so i ve been fairly busy recently designing a servo controller for my hexapod. the new board is currently sporting the following:
USBizi 144 main IC
24 possible servo ports
8 digital ports
8 analogue ports
1 ttl com port
1 i2c interface
USB host
USB client type B - debug
USB Command type B - Serial interface incorporated ftdi232rl
ability via jumper to use seperate power for servo or linked to standard power rail
ability via jumpers to power logic from either usb ports
below are screen shots of the top and bottom layers.
currently i intend to have a prototype built of the board ill then populate it and run some trials. currently im wondering if there is any interest in a board like this as the reason i have designed the board is to essentially build an ssc32 alternative using netMF, rather than having multiple boards to control servos.
Can you make this in the same form factor as the p.Brain board from Micromagic, so it will be a drop in replacement for their hexapods?
Your power traces need to be MUCH bigger. The metal geared servos i use in my hex can stall out at 3A EACH. This would create a smoking trench of your traces if more than one servo stalled
Create a ground plane polygon, it looks like you’re just using traces to ground points. Make sure the top/bottom ground poly’s are well connected with via’s. This will help with electrical interference and any chance of cross talk.
If you need any FT232RL chips cheap let me know, i have a few hundred floating around
Mark is absolutely right. By using 18 servos for my hexapod, I need a 5 amps power supply. Due the fact that they are not high power servos. The ones on order are, (metal gear as mark says) very powerful and I will track amps once they arrive.
in fact i had already beefed up the power rails and sorted out the ground planes,
interesting you mentioned the p.Brain board from Micromagic, as that where i originally found out about hexapods, regarding the form factor it may well be possible to shuffle the design about to accomodate this, however at the moment i intend to stick with the current layout, this is only because i want to debug out any major design issues oh and i have a couple of prototypes already in manufacture, once i get these back i will solder up the components do some code testing [most of which i have already proved via verabord and my domino].
the intention is to then to see how much ik calculation i can run with the servo code running as well.as i have not decided whther to use the board as a complete hex solution or just a netmf servo controller or i have it in the back of my mind to swap out the usbizi for a emx module.
either way my ultimate intention is of course to get my custom hex to work and secondly i intend to free release the gerber files to the fez community.
this is partially why i included a 3rd usb port and ftdi232 chip to enable the board to receive serial commands and essentially be plugged into anything capable of running a usb to virtual serial port, be it a domino, cobra, windows or whatever. oh and to leave the api code open for people who like the hard way.
as this is just afun project for me evenings mainly i don`t currently have a timescale and works keeps getting in the way but thats my plan.
so again thanks for ther feedback and any other comments are greatly received
You might not want to spend hours and hours of work on developing hex patterns, if a controller like the ssc-32 can do that. I believe it would be better to develop something which would make the use of “regular” robots more effective? Since no one really focussed on that as far as I know…
Hi.
This looks really good, and if I was not half way through implementing s solution with a FEZ Mini and a SCC32 I’d be very interested.
Only suggestion I’d add would be to have an SD socket or a wayto plug in a USB drive. That way you can store your sequences in a file and play them back through the micro.
I’ve been playing with a panda and a pololu micro maestro servo board. I have had problems with continuous rotation servo’s. They cause the regular servos to jitter. If I put 2 1 uf ceramic caps across the motor terminals that stops it for the most part. Problem is I can only fit .1 uf caps in the casing. Adding a 100 uf across the power lines solves the problem until I get more than 12 servos. You may want to add a 1000 uf at the power entry to the board and a couple 100 uf at different locations by the servo connectors. Have you tried running 24 servos with the output compare? Can it handle that many?
