You need to replace the ESD shunt with a 36v minimum and replace the LDO with a buck converter. Also make sure the capacitors are at least 50V rated. If you do not, the ESD will short your power supply and the LDO will overheat. Or, you skip the ESD protection and the LDO when you order and provide external MCU supply. I’ve been meaning to re-work it a little to take care of these but will do it when I’m back from vacation.
Making note of this thread for my hammerbot Darn it @Valentine, you’re everywhere doing everything at once
I am starting to become frustrated since I broke my third B-G431B-ESC1 board (died again in split seconds without noise, smoke or noticeable heat, not even load…) and thus I am looking for some more robust replacement. I only need max. 9A right now at 24V (12V should work too) and that is for short term peak load only, but the next level of my simulator will probably require two additional, much stronger motors. @Valentine, would you rate your Hackjammer as a good replacement? It will cause me some mechanical trouble and I never ordered anything form JLCPCB, but that all sounds doable. If others from inside the EU are interested, we might also join and start a bulk order. All I want is something more robust and hopefully repairable if it breaks.
Sorry to hear that. This is very unusual. I would say it may be useful to find the source however that’s a lot of dead boards. You could also have a bad batch, I mean, you got defective boards to begin with?
The HackJammer has a small design flaw where I need to replace the LDO with a buck if you need 24V, but other than that I believe it will deliver 10A without problem. What is the max current you need for the bigger motors?
Ordering from JLC is your least problem. Make sure first you can fit the board. Secondly, the HackJammer is not very cheap, so you need to be prepared. I believe is about $60 per board.
I can walk you through the order, or you can send me the money and I can order and make JLC send the boards directly to you. However, this is last resort because I am not doing this as business. Message me directly and we can talk if you need.
Thanks for your quick reply @Valentine. All three boards died for different cause, the last one now had (at least) a defective Shottky diode in the 10V regulator, but I am unable to replace it with my tools. I went through all the documentation again and perhaps, the 24V I used were too much, since the board is specified for 6s which would boil down to 22.x Volts only. My bad, especially since I do not need to operate it at 24V. I will now play a bit with my last spare board and if it breaks again, I will probably come back to your very kind offer of guiding me through the order process of the Hackjammer. 12V are no problem and the 60$ is cheaper than burning boards all the time and being unable to fix them. If I had some more EU citizens interested in a bulk order, I would probably go straight for it. Except for the size it looks like a perfect and more user friendly replacement for the B-G431B-ESC1. After all, having mounting holes, SPI and connectors instead of tiny soldering pads is a great advantage! I will try to not overstretch your offer for help with the ordering of course, it is very well understood, that this is plain hobby for you too and making you buy it for me sounds like a clear no-go!
How much current I will need for the larger motor is still completely open and will depend on the final mechanical design. I need to trade realism of the (paragliding) simulation vs. cost/power.
Thanks for all the help and the great designs!
Two more questions regarding the Hackjammer, after reviewing the documentation a bit closer:
The MCU has just 64k Flash, which is a bit low for SimpleFOC and too low for me. I guess I could simply use the STM32F103CBT6 instead (B=128k)?
Is there a particular reason for adding the MCP2515T-I/ST CAN bus controller instead of using the one built into the MCU? At least it looks to me as if there was a CAN controller built in, but peripherals on STM devices are a bit confusing.
I am just trying to understand things right now, not sugesting any changes.
Well, you see… this is where it gets really messy. STM32 is not really what you call a “user friendly” CAN controller. They give you the MCU-side CANBus physical layer in silicon. You still need the transceiver, then you need to code your own driver, then your own application layer on top of your driver. This is like going to a restaurant and they give you a bag of potatoes, a live pig and a knife, then show you where the kitchen is.
Also, your driver will be narrowly implementation specific, down to the exact chip.
Much easier to pay $ and get a generic chip that off-loads everything and you just use it like the way you use Serial or I2C.
I believe someone here implemented a scaled-down version of basic CAN, just search the board. I would not use it though. Using the MCP2515T, you can port your CANBus application on any chip that supports SPI.
In that light it makes absolute sense to have a dedicated controller chip on board. I had a brief look at the description of the on-chip controller and it looks different to the one on the G431 CPU. Should you ever look for a CAN bus library which supports the G431 as well as ESP32 and allows you to use the same code for the CAN ID definitions and application layer on both platforms, here it is: GitHub - Igitigit2/SimpleCanLib (I wrote it, but it is based on other people’s code). If I’ll go for the JackHammer, I will extend my CAN bus lib to support the MCP2515T chip too.