Hi Clem! Always a pleasure to follow your projects! Anyway, diodes to put batteries in parallel are a common method also on professional project! On trains, in one of my project, there are three 24v 170Ah batteries in parallel, protected by (slightly bigger) diodes! I just also designed a CM4 carrier board for a retropie console (video soon on my italian channel) but i've choose to use an already functional power bank breakout board! One cell, fast charging and stable 2.5Amps of current :)
I was meaning to do that with a libre computer/raspberry Pi and a case made from laser cut steel with 3d printed bits. I have had other pressing projects which havd gotten in the way but I think you should act on your wish and build your own computer! If you could share your design on the internet I would love to see how you went about making it.
If only there were a couple of videos out there that stress how hard SMPS design (and validation!) is! Still great to see where you are at with this project. I had a feeling you were building up to something with your previous videos.
Hey folks if you are in the know about power supply design and BMS please share your tips on the Element14 community page linked below the video and post helpful links, videos or ressources! hopefully we can get this thing eventually working with the power of the community! YT does not like links and long form discussion with pictures, thats why we have a dedicated page for each video!
For last 2 years I have been trying to understand/build a bms for a laptop around pi4 with almost no headway. Thank you for sharing that it is really really hard. I feel better after watching this vid 😊
If each cell has its own protection (and the flatpacks usually do) and they're all the same chemistry, you could charge or discharge them to approximately the same voltage and just connect them directly in parallel; they'll work like a single large cell.
Thats obe of the things, i mostly dont trust any on pack protection or use bare cells, and balancing still needs to hapoen whike at thexsame time i want to get 3v3,5v and 12v the complexuty adds up quickly. And one should necer assume even discharge across cells, in practice it rarely holds without active balancing
Hi! Compliments for the idea and design, yes, you made a brave decision by designing the final PCB at the first shot! I'm not an electronic engineer, but I like these projects 😊 Anyway, I've got two tips, maybe they can help: 1- why not make a unique step-up converter to generate the most high voltage needed, and then add some step-down regulators, to get every other voltage? (Like 5v, 3.3v, etc..). I don't know if it's a good idea, but I would try to connect the cells to the step-up converter in parallel, in order to drain the power more evenly. 2- Split the main board in sections by their purpose (power, usb/data, audio/video?), then find a connector to join them, even with a flex cable. A single bus maybe would work, including power and data. By following this method, in case of errors (or upgrades!) you don't need to replace the whole thing. Prototyping a module on a breadboard is also easier! The key in every engineering project is moving forward by steps. Doing all in one makes everybody anxious, as you have a single attempt!
Loving this! It's something I as a fellow self-taught hobbyist also have been considering too, trying to building a laptop. Kinda curios as to why not use a dedicated 5S battery charge controller, such as bq2461x from Texas Instruments and equivalent. And use USB C 19V input? then step down from there for the required voltages. The converters look sorta not closely enough packed - you want all components as close as possible! Even to the Pi module. Generally when trying for the first time a new converter I like to add 1 extra of each footprint if I need to add extra caps or something later on. Also double check the calculations for correct capacitor and inductor values and keep in mind tolerances of the used e.g. resistors
Amazing video and amazing effort on trying something that is definitely not easy! It was interested to learn the power supply is the most difficult with the most details for understanding and getting to work perfectly. Also cool to learn how circuits behave differently with a load as that may become not as intuitive. Browning out is interesting to learn more about for myself. Very cool to share your process!
This is excellent!! this pointed me in the right direction with the battery side, but more than that the honesty about the process has given me a boost of inspiration. i love this kind of video.
I think the issue is constant power from the boost converter, due to which voltage increases and current decreases with more voltage ripples during the load. Read the boost converter IC datasheet Min and Max parameters. Try to use a buck converter with 2s or 3s lipo battery pack.
That person who patents what Lenovo marketet 2013-2018 but couldn't get to work would be rich. Dual battery, where internal one holds charge for 30-60 minutes after the external goes out of charge. Also for projects like this MNT Reform laptop could teach a lot. Open hardware & firmware.
@@MAYERMAKES YT channel "Electronics Repair School" taught in his laptop video why such a circuit is so hard to make. The reason comes straight from electromagnetism physics.
