LinuxCNC-RIO - RealtimeIO for LinuxCNC based on FPGA (ICE40 / ECP5)

@meister

Similarly, I can't make the stepper work. This is NOT a good start for me man, ouchy lol

The dir works on all axis, however getting no pulse from any of them. Can you also share a stepgen config for an axis? Below is what I've blindly attempted, but getting nowhere (as usual!!). 

joints need no signal config
EDIT: better using the rio-setup tool
even i now have problems doing this in pure json

 

Wow. I didn't get a pulse on any axis with the original config, that's why I tried with the signals. Neither config pulses. Snap, now what.

Pulse PIN is always HIGH. Dir pin works fine though!

Can you show a functional HAL file with a functional step/dir stepper config? I'll compare it to mine and see what's not updating in mine.

all flashed ? halcompile ?

check it in rio-test gui (do not forget the enable checkbox)

i'm not at home, so i can not test a config and send you, but it should work

Ah man no rush, you're too kind for real. I'm so not making thousand-dollar parts, just testing this gowin pinouts and timings, and so nothing to worry about.

Yeah gateware flashed, LinuxCNC files overwritten and halcomp'd, everything routine after each generate.

I'll play around with it. Obviously something I'm doing wrong. I'll update if I get anywhere.

i had done a major update, but i just compared one with an old version and everything looks good.
Unfortunately I can't tell you what the problem might be at the moment

PIN 13 (x_pulse) is NOT working on my FPGA. w t f

If I swap pins and place the step on PIN 14, I'm getting steps (!!). Told you the problem was at my end, but I'll have to keep you responsible for this one - guilty!! You picked those random pins

In all seriousness, something's not right with PIN13. Let's hope it's just hardware (bad soldering) and not gowin's internals or their place/routing compiler. Yikes. See? That's why I wanted to test ALL PINS on that thing before sending out some PCB work... don't trust - verify!

if you are playing around anyway, you can also set the enable bits of the stepdir instances to 1 in rio.v

but if that works, something is wrong with the connection and the board is in ERROR state

Quick update:

Changed all the pinouts to the ones that works. So far PIN13 and PIN15 are persistently HIGH. As if they were physically shorted but they aren't (good soldering). Just putting this out there, to be aware.

So far got stepgen, spindle pwm, spindle encoder & index, and home switches. [TESTED]. Everything that makes up for the essentials!! Super happy about that.

Those screwed-up pins are discouraging me though. But I'm not coming down to any cynical conclusion yet. I'll have a look in their datasheet again and see if I can find any reason for those being HIGH on crack...