PID issues or calibration help

Hi Guys,

I have verified this phenomenon on 3 different devices. My dial indicators (one digital, one analog) and then the hand control marks on my mill. I'm really confused and need some advice. Long ago I was setting my machine up with the help of another member on this forum but they did most everything and I was just "along for the ride" so I never learned the intricacies of how LinuxCNC operated.

So I discovered this issue when I noticed my small radius circles are coming out oblong and of course I thought ...OK I'm either not calibrated or that's backlash. So I started testing steps in one direction on my X axis. My resolvers are .200" per revolution so very easy math. I started moving .100 at a time loaded up the dials and reset everything to zero. My move was .099" ...OK great so I do a few more just to validate that I was loosing a thousandth each time. -Guess what. Next move it was exactly .200". I try again still moving in the same direction and now we're at .299" then again .400". Keep in mind I verified this was taking place on 3 devices (my hand dials and 2 different dial indicators). So the moves are as I say.

For the life of me I can't figure out how on earth a half turn on my resolver is resulting in a .099" move then by the second half it seems to fix itself. HELP!

I need to go test this on my Y axis as well. The only thing I can think of is my MIN_FERROR = 0.5 is the cause. But if I make it smaller I keep getting follow errors. I think my issue is that my resolvers are not being sampled fast enough (is that the PID Loop?) and so it's getting errors in between full rotations?

You may have a periodic error in the mechanics or resolver or poor tuning

The fact that you cannot reduce the following error limits below 0.5 inches
indicates that the servo system is very far from properly tuned


Have you gone through a standard velocity mode servo tuning procedure like this?:

gnipsel.com/linuxcnc/tuning/servo.html

Hi PCW,

No I never went through a proper servo tuning but I will go through it now. In fact last night I was trying very hard to (as a test) configure a fresh machine using PNCConf Wizard. But I'm running into an odd issue. It appears I have some odd SVRM6_48 files that I think you made me long ago but I wonder if they were done in a hurry at CNCBasher's request. Because I downloaded some stuff from a 2013 post and it doesn't match at all. So this could be why my pncconf wizard is failing to give me the right options.

Can you please provide me a proper link to the 7i43-4 (400k) version of the SVRM6 files? This may help me reset back to base and build my own machine from ground up.

I will go through the Servo Tuning tutorial that you linked for me and let you know what I find. If you want a copy of the firmware I do have I have attached it for your reference. Hopefully this is part of my struggle and proper files will help me get things sorted out!

At this point I would not bother with pncconf or worry about formwaare, it would be a step backwards since the machine is basically working

You should be able to tune this entirely using halscope and the calibrate menu in Axis

OK I had to run out to the store for our groceries but I did mess with HALScope a bit and got it working to a point. I'll get some screenshots but my FERROR is off the chart and I need to figure out how to pull it down into scope.

is there a URL for a tutorial on tuning my drivers with the CALIBRATE menu? I never did this either, obviously.

I think the tuning tutorial assumes use of "calibrate" since it allows you to change
tuning parameters on the fly while testing and allows you to save the updated values
in your current .ini file

( much more convenient than editing the .ini file and reloading LinuxCNC for each tuning iteration )

tommylight wrote:

Have a read through this, it has much more details:
forum.linuxcnc.org/10-advanced-configura...ning-detailed-how-to

I looked through that how-to and seen one concern. The procedure calls for tuning P first THEN FF1. PCW has always said that FF1 was the main acting element and should be tuned first for velocity loops. My tuning adventures concur with tuning FF1 first, then moving on to P.

Its good to have a low P value when tuning FF1 just to prevent drift, but high P values will make FF1 tuning harder. Any I value will also make initial tuning harder.