Need Help: Mesa 5i25 w/ 7i76

@tommylight I Edited the ini and set p to 1000 for all axis. I deleted all instances of maxerror from the .hal file as well. I also had the servo thread at 1,000,000 not 100,000. The machine seemed "normal" if not better. I ran a program on it and cut a part that I have been making a lot of recently. Good news, is that it ran the entire program with out promoting the

hm2/hm2_5I25.0: SMART SERIAL PORT 0: Dolt not cleared from pervious servo thread. Servo thread rate probably too fast. This message will not be repeated, but the hm2_5i25.0.sserial.0.fault-count pin will indicate if this is happening frequent -- error.

I'm guessing that changing the actual servo period to 1,000,000 is what cleared this error. I'm still not sure what deleting the max error and changing the p values did. So basically at this point the machine will home its self, run programs with out errors, everything seemingly works, the parts are just coming out .050" over what they are modeled from. Thanks for all the help.

The P value is calculated by pncconf based on the specified servo thread period
a value higher than 2/servo-thread-period (> 2000 at 1000,000 ns) will be unstable
and cause position oscillations.

Probably pncconf should not allow changes in the servo period without a warning

Over what distance is the 0.050" error?

Thanks for the help and explanation, I'm trying to get my mind wrapped around all this the best I can. The .050" deviation is over 1.5" in x axis. Tomorrow I'm going to run some simple square and circle cuts over a larger area to try and figure out if the error is a result of bad cam work or a machine setting. I'm using fusion 360, which I have had great success with on my other linuxcnc machine. Now there are two post processors emc2 and linuxcnc, a few years back it was just emc2. So I hope I'm not using the wrong post.

Now that I've got the servo thread error ironed out I'm thinking this may be a scaling issue . I didn't want to completely disassemble the gantry on this machine, so I deductively calculated the pulley ratio using a daily indicator. I will double check that the scale factor / ratio is indeed correct.

Alright - I'm back at it. I had to finish a run of a parts before I could really mess with the machine too much, and then the covid. But, I'm going to try and make sense of why the error is in the machine. Its still cutting parts about .050" oversize on a two inch part. I made a run of 30 parts and they all come out constant. That leads me to think this is a scalling issue and not a loss of steps. I'm trying to figure out the best way to trouble shoot this. I'm thinking the first step is to cut a square and a circle at 2", 3" and 5" to see if the error is constantly .050" - which it appears to be. The machine was setup in pncconf. I calculated the pulley ratio by using a gear ratio chart, and a dial gauge magnetically attached to the gantry. I've checked that and I think that its correct. This leads me to if its something in the fusion 360 post processor that could be causing the error. I'm about to go out in the shop and try and get to the bottom of this. Any advice is greatly appreciated. No need to tell me I don't know what I'm doing - im fully aware!

Is it out on both axis?

Hi,

why not simply try a

g91 ; relative coordinates
g21; in millimeters
g1 x100 f100 ; move sloooowwly 100 millimeters along x

and then check if your spindle really moved 10 centimeters?
If this is off, it can‘t be postprocessing but seems to be your stepconf/pncconf ... Thatˋs for me the most likely source of error, : triple check gear ratios, spindle pitch or belt teeth and microstepping settings...

Maecki

Alright I just cut a 3" x 3" square it came out in
x =3.087
y=3.003

This leads me to believe the error is in the setup not the post processing. I'll try the 10 cm move as listed above. I'm not sure the best way to actually measure that.

Ok I think I may have figured something out. I think the issue is in the scaling / gear ratio of the machine. I didn't build this machine, I bought it from china, ripped off all the mach3 stuff and am trying to get it to work with linuxcnc. (dumb I know). I can't get to the pulley's to actually count the teeth. ( I could but it would require some pretty serious disassembly) The way I calculated this was by using a dial indicator. The problem with that is the short throw of a dial indicator. I had the Y axis, which seems to be really close @ 30:18 and the x at 26:16. So basically .6 and .615. Very close ratios. I'm thinking that I miscalculated this and didn't use a large enough test area to prove that out. Also I doubt they would have used different pulleys on the x and y axis.

The simple thing to do would be to change that and see what happens. An issue I'm having though is I don't understand what the shortcut to launch linuxcnc references. When I run pncconf I have the option to create a desktop shortcut. When I run pncconf I also have the option to modify an existing setup or create new. If I modify an existing setup, do those changes post out to the desktop shortcut. IE the file I'm editing is in the linuxcnc/ config directory. If I change that file via pnconf does that change the previously created shortcut? I'm getting lost in my versions. I think my next step is to screenshot all the previous configs, erase them, and start from scratch. THEN try and mess with those ratios.

UPDATE: I looked at the properties of the desktop shortcut to figure out what config file they were pointing to. I ran pncconf, modified an existing setup, changed the x axis scale to match the y (30:18). Changed nothing else. Ran that config on the machine, and when I attempted to home the steppers sounded horrible and were going really slow. I stopped the machine, reopened pnc conf changed the number back to the original (26:16), changed nothing else. Reopened the Machine, attempt to rehome again and the same thing happened terrible noise. I don't understand how changing the stepper scale ruined this config. Modify an existing configuration seems not to work for me. I think starting from scratch may be the best option.

Also in the axis config. I've got all my stepper on time, step space, direction hold and direction setup set to 10000. However when I go to test tune the axis, it only allows me to have a max of 5000 which causes the machine not to hold. Also I just crashed my machine pretty hard, I'm guessing limits don't work in test tune.

Alright. I think I figured out the problem. I created a new configuration, based off the previous settings in pncconf. I changed the x axis ratio to match the y axis ratio, and I'm coming up much better, around .005" big. So I'm pretty sure I just had that gear ratio wrong on the x axis.

I do not understand why creating a new configuration fixed this but when I tried to edit an existing it did not. I compared the Hal and ini files of the two old configs and couldn't find any differences.

I'm guessing that last .005 is backlash and will need to be compensated for. Next thing to figure out... But I'm finally making progress so thats good.