would essentially be the same as synchronous but what I want to know is if there are any that will work with high frequencies for PSU switching applications.
I just remember Julian playing about with ideal diodes a long time ago, I think in a postbag vid and there was some issue with the ones he got, but they never reappeared for further investigation
"ideal diodes" are not real components. There are ICs that act like an ideal diode, but they´re basically just a small op-amp circuit that needs power and cannot offer any useful current capability, and even if they could they would be horribly inefficient.
im pretty sure the ebay boards use mosfets in a "battery protection" arrangement N-channel and P-channel in anti-series to each other
I looked up the video I'd seen it in, it was the last item in postbag #74 way back in Nov 16. the one that was shown had 2 n-channel fets in a configuration that didnt seem to do anything. it was just a thing I'd never heard of before
why don't you use o low side switch for your Muppet?
Ilias because to get easy feedback circuit with common ground
Not easy by common positive side
Well...all he needs to know is if the battery is more than 12V(or whatever the maximum voltage is); which he could easily do with an optocoupler...That mosfet driver does not look very promising...
"Not easy by common positive side"
birds-are-nice.me/ipfs/QmeHTKsAWhaKeygAqRBkAUkt5t2e9NPZcHT2ZcuyZcXcwy/solar_optimiser_schematic.png
I made a Muppet of my own months ago. Common positive isn't that hard. Trick is to measure the inverted output voltage, then subtract it from the input to get true output. Common positive may be the way to go - it certainly simplifies drive. No messing around with opto-isolators.
Yea... That is a way of doing it but there will always be a current path through the voltage divider you used to measure the output voltage... Which basically will charge the battery even if the mosfet is closed... I am not that experienced in battery technology but that does not seem like a good thing
If you use a load on the output which takes care of the overvoltage... That would be beter I think
На який ток розрахований діод ?
yes
Now try putting 3 schottky diodes in parallel, not to greatly increase "safe" current capacity but instead, lower forward drop at any current that's a significant % of the diode rating, since ultimately the diode rating depends on the heat generated in that package size, which is based upon the forward drop at that voltage (and of course, heatsinking in a package that allows for it. Purists would hate the idea of 3 diodes in series but stop to consider how inexpensive a bag of diodes is these day and how precious efficiency can be on battery or solar power.
ohh… cliffhanger…
Yeah, I'll explain why in the next video :)
I can't wait that long!!!!!!
🎇BOOM!!!🎇😜
sure there's a reveal coming up...
*possible spoiler below*
:
:
:
:
10sq045 *IS* a schottky according to datasheet I found on the internet.
is it just me or is Julian giving us the finger when he points!
u ever notice Julian is always flipping us off.
and where is the Germanium
@2:40 "EFFIENCY" That's my kind of spelling...sadly.
Oops :)
nice
The wife turned up :)
With no diode you probably damaged your MOSFET.
Fortunately they survived :)