Brother TC-225 / TC-229 adventure!

I did further tuning, scaling is now correct and I get values of order unity for FF1. P on the servo amp is 50% and I at 100% of the potentiometer scale. There are however still some peaks in ferror. The are always ~60µm no matter how much D or FF2.
I wanted to know the frequency response of the system, so I hooked up my oscilloscope and did some bode plots. The plots show servo motor velocity (or whatever else pin N on the servo amp is) vs. input velocity.
The second plot shows the y axis motor which was left unconnected (had a little less P and I on the amp).
From the plots I would say that there's a resonance between 100 and 150 Hz, maybe it's a good idea to turn P a little down so it's not that pronounced anymore. I am still asking myself whether there's a jerk limit that can be tuned. There is "UZ" and "WZ" as potentiometers. I haven't yet found out what they are good for. Would I be able to see such a limit in the bode plot for small wobble amplitudes and hence only small 2nd order derivatives?

EDIT: The plots show intensity (or power, if you want). divide by 2 to get the amplitude response

here's another picture of torque out vs. input velocity.

looks basically identical to mine. the cachunk at the start and stop of acceleration, then fairly smooth for linear motion.

I broke one of the oldham couplers during a small crash. I ordered replacement disks and will exchange them tomorrow. Compared with the new disks, the old one looked a little worn out. Hopefully that will improve the frequency response.

since the encoders are on the motor, i dont think the tuning is effected by tiny mmounts of platy in the system. but id be curious to see what happens. my machine has a tiny bit of backlash in each axis.

The stiffness/backslash certainly does affect the tuning. I did some calculations and the first dip in the response curve corresponds to a stiffness of less then 20Nm/rad with the moment of inertia of my ball screws. So there is something in the chain between motor and load (table) that is not very rigid. Could be the coupler, could be backslash in the ball screw nut or something else.
For comparison: the Huco couplings I found in my machine have 200Nm/rad stiffness according to the datasheet.

the machine runs perfectly smooth under the stock control. so, jerk control smooths it out?

im not sure. many other converted machines run fine with bad backlash, however, i know most of them dont have near the same acceleration and the brother..

Sure they run fine with backslash, but who knows what servo bandwidth they run with and what overshoot or oscillations they have. While tuning the motor you see just the motor encoder velocity but not the table movement. For example, in the bode plot above there is -30dB @ 72Hz i.e. ~1/30 of the amplitude compared to 20Hz (plots show power, divide by 2 for amplitudes). If you put your hand on the table you can feel, that it is vibrating as hell. Probably much more than at 20Hz.

Maybe I am putting too much effort in this, but if I can solve the problem for 6$ (for a pack of 10 disks) I'll give it a try. It won't hurt.
Speaking about jerk control. I should have measured the velocity ramp of the stock control, when I had the oscilloscope in the workshop..

so ill probalble stick these in storage for a bit, while i seek out a new shop. im sending the housings to be powder coated, and i will pull the tc225 ball screws to re pack them with new balls.

i think what i may do it take my computer and mesa gear, and my bekhoff system, the servos for size reference, and some other bits, and play with them offline in an attempt create a stable system to apply later.

the only thing that definitely works fine is the spindle motor / drive on the tc225, so we will just leave that in the machine.

we have a few options for new servos:

1 is to keep the sanyo servos, and give them new drives. this im told isnt always smooth if you want to keep the high end performance.the delta asd a2r is a drive we could use, at about $600us each.

2 we just buy new delta drives and motors. $3000ish, analogue and pulse and direction. they make a 750w and 1kw model to fit, but they only run at 3000rpm, which would lower the speeds on both machines.

3 we buy different sized 750w and 1kw servos from delta or yaskawa and make them fit with adapters and new couplings. these would be 5000-6000rpm, and increase the rapids. these motors dont have the low end or intermittent performance of the sanyos though.

4 we buy the brand new equivelant sanyo motor and drive kit, at a cost of around $1800-$2400 per axis. basically whats in a brand new brother speedio, designed to work in this machine. as you may have read earlier, i do in fact have 3 newer-ish sanyo rs1 drives that take analogue or step with incremental encoders which could be used with brand new motors.

hmmmm... any other ideas?

Oh, maybe this could work for someone. i have a TOS FNK25 manual mill, kinda like a bridgeport, but bigger. its a good cnc conversion candidate. i cant take it with me when i move out of the shop, i have no room. maybe someone here would like it, either to buy, or to trade for some ac servos and such that could be used in my brothers.