Sorry for the delayed reply. I did some work on the scope on Sunday morning and have attached some pictures of the results. Let me know if they don’t come through.
In sort, the results appear to show a roughly 80kHz noise when the machine is turned on, with spikes pretty frequently exceeding 2V (where my trigger was set for channel 1, the index channel. Channel 2 was the A channel of the encoder and the noise also appeared on there. Results were identical regardless of whether the encoder was connected to the USC board or not. Spindle on/off appeared to have little or no impact on the noise. I also got a picture of the Hal Scope for the index pin and it is triggering repeatedly as discussed before. I wasn’t able to reproduce the sawtooth on the motion.spindle-revs during G33.1, but it was still randomly failing to reverse and retract out of the hole occasionally. I ran a program with numerous cycles and it’d e-stop the machine after a few G33.1’s for reasons unknown. The encoder wires are twisted pairs and they’re only separated to allow connection to ground. Given these findings, it appears that I’ll be rewiring the grounds to a single point in the near future. I’ll re-test once that is done, but it’ll be a few weeks. I was looking it over and while I didn’t connect everything to a single point, I did do a fairly thorough job of making sure each one was on a separate bolt on the case of the respective enclosures. I’m wondering if there’s any value in isolating power and logic grounds or isolating either one from the machine. > On Aug 2, 2020, at 11:34 AM, Jon Elson <el...@pico-systems.com> wrote: > > On 08/01/2020 01:37 PM, Matthew Herd wrote: >> Thanks Gene, I hope you’re well also. I disconnected that grounding wire, >> no difference observed in the ppmc.0.encoder.03.index behavior. The noise >> seems the same both when spindle is running and stopped, with a tendency >> strongly toward "true" than "false." Pulses seem to be both long and short, >> but I’d guess they’re about 80-90% true. Not at all what I’d expect, even >> with noise. >> >> > This does not make any sense at all. The encoder.xx.index reports detecting > a falling edge on the index (Z) input. it is reset every time it is read. > Long true pulses mean it was triggered at least once every millisecond, ie. > faster than the servo thread samples it. > > So, the encoder.xx.index is NOT looking at the STATE of the Z input, but > whether a high-to-low transition occurred since the last time it was read. > > Jon > > > _______________________________________________ > Emc-users mailing list > Emc-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-users _______________________________________________ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
