Brother TC-225 / TC-229 adventure!

More
29 Aug 2017 03:56 #98130 by ihavenofish
nope. no combination of settings smooths it. even p at some low number like 10 doesn't. it just makes bigger oscilations.

the biggest factor in getting the z for squealing to just short oscillations, was deadband. I kept making it smaller, and it got better. but there seems to be a limit, and basically, all 3 axes are roughly behaving the same right now.

I think there may be something deeper wrong here. the whole analogue output mimicking the follow error, seems weird to me.

hmmmm
More
29 Aug 2017 19:49 #98162 by ihavenofish
so, any more thoughts? haha.

ive recovered from the novocaine haze, and would like to try and get further on tuning tonight. :)
More
29 Aug 2017 20:27 #98163 by ihavenofish
how would i make the motor just move smoothly, disregarding error?

is that p=0, ff1=1?

basically saying move at x velocity, but dont attempt to compensate kinda like open an loop.

thanks
More
29 Aug 2017 21:10 #98164 by PCW
You would not use FF1= 1, you would use the value you determined from tuning

(you can also use a very small amount of P to prevent drift for this test)

(FF1 _would_ be 1 if you set the analog output scale to the velocity in machine units per second at 10V)
More
29 Aug 2017 21:15 #98165 by ihavenofish
ah, right. although it would do the same think just offset from the desired speed.

basically, im thinking that i can start with smooth, then slowly up p til it goes bad. maybe then we will notice some clues.

or not. :P
More
30 Aug 2017 00:51 #98173 by ihavenofish
ok, so no plots cause I forgot my phone.

buuuuut.

with P at 1, ff1 at .67 (or whatever, it never would center at this level P) I get a very smooth analogue output in the scope.

encoder velocity is still all over the place.

so, it must be noise. either noise on the analogue out, or noise on the encoder in, or a combination of both. based on the motion "feeling" and sounding smoother, I'm going to guess it might be more encoder noise.

my cables are 24 inch long patched in extensions onto the originals. they are twisted in pairs on the encoder and on the analogue speed/ground. there are not however shielded. maybe the analogue should not be twisted?

sadly I don't have a real scope, so I cant check that way.

any specific thoughts on how to figure this out? I was thinking of getting an Ethernet cable and running the encoder and analogue speed/ground through that. it would easily tell me if it helps in any way. I could even do encoder, then analogue separately.

thanks
More
30 Aug 2017 03:16 #98174 by PCW
To check for encoder noise issues you can enable encoder quadrature error detection
and then check if you ever get an error, that is:

setp hm2_5i25.0.encoder.00.quad-error-enable true
and then watch or halmeter hm2_5i25.0.encoder.00.quad-error

Do the drives have differential analog inputs?
More
30 Aug 2017 03:48 #98176 by ihavenofish
drive has analogue speed and analogue ground. in my head they would have been paired, but maybe thats wrong.

anyhow, how does error detection work? what is it detecting? incomplete pulses?
ill try that tomorrow.

thanks
More
30 Aug 2017 04:00 #98177 by ihavenofish
I see that we can also just scope the raw a/b encoder channels. I can manually spin the screw and see if its smooth too.

I think encoder error makes a lot of sense. adding or subtracting a pulse every few moments would give a 'slightly" bad velocity feedback and then our perpetual compensation loop.
More
30 Aug 2017 05:03 #98179 by ihavenofish
I went back. Ha. So there's basically what looj like missed pulses. And they correspond to f error.


But....

If I push the axis by hand its clean. But it may be just appearing that way.

Quad error seems to stay at 0.
Moderators: cncbasher
Time to create page: 0.124 seconds
Powered by Kunena Forum