CNC Bridgeport retrofit

[Erik Brewster]

Should I run the current test with different low pass filter values?

It shouldn't matter one bit (as you can see from your plots being the same before and after) -- the low pass is in the velocity loop, not the current loop.

Is 330Hz where I want to leave things before I move on to position?

330 Hz low pass is probably a good compromise value if you are running a low-ish resolution encoder (< 50,000 counts / rev or so)

[Erik Brewster]

Running some longer tests lettings things get faster, getting nothing but under voltage faults.

Manual says: Supply voltage gone below lower voltage limit. This may be caused by an undersized power supply.

The suggested settings from GD for this ps are 160000mV upper and 100000mV lower limit. I lowered the lower to 50000mV an still getting the fault.

This is telling you that you are out of current capacity in your power supply.You will see the same behavior until you lower the accel limits enough and then, it will completely disappear, with a smallish accel reduction.

Remember that you are not moving under load. This is just to move your system around. When you are under load, you will need to lower your accel more (50% for a SWAG).

[Erik Brewster]

Regarding your guys posts about lowering my acceleration limits.

In tq mode the only thing I can change are PI gains, commanded current, sample length and sample rate. Accel and vel limits are not available until the mode is switched.

Lowering the sample rate gives a longer test duration, which gives the motor time to reach higher speeds. However I can't estimate how fast the axis is moving at a given commanded current.

It looks* not too fast

The accel limits I was talking about was for velocity and position loop tuning. If you are getting over current / under voltage faults in current loop tuning, lower the commanded current. It shouldn't matter too much if you have a big or small-ish current step. Your current step data is so clean that you could drop it to 25% of your present current step size with little compromise.

In general, small moves (short pulse duration) is fine for current loop tuning. The longer moves are good for velocity and especially position loop tuning.

FYI, many servo drive / amplifier companies suggest you mechanically lock the motor shaft when doing current loop tuning - the motion is not important in this step. Big motion, small motion, no motion is ok at current tuning step. Bigger motions are for velocity and position tuning steps.

[macona]

You can install a larger capacitor before the drive. What do you have now?

[legendboy]

It shouldn't matter one bit (as you can see from your plots being the same before and after) -- the low pass is in the velocity loop, not the current loop.

330 Hz low pass is probably a good compromise value if you are running a low-ish resolution encoder (< 50,000 counts / rev or so)

Got it

This is telling you that you are out of current capacity in your power supply.You will see the same behavior until you lower the accel limits enough and then, it will completely disappear, with a smallish accel reduction.

Remember that you are not moving under load. This is just to move your system around. When you are under load, you will need to lower your accel more (50% for a SWAG).

Got it. It sounds like a larger power supply would also allow me to run higher accel

The accel limits I was talking about was for velocity and position loop tuning. If you are getting over current / under voltage faults in current loop tuning, lower the commanded current. It shouldn't matter too much if you have a big or small-ish current step. Your current step data is so clean that you could drop it to 25% of your present current step size with little compromise.

In general, small moves (short pulse duration) is fine for current loop tuning. The longer moves are good for velocity and especially position loop tuning.

FYI, many servo drive / amplifier companies suggest you mechanically lock the motor shaft when doing current loop tuning - the motion is not important in this step. Big motion, small motion, no motion is ok at current tuning step. Bigger motions are for velocity and position tuning steps.

Ah ok that makes sense.

You can install a larger capacitor before the drive. What do you have now?

This just reminded me that at one point I had a 330uf 200v cap tied across HV+- on each drive as recommended because my HV bus wiring is longer then 30cm.

I can’t even remember when I removed them. I should likely be re installing them? What kind of cap did you have in mind?


Again, thanks so much for all the help guys

[macona]

I would say a couple thousand. Make sure you put a bleed resistor and a guard over the terminals. That is some serious voltage there. I am not sure how these drives are designed. Some drives feed the motor current from decal into the DC bus. Putting a larger cap in there will give it somewhere to go as it can push the bus voltage higher than normal during dynamic braking. Do these drives support an external brake resistor?

Your max accel is going to be limited by the current capacity of the drive. Nothing you do will change that, a different power supply won't help with over current faults.

From your graphs it looks like it is good enough. You are chasing shadows at this point and you should try running code with it. Watch the follow error as you run the code and see how well it does. All servos will have follow error to some degree, so don't worry about seeing some counts. How many will depend on a lot of things like how many lines the encoder has. On some of my high resolution motors I can be off a couple hundred lines. That sounds like a lot but when you consider the encoder has 131072 lines it is not a big deal.

[legendboy]

I have a bunch of capacitors left over from the boss 8 control. I will look and see what exactly I have.
These drives dump decal into the dc bus as max pointed out earlier. I do have a large regen resistor installed on the bus. It’s a 150w 200ohm if I remember.

Hopefully my better tuned current loop will eliminate the OC faults I was seeing before. I was getting under voltage faults which I thought my be an indication of undersized ps.

I spent last night re wiring my ribbon connectors between the drives and controller.
SOOO annoying winding axis back by hand when they get too close to EOT from faulted step response tests. (I had ripped apart these connections for a ribbon cable to setup the estop for tuning)
Now all I have to do is swap header pin connectors and move the axis back to a safe place with the controller.
My weekend is looking pretty busy but I may get some time tonight to go back and get position loop tuned (hopefully)

[legendboy]

So I have a 5000uf 200v capacitor

I just need to tie the hv bus across the terminals?

[legendboy]

A good test of the servos and getting them to fault is running the roadrunner g-code example in mach. Lots of short moves that can trip up a poorly tuned machine. Do kick up the feed rate.

Hey macona can you post or send me the roadrunner gcode?

I have it running pretty good but I'd like to try this

[macona]

send me your email addy.

-Jerry