Compensating for multiple index pulses per spindle rotation

I'm working on a little Taig CNC lathe and for some reason I thought it was a good idea to put a 2:1 belt reduction between the spindle and encoder, so I'm getting 2 index pulses per rotation of the spindle. As a result, there's a 50% chance that each thread I cut lines up with the previous pass.

Is there a good way to compensate in software? I thought about writing a little module to count index pulses and emit one pulse for every two it sees, but that seems hacky and I was hoping there was a better option.

I found this thread:  forum.linuxcnc.org/49-basic-configuration/40459-config-dump where @PCW mentioned the index mask as being used when an encoder is generating multiple pulses per revolution, but I'm not sure how to use it or if the MESA 7i96s supports such configuration out of the box.

Should be easily doable in HAL using a flipflop with one output feeding back to one input.
That should make a divide-by-two counter.

Ha! Thank you, that was exactly the tip I needed. Ended up being a super simple fix.

For anyone finding this thread in the future, here's what I did:

Set up a flipflop:


Then replace the existing connection with a detour through the flipflop:

Hooray! Now my threads will have a predictable number of starts.

Thanks tommylight!

Wait a minute. That didn't work at at. Hmm I think trying to use input-index was probably a mistake.

Well that took way too long to figure out. I don't know if I could have done it with a flip flop in the end because I believe I needed to store 2 bits of state, but the flipflop was confusing me to no end, so I just wrote a new component for n clear deduplication:



and then linked it up thusly:

I don't know if this is the most idiomatic way to do this, so I'd love feedback, but it appears to work.[/code]

You were on the right track....
If i remember correctly, the Q- should be wired to D, clock to index from encoder and Q output to LinuxCNC index.
Do check that, misty on everything lately.
Also use "show hal configuration" from the machine menu to watch the pins.

I think that would have worked if the servo thread was running fast enough to catch the encoder index pulse, but the pulses output by my encoder are so narrow that it pretty much always misses them. As a result I had to use the encoder's input enable interface which just added a lot of complexity.

I tried threading a part again today and got an accidental double start again, so the solution I posted clearly did not solve my problem. The issue I'm realizing now was that I was counting one index pulse for every two that came in, but only once the spindle component had it's index-enable pin set. That meant that I could start counting to 2 on either the 0 degree or 180 degree pulse. 

In order to avoid that issue, I think I'd need my component to be setting the index enable pin on every half rotation, which isn't going to happen if there are 6K index pulses per second.

So I just ordered a bigger pulley. Oh well.