Wow, this video is really motivating😃. I'm not familiar with the raspberry pi but is it possible that it requires some sort of power sequencing like how an fpga requires? Reason for power sequencing is to ensure the soc/som turns on properly so that its function can turn on accordingly. Just my 2 cents😄
Oof this was painful to watch, I admire the attempt, but as an electronic engineer that does very similar things daily i have a couple of suggestions. One thing is for sure you can by no means remove the diode from the SMPS, its part of its core working principle. Also, with the video output, I have a feeling it could be signal integrity and I am just curious what type of video signals those displays use?
There are multiple sets of diodes, the ones integral to the smps where left alone. please share you pro input on our community page , we can all learn from a pro.
the cm4S is actually just for legacy compatibility with Cm3 projects, and only available in higher MoQ so the Cm4 for mfactor still holds as the new generation. looking back the dual mezzannine conenctor might not have been the most sensible choice.
@@MAYERMAKES I know but it was the only comment I could think of that would fit the HAL9000 quote into. Nice work and as Bryant Addams once sang "Dont give up"
Ok, so, a few questions in my head after watching this episode: 1. Why not use a Raspberry Pi 4b with 8GB of RAM and an HDMI output? It's your first laptop, so at least would be useable ;). Than you can start to redesign thing from scratch one by one. 2. An LVDS is more of a TV type from mainboard to t-con or an LCD panel. In the low power devices like phones/tablets, we have different standards. You can google more about them. 3. Why not use in a first project ready made step-up and back converters? This way you have a working laptop quicker :). Same as point no.1. 4. BMS you can buy cheap for 18650's on Aliexpress for pennies. I could recommend you using from some old laptop :D, but, this can go sideways ;). I'd avoid those flat batteries, but focus on some outdated and used 18650 from old laptops which have some life in them. And, a spot welder you'd need which is cheap on Aliexpress as well. 5. Why not forget about the case until the whole thing is not working? And, PLA first, not a resin. I have a resin printer and there's a difference in between them. So, unless you have a ready design, no point to waste resin IMHO. I've got more, but this is already a lot :D. So, if you'd like to talk more about this project, my contact details are on my YT channel here. Thx.
I know how to build a rasp pi killer Just take a Intel atom line or similar cpu, add 4-8gb ram and pcie, and add a gpu with 1080p support, with its own 2-gb of ram, and share work ram with gpu dynamically Add a cpld to handle fast gpio and an ac'97 audio chipset A sbc that runs windows natively and all parts have windows drivers available from manufacturer And can run most games not at full settings, depending of the specific cpu, single, dual or quad core with hyperthreading. 32 or 64 bit
Wow. I think boost converters are one of the simplest things you can design. I made one by letting an avr microcontroller (arduino software, but bare chip) put out a fixed PWM waveform, feed that into a mosfet gate, add schottky diode, inductor and output cap and it works, until it overvolts. But for that purpose I connected a zener diode from output voltage to an IO pin (with pull down resistor). Now it goes digital HIGH if voltage is high enough, otherwise LOW. The microcontroller simply switches the PWM on and off according to that pin state. Why not use the adc to measure output voltage? I think that would be too slow (or would need a huge output capacitor) and the output voltage would overshoot more and be more unstable. btw that controller also does battery discharge protection by occasionally measuring VCC and if the battery goes too low, it also switches off the PWM. And if you want high efficiency, always use overkill inductors. I got 98% at a 18V->48V converter using a totally overkill inductor (it does not get warm even one bit at my definition of 100% load, while output schottky is pretty hot and mosfet is lukewarm)
Hi Clem! Always a pleasure to follow your projects! Anyway, diodes to put batteries in parallel are a common method also on professional project! On trains, in one of my project, there are three 24v 170Ah batteries in parallel, protected by (slightly bigger) diodes!
I just also designed a CM4 carrier board for a retropie console (video soon on my italian channel) but i've choose to use an already functional power bank breakout board! One cell, fast charging and stable 2.5Amps of current :)
Grest to hear my approach is not as unprofessional as I thought😂
Beautiful boards. Looking forward to when you get it working!
Thx😂
This is a very ambitious project, Clem! I wish you the best of luck and can't wait for a part 2!
