Z axis accumulating negative error (Resolved)

[slodat]

When I operate my Z axis by jogging or through MDI, the more I operate up and down, the more it creates a negative error or offset (not sure what else to call it).

This is my test setup:

The more I jog the Z up and down, the bigger the magnitude the Z axis value is in the negative direction. For example, I jog down until the metrol switch trips. I back off 0.001 and switch clears, 0.001 down and it trips again. At this point I zero G54 Z axis and note the dial indicator reading (1.872). Then, jog the axis up and down and check. The DRO will read a negative value, say -0.002. Jog more, go back down to switch. Dial indicator reads the same value, now the DRO will be a bigger negative value -0.006. Jog more and it goes more negative in this position.

I have done a lot of testing on the Z axis. I **think** it is dialed in pretty well. It has a tiny amount of backlash, measured using Marc's procedure. I dialed in 0.0002 backlash comp. The axis repeats back to the same position after a lot of moves.. like 40. I double and triple checked the axis calibration with the 2" Starrett dial indicator.

What got me looking into this is my parts are coming up small in the Z direction. At first I thought it was my Metrol tool setter. It's not moving and the dial indicator returning to the same dial location has me feeling like that's not it. I did change the max frequency in the wizard to 100khz.

Obviously something is causing this.

edit: I’m pretty confident this is caused by the motor revolutions per inch being slightly off. Got my tenths indicator and a 123 block ready to try in the morning.

[Muzzer]

edit: I’m pretty confident this is caused by the motor revolutions per inch being slightly off. Got my tenths indicator and a 123 block ready to try in the morning.

That can't explain the issue. Regardless of the revs per inch, it should return to the same position. Same number of pulses should be issued by Acorn in each direction.

Sounds more likely that you are losing (or gaining!) some steps between the Acorn and the motor driver. A couple of obvious things to try would be:

* Slow down the step pulse rate eg by simply reducing the feed rate. If you are losing pulses because they are arriving too fast for the driver to register, this might fix the problem, showing what the root cause is. There is also a setting for max pulse rate in Acorn Wizard - does that exceed the max pulse rate specified in the driver manual?

* Are the accel / decel rates very aggressive? You might lose some position if you ask the motor etc to accelerate unreasonably quickly, even if the pulse rate is within spec. You could slow them down and see if that improves the outcome. This would affect an open loop stepper system but not a servo or closed loop stepper.

* Failing that, check you have followed good practice on the signals wiring between the Acorn and the driver, to avoid picking up noise. It's vaguely possible that you could pick up enough noise to acquire additional pulses, although whether this effect would be more active in one direction than the other is anyone's guess. Perhaps some pics would help give an idea of how good the installation is from that viewpoint.

[martyscncgarage]

When I operate my Z axis by jogging or through MDI, the more I operate up and down, the more it creates a negative error or offset (not sure what else to call it).

This is my test setup:

43DEEF79-4B12-43EF-8FAE-F438248DBDD1.jpeg

33F58802-A9C8-4B23-ACDA-B0C49F16013F.jpeg

The more I jog the Z up and down, the bigger the magnitude the Z axis value is in the negative direction. For example, I jog down until the metrol switch trips. I back off 0.001 and switch clears, 0.001 down and it trips again. At this point I zero G54 Z axis and note the dial indicator reading (1.872). Then, jog the axis up and down and check. The DRO will read a negative value, say -0.002. Jog more, go back down to switch. Dial indicator reads the same value, now the DRO will be a bigger negative value -0.006. Jog more and it goes more negative in this position.

I have done a lot of testing on the Z axis. I **think** it is dialed in pretty well. It has a tiny amount of backlash, measured using Marc's procedure. I dialed in 0.0002 backlash comp. The axis repeats back to the same position after a lot of moves.. like 40. I double and triple checked the axis calibration with the 2" Starrett dial indicator.

What got me looking into this is my parts are coming up small in the Z direction. At first I thought it was my Metrol tool setter. It's not moving and the dial indicator returning to the same dial location has me feeling like that's not it. I did change the max frequency in the wizard to 100khz.

