[3.6.0-rc.3] sticky probe with 1HCL board

dc42

@ironhydroxide thanks! I worked out what was probably causing the original problem and I am confident that the new code has fixed that.

I would like to reduce the debounce latency before we release 3.6.0 stable so I'll probably as you to test another build later today or tomorrow if that's OK with you.

PS - do you have any feel for what probe trigger latency would work best for you? Currently it is 224us but I have it in mind to reduce it to 112us.

ironhydroxide

@dc42 I'll gladly do more testing.

As for trigger latency, I have not had the chance to drag my scope in and connect it, so I don't have enough data to even guess at a range.
I will attempt to get it connected today and get some traces, but no promises there.

dc42

@ironhydroxide the latency due to debouncing is 212 to 244us in this version. ~~This is somewhat lower than the maximum CAN latency so I think I will leave it as-is.~~

The effect of the latency is that the probing move will go a little further; but the amount is small. For example, if you probe at 5mm/sec then the extra 212us latency will cause the probe to descend about another 1um.

dc42

@ironhydroxide I've halved the latency to bring it oto line with the 6HC and 6XD boards. I doubt that this will affect your system, however if you wish to test it you can find new binaries at https://www.dropbox.com/scl/fo/vg4yfqcuk1u7oofsiogd6/AMDAS-LZ44linkaLjtrkCAY?rlkey=ptopknaa5xr4nhtacr94t6mw3&dl=0.

ironhydroxide

@dc42 Sounds good, I'll load this last one in and start the test.

I am probing below 5mm/s, though would like to keep the overshoot less than 1um, so that's a good "limit" value for me as I try and increase the speed of the system.

Thanks again for your hard work in this. It really is amazing what the Duet team is doing.

ironhydroxide

@dc42 Ran a full cycle of my test without a sticky probe.

on the start of the second cycle I had a weird thing happen, where the G30 did not return, I was watching the movement and it seemed to be due to my secondary attempt speeds being too slow and the debounce was ignoring the contact (blade moved down, contacted, blade moved up and started moving down slowly, but the probe never triggered again and the system hit the bottom stop)

I'm still calling this probe sticky issue no longer an issue. Again, thank you very much for your work in this.

While i have your attention though, would it be possible to get the encoder counts from 1HCL boards into the object model? or is that a huge change? (would make my tests of skew much easier)

dc42

@ironhydroxide thanks for your feedback.

Regarding getting encoder counts back from the 1HCL boards, unfortunately it is quite a large change involving all f the 1HCL firmware, the CAN message format and the main board firmware. Firmware 3.6 is too close to release for us to consider adding it before then. Perhaps you would like to create a feature request at https://github.com/Duet3D/RepRapFirmware/issues.

dc42

@ironhydroxide I'm sorry to bother you again, however I've had to revert one of the earlier changes I made to fix this issue because it had an unfortunate side effect. I think the issue should be fixed without that change; however I would be grateful if you could test the candidate 3.6.0 firmware for the 1HCL which is now at https://www.dropbox.com/scl/fo/etf8gf0vxx8ro102juita/AHzXcQJL_3Io_1C4LuDwpxI?rlkey=1hbgwbogzy1r44y1897a7kqds&dl=0.

ironhydroxide

@dc42 Sorry for the delay, I've been out of town the last 4 days.

I have loaded the provided build and have ran over 10k probes without a sticky probe issue.

I'll continue to test(mostly because of skew i'm seeing), but it looks like you got the sticky probe issue whipped.

Thanks again!!

dc42

@ironhydroxide thanks for the confirmation!