DMM-TECH Analog Tunning

[tmday7]

Hi Tom,
So i made some changes as you said and checked voltages. The voltages looks like the highest is 5V and average around3.5-4v on the Analog screen with DAC highest at 1100 and average about 550-650. This was done running a test move of 5" or 203200 counts.

I have also made some other changes. The drive resolution is now set to 40640 counts per inch.
And after you said earlier about the 886ipm?? And mmurray pointed out the speed for this small of machine. I figured it takes a few inches to reach the 800IPM i was shooting for which means slower ramp up and down.And that it WOULD be best to lower rapid rate and shorten the distance to hit top speed, decreasing the ramp up and down time. Unless iam missing something this would actually make the machine faster, especially on short moves, which is what the most of this machine is.

Then i did some more tuning .Attached is the results of latest configuring and tuning for X and Z.
What do think of the results?

Thanks,
Troy

V4XaxisVel340AnalogDr2000.zip

(410.38 KiB) Downloaded 282 times

V4XaxisVel340AnalogDr2000(mot).zip

(639 Bytes) Downloaded 292 times

V4ZaxisVel340AnalogDr2000.zip

(420.04 KiB) Downloaded 291 times

V4ZaxisVel340AnalogDr2000(mot).zip

(642 Bytes) Downloaded 295 times

[TomKerekes]

Hi Troy,

So i made some changes as you said and checked voltages. The voltages looks like the highest is 5V and average around3.5-4v on the Analog screen with DAC highest at 1100 and average about 550-650. This was done running a test move of 5" or 203200 counts.

That matches the Output levels in the Plots. But I was more referring to the issue where you hold off the ball screw by 5 counts.

Here are the error plots. Not bad. You can probably use Feed Forward to reduce the errors further.

I noticed you have a 1st order Low Pass filter at 500Hz. Normally we use a 2nd order 500Hz Q=1.4 to suppress the spikes in the output a bit more.

I noticed both axes have a strange, small, ~1500Hz oscillation. Probably not a big issue as the Motors/Drives won't respond at that frequency. I suspect that is a quirk with the drives. They might be reading their absolute encoder every so often and output a burst of pulses periodically to simulate an incremental encoder or something.

[tmday7]

Hay Tom,
Forgot to mention that after increasing P and I that the servo holding stiffened up nicely. I can no longer rotate ball screw by hand when ad idle.

When reading the error plot which should i focus as the overall error, the height of tallest spikes or the difference between the 2 lines on the graduation at 0 on left of window? If iam figuring it correct i have between .0002" to .0005". Is this what you get?

The Low Pass filter is what i copied from mmurray's setting he has on his mill.And i have no clue what it does:) Should i add the 2nd order filter or replace the Low Pass filter with a 2nd order?

Do all Absolute encoders only use 4 wires for all A,B and Z channels? DMM calls these encoders digital. Isnt this the magnetic type?

Thanks again,
Troy

[Moray]

Do all Absolute encoders only use 4 wires for all A,B and Z channels? DMM calls these encoders digital. Isnt this the magnetic type?

Absolute encoders output their current position as a numerical value when requested, which for a typical rotary encoder will be in degrees. They don't output ABZ like a typical incremental quadrature encoder. The output format can be serial or parallel, but serial is far more common, and relies on the drive requesting the position regularly.
Digital is a misleading term, as even incremental/quadrature encoders are technically digital, as they are all working on 1's and 0's, just in different formats.

Magnetic types can use an analogue reading method, but the output is nearly always digital.

[TomKerekes]

Hi Troy,

When reading the error plot which should i focus as the overall error, the height of tallest spikes or the difference between the 2 lines on the graduation at 0 on left of window? If iam figuring it correct i have between .0002" to .0005". Is this what you get?

Well the plot is of the error so the value is how much the axis is out of position (according to the encoder). So when the value spikes to +80 that means your axis is in the wrong position by 80 counts. So you might want to consider the biggest spike the biggest error. But it all depends on your requirements. Maybe you don't care about when the axis starts and stops. Or maybe since this is a rapid you don't really care at all as long as it is in position and settled at the end.

I see +/-80 count which would be +/- 80/40640 ~ +/- 0.002 inches

Again Feed Forward should reduce the errors during the accelerations.

The Low Pass filter is what i copied from mmurray's setting he has on his mill.And i have no clue what it does:) Should i add the 2nd order filter or replace the Low Pass filter with a 2nd order?

Replace the 1st order with the 2nd order.

[tmday7]

Hi Tom,
Did some more tuning now that i know i was looking at the error plots wrong. Added the Feed Forward and replaced filter with 2nd order. This filter made motors quieter ,idle and running. Feed Forward helped a lot.
Now i figure it is within .0005". This is what i was expecting out of this machine, before its been hand scrapped and glass scales on each axis.
Attached is of Z axis.
Thanks,
Troy

[TomKerekes]

Nice!