Thanks
RUclips is currently reading my mind and I was thinking of building the laptop from scratch and this popped up 😮
Hurray to the almughty algorithm I guess😂
I was meaning to do that with a libre computer/raspberry Pi and a case made from laser cut steel with 3d printed bits. I have had other pressing projects which havd gotten in the way but I think you should act on your wish and build your own computer! If you could share your design on the internet I would love to see how you went about making it.
probably picked up from whatever you said..
the alphabet boys been doing that since 2016
If only there were a couple of videos out there that stress how hard SMPS design (and validation!) is! Still great to see where you are at with this project. I had a feeling you were building up to something with your previous videos.
It was worth a shot, but that curve is hard to climb
Yes, I was also thinking there were some prior E14 videos on switch mode power supplies. PCB layout is importanr if I remember
Hey folks if you are in the know about power supply design and BMS please share your tips on the Element14 community page linked below the video and post helpful links, videos or ressources! hopefully we can get this thing eventually working with the power of the community! YT does not like links and long form discussion with pictures, thats why we have a dedicated page for each video!
For last 2 years I have been trying to understand/build a bms for a laptop around pi4 with almost no headway. Thank you for sharing that it is really really hard. I feel better after watching this vid 😊
I feel you.
If each cell has its own protection (and the flatpacks usually do) and they're all the same chemistry, you could charge or discharge them to approximately the same voltage and just connect them directly in parallel; they'll work like a single large cell.
Thats obe of the things, i mostly dont trust any on pack protection or use bare cells, and balancing still needs to hapoen whike at thexsame time i want to get 3v3,5v and 12v the complexuty adds up quickly. And one should necer assume even discharge across cells, in practice it rarely holds without active balancing
Hi! Compliments for the idea and design, yes, you made a brave decision by designing the final PCB at the first shot!
I'm not an electronic engineer, but I like these projects 😊
Anyway, I've got two tips, maybe they can help:
1- why not make a unique step-up converter to generate the most high voltage needed, and then add some step-down regulators, to get every other voltage? (Like 5v, 3.3v, etc..).
I don't know if it's a good idea, but I would try to connect the cells to the step-up converter in parallel, in order to drain the power more evenly.
2- Split the main board in sections by their purpose (power, usb/data, audio/video?), then find a connector to join them, even with a flex cable. A single bus maybe would work, including power and data. By following this method, in case of errors (or upgrades!) you don't need to replace the whole thing. Prototyping a module on a breadboard is also easier!
The key in every engineering project is moving forward by steps. Doing all in one makes everybody anxious, as you have a single attempt!
Loving this! It's something I as a fellow self-taught hobbyist also have been considering too, trying to building a laptop.
Kinda curios as to why not use a dedicated 5S battery charge controller, such as bq2461x from Texas Instruments and equivalent. And use USB C 19V input? then step down from there for the required voltages.
The converters look sorta not closely enough packed - you want all components as close as possible!
Even to the Pi module. Generally when trying for the first time a new converter I like to add 1 extra of each footprint if I need to add extra caps or something later on. Also double check the calculations for correct capacitor and inductor values and keep in mind tolerances of the used e.g. resistors
Mostly because i tried to iterate in what i usually do. But in hindsight the way you desctibed is a good option to keep in mind.
Amazing video and amazing effort on trying something that is definitely not easy! It was interested to learn the power supply is the most difficult with the most details for understanding and getting to work perfectly. Also cool to learn how circuits behave differently with a load as that may become not as intuitive. Browning out is interesting to learn more about for myself. Very cool to share your process!
thanks, its all about the process!
You could use the 24bit DPI RGB from the GPIO to drive a RGB LCD display!
a possibility to explore!
Simply kudos for trying and realizing: the very power supply design for this endeavor: not easy (leave alone trivial).
you never know until you try.
This is excellent!! this pointed me in the right direction with the battery side, but more than that the honesty about the process has given me a boost of inspiration.
i love this kind of video.
Thanks, that is what i hoped for!🎉
I think the issue is constant power from the boost converter, due to which voltage increases and current decreases with more voltage ripples during the load. Read the boost converter IC datasheet Min and Max parameters. Try to use a buck converter with 2s or 3s lipo battery pack.
That person who patents what Lenovo marketet 2013-2018 but couldn't get to work would be rich. Dual battery, where internal one holds charge for 30-60 minutes after the external goes out of charge.
Also for projects like this MNT Reform laptop could teach a lot. Open hardware & firmware.
hmm interesting idea , reminds me of the backup battery the atari portfolio had to retain ram memory when you had to change the AA batteries.
