Mesa 5i20/7i29 trying to get more out of servos

Machine is a Lagun FTV-3 that I've retrofitted with an intel Atom mobo, Mesa 5i20 FPGA, 2x 7i29 servo drives. Upgraded the servo motors from original Electrocraft 90V units to Baldor 150 VDC motors at 2:1 reduction from a newer Lagunamatic mill that was parted out.

Right now in lagun.ini X/Y/Z axis are all set as following:
Output_scale = 100
Max_Output = 95

Tonight I upgraded my power supply situation so instead of putting out 90 VDC (as I had limited the old power supply to the max voltage of the old servos) I'm putting out 150VDC unregulated using a 2.8 kVA transformer, feeding thru rectifiers into 2 banks of 2 ea 56000 uF 80 VDC capacitors (4 capacitors total wired series-parallel).

I want to get faster acceleration out of these motors....Right now I'm lucky to get 200 and I know that these motors, with this reduction on a Lagunamatic 310 is capable of 350 IPM.

I can't help but feel like the torque they exhibit when I resist them at the ballscrew shafts just seems weak. I haven't put a torque wrench on the servo shaft or ballscrew shaft to verify actual in/lbs but will probably do so tomorrow.

I have put a voltmeter on my capacitor bank to verify that with the servo fully resisting me I get ~12 VDC drop. So from 150 VDC resting voltage I get 138 VDC with a single servo fighting my wrench. I have not yet put an amp clamp on the motor lead to see how much its actually drawing. I know brushes are good, and motors do not smell or appear to have ever overheated or cooked.

A couple of thoughts I had - I know the Mesa 7i29's have a current limit setting, but the manual says that the default setting from the factory is the higher one (22.5A) and I have to pull the heatsink to change this setting. I've never had the heatsinks off so unless they came preset to the lower one I doubt this would be the case.

I know I can bump my max_output to 100 from where its currently sitting at 95 but assuming this scales linearly, I don't think I'm going to nearly double my IPM or torque from a 5% increase in that value

I could bump my voltage up higher to where the motors are seeing 150 VDC while being resisted, but I suspect my problem is current and not voltage-related. As I understand it current is where my torque comes from, not voltage.

I could bump PID values to a higher P value but I think I'm pretty well optimized there as far as keeping overrun and oscillation at bay. For the Y axis I was resisting, current values are P 12000 I 6 D 15 - do these seem reasonable?

Thankful for any other help or ideas to test out what I can/should be doing to improve acceleration and torque out of my setup.

As far as servo tuning goes, what are your FF1/FF2 settings, these may help with your accleleration, but not your lack of stiffness. If it isn't stiff enough then you probably need more P (if you power supply isn't running out of oomph.)

A couple of things:

Maximum velocity is determined by voltage, not current (acceleration is limited by current but you
are talking about IPM which is is velocity)

You need to bound the maximum PWM setting with the
7I29 so the max output = 95 is correct DO NOT change this
(this is needed for the high side gate drivers to get power)

Bare Hbridge drives (like the 7I29) require FF1 to compensate for the motors backEMF

Have you calculated the maximum speed based on the ballscrew pitch. belt reduction and motorV/KRPM?
maybe you simple cant get to 350 IPM with your current setup

PCW wrote:

Have you calculated the maximum speed based on the ballscrew pitch. belt reduction and motorV/KRPM?
maybe you simple cant get to 350 IPM with your current setup

Double check my math here:
I originally put reduction is 1:2, double checked that wasn't quite right - on X axis I've got 14:26 reduction so 7:13 simplified

I have 5 TPI ballscrews

Max RPM on my motors is listed at 2500

So at 2500 RPM x (7:13) = 1346 (rounded off) RPM at the ballscrew

1346/5 = 269.2 max IPM possible. So 350 IPM is not possible, but I should be able to do better than 200 IPM rapid

I'm going to try upping the FF1 and continuing to try and improve PID tune.

Remember that you had only 138V under load *0.95 PWM max so RPM is about 2185 max
This works out to about 235 IPM but you need some voltage margin for the motors, at least 20%
so 200 IPM is about the expected maximum speed

Some extra testing:
I put a digital torque wrench on both X and Y axis screws to see how much torque I had.

Got ~10.3 ft/lbs on X, ~10.5 on Y.

For the sake of consistent numbers, multiplying by 12 gets me 123.6 in/lb on X, 126 on Y
Both motors are rated at 28 in/lb stall torque
Since I'm gearing down with the toothed belts I'm under the impression torque should be (26/14) * 28 in/lb which gets me 52 (rounded) in/lb

So now I'm confused. I still think my torque should be a bit higher (for comparison sake, Servo sells a knee mill power feed for X & Y that peaks at around 140 or 150 in/lb of torque, depending on model). But it would seem that in real world measurements I'm getting significantly more torque than these motors are rated for, so maybe I need to upgrade motors to higher torque models in order to get better acceleration and rapid speeds out of this mill. Thoughts?

Stall torque (your 28 inch/lb) is a thermal rating for continuous load.
Peak torque (which is what you are measuring) is 3-5 times
this and determined by the motors and drives maximum current rating

PCW wrote:

Stall torque (your 28 inch/lb) is a thermal rating for continuous load.
Peak torque (which is what you are measuring) is 3-5 times
this and determined by the motors and drives maximum current rating

If peak torque should be 3-5x the label rated stall torque, then doesn't my numbers I'm seeing at my ballscrew seem low? Checking numbers here
3 * 28 in/lb = 84 in/lb at servo * (26/14) = 156 in/lb at screw
5 * 28 in/lb = 140 in/lb at servo * (26/14) = 260 in/lb at screw

These numbers seem a lot more along the lines of what I want to see for torque numbers at screw.

andypugh wrote:

greenbuggy wrote:

Since I'm gearing down with the toothed belts I'm under the impression torque should be (26/14) * 28 in/lb which gets me 52 (rounded) in/lb

If you want faster rapids you could consider altering the belt ratio. You need to be sure that you have enough peak cutting force still, but with a different gear ratio you will be able to reach a higher peak speed, but possibly at a lower accelleration.

Original servo/resolver setup on this mill when it had a Bandit controller did actually have the X-axis directly driven from the servo. Don't know about the Y-axis as it was MIA when I got the mill. Still, if my motors aren't hitting the torque numbers they should be coupling them directly to the ballscrews isn't going to do me any favors.

I did add in some more FF1 and FF2 starting with numbers suggested here: gnipsel.com/linuxcnc/tuning/servo.html

Now at 500 FF1 2 FF2 on X

My acceleration is definitely better, have not put a torque wrench back on the screw yet to see if my torque numbers changed at all

Point-to-point times for short moves might actually be longer with a higher peak speed and lower acceleration. How far do your axes have to move at the current settings to reach max speed?

s = v^2 / 2a. Use the INI file numbers for the calcs.

Aren't the numbers in my ini file under axes maximum accel? I can specify far higher accel numbers than the machine seems capable of actually hitting, right now its set at 2.5. How would I go about determining actual acceleration numbers?

If you are wasting a lot of time on rapids, you might want to look at optimising the G-code first. For example does your CAM allow you to switch off full-retract? Certainly in Fusion 360 you can opt to leave the tool down at the cutting height when re-positioning, and it is clever enough to move around the remaining material.

What I've been doing has mostly been 2.5D work so haven't been zipping the Z axis around a lot, but hoping that troubleshooting the accel/torque issues on the heavier-to-move axis on this mill will help me to improve Z as well. Dreaming of a mill I can do HSM on...