Obviously something is causing this.

edit: I’m pretty confident this is caused by the motor revolutions per inch being slightly off. Got my tenths indicator and a 123 block ready to try in the morning.

I wished you would have put this in your original post. I can't remember what you were using for AXIS drives and motors.
If you are picking up or loosing steps, it may be due to noise.
What frequency are you using in the advanced tab? Lower it
Slow down your rapids as mentioned and see if it improves.
Don't rely on a tool setter.

Follow the guide on fine DRO tuning attached and check backlash adjustment: https://www.centroidcnc.com/dealersuppo ... ads/37.pdf
Both can be found in the All in One DC or OAK installation manuals (we're trying to get back on one for Acorn)

Follow the above processes first. They should be repeatable when done correctly.

Marty

[Gary Campbell]

edit: I’m pretty confident this is caused by the motor revolutions per inch being slightly off. Got my tenths indicator and a 123 block ready to try in the morning.

This is an incorrect assumption. Even if the steps per rev or turns per inch were half (or double) the correct number, the machine would return to the same start point

Did changing to 100,000 pulse rate make a difference?

[slodat]

Ahh! Thank you! I was in the shop past bedtime for sure last night.

I’m losing steps. I’ve been through this before with DYN4 and Acorn. The fix was to move to the DB25. I proved it without a doubt when I had DYN4 on my router. I’ll make that change when I get to the shop this morning. And see where I’m at.

In my first post I mentioned I went to 100khz. It didn’t make a difference I could measure. I’m not “relying” on the Metrol tool setter. It is a second method showing repeatability of the issue though. Ultimately I will be relying on it for setting tool heights and I do have more confidence in it now.

There’s a link to my mill build in my signature. Didn’t think about posting a link. My apologies.

I appreciate the advice guys!

[slodat]

Thank you guys so much for pointing out the obvious!

I homed the machine. I use the DYN4 ZRI pulse for as accurate homing as possible with what I have. Then I ran the following g code. Slow moves at 25 ipm.

Code: Select all

g01 x2 y2 z-1 f25
g04 p1
g01 x28 y13.5 z-5 f25
g04 p1

g01 x2 y2 z-1 f25
g04 p1
g01 x28 y13.5 z-5 f25
g04 p1

g01 x2 y2 z-1 f25
g04 p1
g01 x28 y13.5 z-5 f25
g04 p1

g01 x2 y2 z-1 f25
g04 p1
g01 x28 y13.5 z-5 f25
g04 p1

g01 x2 y2 z-1 f25
g04 p1
g01 x28 y13.5 z-5 f25
g04 p1

g01 x2 y2 z-1 f25

I have a significant amount of lost steps:

I'm open minded about it being noise or some other explanation.. however, my experience last year taught me that the DYN4 drives do not like to work with the screw terminals. Yes, I'm using DMM's expensive cable made for this purpose. I will be removing the resistor and connecting +5vdc to the power supply and step/direction to the DB25. I bought a DB25 break out board when I ordered everything in for this conversion. I'll report back.

[mick41zxr]

Slodat,
I had the exact same issues as you & ended up running via the DB25 port with differential drivers. I found this this to be most repeatable.
I am also currently testing CNC4PC C86 board with some success but still early at this time, however it needs 12v supply. Even though its supposedly rated for 24 I & others have had some issues at 24v.
Regards,
Michael

[slodat]

I removed the resistor from the DMM DB25 cable, connected the white wire (was brown in previous version of this cable) to +5 vdc, step and direction to the Acorn DB25 outputs, changed the GEAR to 2048, steps/rev to 8192, max frequency to 400 kHz and ran the test again. Ran it a couple times, the last with feed rate of 125 ipm. Encoders are returning to their origins within a few counts on all three axis.

I will now go in and check the backlash and hopefully make a part tonight!

[martyscncgarage]

Since you have multiple threads going on this subject, you have resolved this issue and we can mark this topic resolved?
Marty

[slodat]

Just the one thread on this issue. I added a blurb about it to my machine thread based on your recommendation.

My machine is functioning. I still feel there’s an inherent issue with DYN4 connected to the open collector outputs.