Bridgeport Torq-Cut TC4G Retrofit

chopper79 wrote: They said that the DX32 controller used a PID to orient the spindle. Basically just like I am attempting in LCNC.

What does the VFD orient to? Is there any way to change the VFD orientation angle?

What might work would be to use a LinuxCNC PID to control the VFD orient position to achieve the desired spindle position.

I am going to create a different config and focus on just the spindle for a bit. This will eliminate any odd items I have in my HAL now that may be conflicting with the orientation. I will do this tonight and post what the results are. Spoke with Yaskawa again today and did verify a few things on the VFD and how this was being controlled. It was being oriented via a PID, so I need to just focus on a single gear and see what happens.

Having a funny feeling there is something in my config causing the conflict, but then again maybe not.

Yes I can change the orient angle to what ever I choose between 0 and 360. The angle is set in the VFD constants. I thought about seeing if there was a way to call set the orient based on the spindle encoder and PID. Did not know if I could or not.

Created a new config last night.

* Spindle is functioning in both directions
* Spindle seems smoother now also (lack of better words is seemed a little noisy and choppy before)
* Speeds are all matched everywhere
* Encoder is working and signed correctly
* Spindle still turns when I press CW or CCW (same issue as before)
* Analog 5 has ~ 0.006vdc on the outputs when I press CW or CCW which is causing the rotation

I went through the settings a bit on the VFD and found that the VFD is running is servo mode. This makes me think that the spindle was being used in a closed loop arrangement by Bridgeport and not in open loop that it is currently being used in.

So now with my current test config (generated by pncconf) I should be able to adjust the PID for closed loop operation and remove the FF0 = 1 value or am I incorrect on this?

I will reconstruct the config to add items in slowly and confirm results each time to see if I can find what ever issue the old config may have had. Need figure out the closed loop bit first to test.

chopper79 wrote: Yes I can change the orient angle to what ever I choose between 0 and 360.

How? ie can you change it while the system is "live"? If it is a front panel parameter then it probably is not possible to control the orient angle via a LinuxCNC PID.

The VFD can only be changed when stopped, in digital operator mode (keypad), and after a tune up process. This rules out the possibility to control the orientation angle of the VFD via LCNC I would think.

Seems the only way is to do a PID loop for orient or purchase an obsolete orientation card.

Have you looked at the output of the PID with halscope?
Do the numbers coming out look reasonable compared to normal command signals to the VFD?

Played around with the orient some more and I have not found a way to keep the spindle from oscillating once oriented, even with the new minimal configuration.

I have settled on using the is-oriented signals which once oriented the commands stop. This keeps the spindle from oscillating and achieves the target by +- 0.2°. Then when the tool release cylinder is activated it locks the spindle anyway so everything should work out fine I would think. Need to purchase some tool holders to test with as I do not have any at this moment.

I will reassemble my config over the next couple days and test again.

andypugh wrote: Have you looked at the output of the PID with halscope?
Do the numbers coming out look reasonable compared to normal command signals to the VFD?

I have not yet, just used halshow and monitored the positions, command, feedback, and error. I got the error down pretty well by not using so much P and still achieving decent orientation time.

Main reason for not using halscope is that I forgot how to use it... Need to brush up on it again.

chopper79 wrote: Played around with the orient some more and I have not found a way to keep the spindle from oscillating once oriented, even with the new minimal configuration

Did you try deadband at any point?