Splitting the topic here since this is not Lepton anymore but a completely new board. Keeping with naming, QVADRANS (worth two Leptons from Antiquity c.a. King Alexander Jannaeus of Judah from 103-76 BC).
Got a little time today and generally finalized the layout, pinout and component selection. Cost per board before shipping is $14/board for a batch of 10, I believe this is a very favorable value compared to B-G431b-ESC1 with very similar parameters.
I have no idea when I will finish this, my time is of very short supply.
nice! is it better to post an easyeda link or it’s not done enough for that? I’m definitely committed to using this kind of board, I don’t need nearly as much current capacity but that part isn’t very expensive. I don’t think I should try the mosquito because the processor on the lepton really was too slow and small unfortunately and I would fear similar issues with that. I need the fast g431 with it’s m4 core.
I have found uses for current sensing, for safety in the event of motor stall and other systems malfunction, and also for measurement of motor characteristics and some other stuff, plus the possibility of sensorless drive. I think sensorless drive is underrated, when the motor and sensor are a single unit you eliminate some important error sources and complications, not just cost. I think it is percieved as the cheapie approach but it’s not really, actually. Especially with IPD and stuff that works at zero rpm. Imo it’s the future in many ways, it’s largely a challenge of figuring it out, not from a technology standpoint but just understanding existing publications and stuff and writing some open source code to make it happen, and some adequate current sensing. To be fair, I am unclear on the precision etc. needed by the current sensing, so I would either over or under specify the characteristics… that’s a problem. But just leaving space on the board and some pads for shunt resistors, and wiring them to the op-amps and stuff on the g431 chip seems like a reasonable hedge. The b-g431-esc1 board seems to work ok.
The board is fully routed. Need to finish the silk screen, but electrically it’s complete. Will need to manufacture and test unless others would be interested in risking an order to test.
A batch of 10 would cost $11.80 per board (excluding the headers and shipping) which is really impressive considering you get functionality similar if not better than the $20 STM subsidized B-G431b-ESC1 board. All components in JLC stock.
30A max thermal and copper limited (60A MOSFETs)
Inline current sensing
All pins exposed
Split power and digital ground (reduce noise)
Amazing, I did say I would be willing to pay for a prototype order, but I would have to double check it meets the major points I need to meet and possibly make minor modifications like an indicator led or whatever. To be honest my ERV project is taking a significant detour and I might be selling all the tech and never actually make any units so this is not a priority, but I can throw the $ on the prototypes in order to keep my promise. I was foolish to order all those g431 boards, I could have saved the money and spent it here instead. There was a change of plans, can’t always know…
We never figured out why the lepton 2.0 had noise issues, in order to prevent it from being inherited here, I think it might have been a manufacturing issue, a cold solder joint somewhere. I would like to test the other leptons I have, if I can get a minute. Runger, you had some leptons, did you never notice any noise issues with serial port or interrupts? Interrupts would spontaneously fire at high frequency and only 14/15 characters got through the serial port. This later got even worse till the serial port didn’t really work at all, which makes me suspect a manufacturing issue.
Fair enough, thanks for weighing in. I also found it to execute code without issues, the problem was getting info in and out, regarding the noise. My plan of attack to clear this up is to inspect the pcb closely and also to test as quickly as I can the communication over UART. I have to resurrect my test rig with the raspberry pi pico though, or maybe I can just write a python program. I think it is likely to be a manufacturing issue, but that is only a suspicion. I would prefer to nail any suspects if it were me doing the Lepton 3.0, for instance I would replace the buck regulators with linear regulators as they are powerful oscillators, but I seem to remember Valentine indicating his actual experience indicated they weren’t an issue.
@Valentine - I’ll be tracking the testing on this board. I really wanted to like the stm32 esc1 dev board but it was far too fiddly.
Do you have schematics? What did you design it in?
I’m working on my own gimbal board but would be happy switching, if this ticked boxes.
I think what is lacking with (and perhaps the cause of) all these custom boards is the lack ability to influence the design.
This to me might be useful as a dev board but I think I’m looking for something a bit more opinionated on the ports e.g. whilst all the pins are broken out, wiring CAN bus or even rx/tx and jtag requires referring back to both the schematics and to the datasheet for the chip. Also I eventually want to wire 6 of these together and control through can and wiring might get messy.
Perhaps we should be collaborating more on standardising on common chip and pin choices. I’m using similar pins for pwm and same chip. Additionally we could co-brand two variants (qvadrans-power vs qvadrans-lite).
Yours is more flexible, for example my motor mounting holes and onboard sensor limits the type of motor it will be attached.
It is surprisingly difficult to collaborate when the use cases can be so different!