@@MAYERMAKES YT channel "Electronics Repair School" taught in his laptop video why such a circuit is so hard to make. The reason comes straight from electromagnetism physics.
Wow, this video is really motivating😃. I'm not familiar with the raspberry pi but is it possible that it requires some sort of power sequencing like how an fpga requires? Reason for power sequencing is to ensure the soc/som turns on properly so that its function can turn on accordingly. Just my 2 cents😄
It does but it happens internally. This requires the 5v to be really stable.
Oof this was painful to watch, I admire the attempt, but as an electronic engineer that does very similar things daily i have a couple of suggestions. One thing is for sure you can by no means remove the diode from the SMPS, its part of its core working principle. Also, with the video output, I have a feeling it could be signal integrity and I am just curious what type of video signals those displays use?
There are multiple sets of diodes, the ones integral to the smps where left alone. please share you pro input on our community page , we can all learn from a pro.
*HAL9000 voice* I'm Sorry Dave, there are two different CM4 units from RPI now. The CM4 and the CM4 SIL!
the cm4S is actually just for legacy compatibility with Cm3 projects, and only available in higher MoQ so the Cm4 for mfactor still holds as the new generation. looking back the dual mezzannine conenctor might not have been the most sensible choice.
@@MAYERMAKES I know but it was the only comment I could think of that would fit the HAL9000 quote into.
Nice work and as Bryant Addams once sang "Dont give up"
@@AJB2K3 and as Skinless once sang: life sucks and then you die.
Ok, so, a few questions in my head after watching this episode:
1. Why not use a Raspberry Pi 4b with 8GB of RAM and an HDMI output? It's your first laptop, so at least would be useable ;). Than you can start to redesign thing from scratch one by one.
2. An LVDS is more of a TV type from mainboard to t-con or an LCD panel. In the low power devices like phones/tablets, we have different standards. You can google more about them.
3. Why not use in a first project ready made step-up and back converters? This way you have a working laptop quicker :). Same as point no.1.
4. BMS you can buy cheap for 18650's on Aliexpress for pennies. I could recommend you using from some old laptop :D, but, this can go sideways ;). I'd avoid those flat batteries, but focus on some outdated and used 18650 from old laptops which have some life in them. And, a spot welder you'd need which is cheap on Aliexpress as well.
5. Why not forget about the case until the whole thing is not working? And, PLA first, not a resin. I have a resin printer and there's a difference in between them. So, unless you have a ready design, no point to waste resin IMHO.
I've got more, but this is already a lot :D. So, if you'd like to talk more about this project, my contact details are on my YT channel here. Thx.
I know how to build a rasp pi killer
Just take a Intel atom line or similar cpu, add 4-8gb ram and pcie, and add a gpu with 1080p support, with its own 2-gb of ram, and share work ram with gpu dynamically
Add a cpld to handle fast gpio and an ac'97 audio chipset
A sbc that runs windows natively and all parts have windows drivers available from manufacturer
And can run most games not at full settings, depending of the specific cpu, single, dual or quad core with hyperthreading. 32 or 64 bit
Add external USB 3.0 support via chip or just use USB 2.0
And an ethernet chip
Use a 4 layer pcb and populate both sides
You just described the lattepanda delta , which did nit kill the rpi and wont, but its actually a part of an upcoming project.😂
Wow. I think boost converters are one of the simplest things you can design. I made one by letting an avr microcontroller (arduino software, but bare chip) put out a fixed PWM waveform, feed that into a mosfet gate, add schottky diode, inductor and output cap and it works, until it overvolts. But for that purpose I connected a zener diode from output voltage to an IO pin (with pull down resistor). Now it goes digital HIGH if voltage is high enough, otherwise LOW. The microcontroller simply switches the PWM on and off according to that pin state.
Why not use the adc to measure output voltage? I think that would be too slow (or would need a huge output capacitor) and the output voltage would overshoot more and be more unstable.
btw that controller also does battery discharge protection by occasionally measuring VCC and if the battery goes too low, it also switches off the PWM.
And if you want high efficiency, always use overkill inductors. I got 98% at a 18V->48V converter using a totally overkill inductor (it does not get warm even one bit at my definition of 100% load, while output schottky is pretty hot and mosfet is lukewarm)
I am currently trying to build Ben Eater's style 8 bit computer just waiting for some components to arrive