On Wednesday 11 January 2017 00:29:01 Peter C. Wallace wrote: > >> AFAIK there are no significant issues with HM2 encoder velocity > >> estimation > >> > >> Heres a 100 Hz quadrature velocity trace: > >> > >> http://freeby.mesanet.com/100_Hz_quadrature_velocity.png > >> > >> Notice theres no discernible ripple in the velocity even though > >> theres considerable sample jitter (green trace) > > > > Show me that same trace out of a 7i90 running on the spi bus driven > > by the rpi 3b. > > I expect it would be the same since all hm2 encoder counters are the > same regardless of interface, and AFAIK the encoder driver code above > the data transfer level is the same for all interface methods. > > If you wish to determine if your error is driver/SPI hardware _or_ > simply bad quadrature you can use a hardware stepgen in velocity mode > looped back to the encoder counter since it generates basically > perfect quadrature. > I'll see if I can rig my hal file for that tomorrow (later today actually). Some sort of a set of mux2's so I can dynamically switch it back and forth.
The encoder is a triplet of ats667's watching a 6" diameter, 60 tooth gear, so thats 240 transitions a second*rps. The data sheets say the normal duty cycle is between 48 to 49 % as a result of the internal gain controls that hold the switchpoint pretty constant in the face of and eccentricity in the target. I have dialed that gear, and its eccentricity is less than a thou. It doesn't say what effect any residual magnetism of the steel of the gear might do but I tried to neutralize that by bringing a mag tool base such that it alternated as the gear turned, close enough to effect that, then slowly removed the tool base so the mag field faded leaving the gear as demagnetized as it was going to get. No effect that I could detect. The lack of a 50% symmetry in the wave form is unavoidable when using the 667, and when I remachined the mount, I tried to get good spacings, and my scope confirms at most a 2 thou error in a 90 degree quadrature. Thats not much but I'll confirm that on the analog scope tomorrow before I start mucking around in the hal file. Does the stepgen in that mode also generate an index? And should I switch that too? I have to assume no, its not mentioned in the manpage. >From the manpage I gather I should set a base thread since a step change at 1 kilohertz would be /4, or 250 teeth a second /60 teeth=4.16666666667 rps. Thats about as slow as it will run. ;-) I also think that latency in the generated waveforms would be a self compensated error, and invisible to this test. It sounds like to do this right, I'd need a dual channel Arbitrary Waveform Generator. The last time I priced one of those, its was about 10x my paygrade. But 10 years later a dual channel DDS version that can run at 6MHz is now around $55 USD, but of course 3 weeks away, from China. I have some hal code that will do a /4 on the noise, but it would also be 2 to 4 counts behind at a G33.1 reversal. I'll need to establish about a 10 thou backlash in the z, somewhat more than it measures. > If this generates noisy velocity signals then the driver/SPI interface > needs to be looked into. > > Peter Wallace > Mesa Electronics > > > > ---------------------------------------------------------------------- >-------- Developer Access Program for Intel Xeon Phi Processors > Access to Intel Xeon Phi processor-based developer platforms. > With one year of Intel Parallel Studio XE. > Training and support from Colfax. > Order your platform today. http://sdm.link/xeonphi > _______________________________________________ > Emc-users mailing list > Emc-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-users Cheers, Gene Heskett -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author) Genes Web page <http://geneslinuxbox.net:6309/gene> ------------------------------------------------------------------------------ Developer Access Program for Intel Xeon Phi Processors Access to Intel Xeon Phi processor-based developer platforms. With one year of Intel Parallel Studio XE. Training and support from Colfax. Order your platform today. http://sdm.link/xeonphi _______________________________________________ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
