On Sunday 15 July 2012 23:29:40 Jon Elson did opine: > Gene Heskett wrote: > > Sure there is Jon. Timestamp the last 4 edges so you can develop an > > average speed over the total time period of those 4 stamps. > > Sounds great at first, but then you are basing the control loop on past > history, which makes > it even slower to respond to what is happening NOW. Sure, it will > smooth out the growling > due to a very coarse encoder, but it is a real compromise. > > > Then add the > > > > last few digits for the position it is expected to be at the servo > > threads time. The only thing to complicate that would be a motion > > reversal within that 4 sample period. > > Yup, that would foul things up a lot, if you ever want to do rigid > tapping. The > reversal would be seriously smoothed out and you might break taps or > mangle the threads. > > > Or maybe I don't fully appreciate the problem. No explanation offered > > so far seems to explain why the encoders rps output, at a mechanical > > speed of 20 rps, has +- 3.xxx or more from the 20 its actually doing, > > flickering in the halmeter output. And this 'noise' does not seem to > > be all that much effected by the actual speed regardless of the speed > > as long as its over about 1 rps. > > Well, every sample period will have a guaranteed +/- 1 count jitter. > And, they are correlated, > so if the extra count is in this sample, then the next one will probably > be short one count. > So, that explains jumps of +/- 2 counts between adjacent samples. Yes, > this jitter of +/- 1 > count per sample will remain constant and NOT scale with velocity, as it > is a timing > relationship between when the counts occur and when the position is > sampled. So, I expect that part. > > Now, I'm not too clear on the +/- 3 fluctuation in terms of REVOLUTIONS > rather > than COUNTS, which you seem to be saying above.
That is exactly what I am saying Jon, using the hal encoder module. > I completely expect > jumps of +/- 2 counts/sample, but that should be divided by the number > of encoder counts/rev. If the measured speed is jumping by +/- 3 RPS, It is. > something seems to be really wrong. It is, even when turning 20 rps, I see readings that flicker into view that range from 17.xxxxxx to 23.xxxxxx in the halmeter. But the spindle isn't making any wow wow noises, its dead steady, it can't be otherwise when the chuck weighs a good 4 or 5 pounds, that is one heck of a flywheel. Its a 5" 4 jaw. > Can you look at the encoder > velocity in count units? I believe I have, several times, but IIRC that value never zeros, it is incremented or decremented continuously according to the direction the spindle is turning. I have not been able to see/find a signal with halscope that zeros itself on the index pulse. > (That is ppmc.0.encoder.x.delta on pico > systems boards.) This value DOES jump to zero for one sample whenever > the encoder syncs to the spindle, by the way. > The encoder velocity pin (scaled from counts to rotations/second for a > spindle > axis) is fudged, it retains the previous value for that one sample where > velocity > can't be known when the spindle syncs. > > And, finally, I think you may need to get an oscilloscope on the encoder > outputs to see if they are picking up noise or possibly failing to > follow the pattern on the encoder disc at higher speeds. If these > sensors need a pull-up resistor, then possibly the value of that > resistor needs to be changed. No pullups or pulldowns needed, this particular sensor outputs a very solid, no noise visible in my 100mhz dual trace Hitachi V-1065, logic that runs rail to rail (+-10 millivolts of the rail, its output stage is cmos) in about 15-20ns for its rise & fall times, at any rotational speed including high gears 2500 rpm, or 40 rps. These sensors have some hysteresis so despite looking for it with the lights turned off, I have never seen a 'contact bounce'. Its getting late now, but I'll get my current .hal file where it can be grabbed tomorrow. Maybe I have that foobared somehow. Thanks Jon. Cheers, Gene -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author) My web page: <http://coyoteden.dyndns-free.com:85/gene> is up! "Gravitation cannot be held responsible for people falling in love." -- Albert Einstein ------------------------------------------------------------------------------ Live Security Virtual Conference Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ _______________________________________________ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
