Gerber 408 retrofit

Rebelx:  Do you happen to have your old config files? I did pretty much the same as you and am pretty pleased with my tuning but always wondering how others have done it. This is what I ended up with.

P = 150
I = 0
D = 3
FF0 = 0
FF1 = 1
FF2 = 0
BIAS = 0
DEADBAND = 0

Hi SkyMoCo. Yes, I still have some of the config files. They are not the most recent, as I lost one of tthe backup drives with while moving abroad, but I believe these settings were the most successful for my machine:

X-Axis:

P = 10
I = 0.5
D = 0.02
FF0 = 0
FF1 = 0.2
FF2 = 0
BIAS = 0
DEADBAND = 0

Y and Z axis: 

P = 5
I = 0.5
D = 0.01
FF0 = 0
FF1 = 0
FF2 = 0
BIAS = 0
DEADBAND = 0

The setting Forest provided me with for his machine are listed below. Unfortunately those settings did not work for me, but they helped me a lot to debug the problems with my setup:

P = 40.0
I = 0.0
D = 0.1
FF0 = 0.0
FF1 = 0.106
FF2 = 0.0
BIAS = 0.0
DEADBAND = 0.0

Please keep in mind that the settings for your machine may be completely different. The teflon coated screws on my machine were quite worn, and as a result of that also nuts were in a less than optimal condition. That may have had quite a significant impact on the dynamic behavior of the system and therefore the PID settings I used may not work on other systems. It may be best to use the settings of Forest's machine as a starting point.

Thanks. I have a couple hundred hours on mine so it's a bit past the starting point . According to the inverter, when I got my machine it only had about 600 hours on it total so the ball screws are still pretty tight so I guess I got lucky there.. I have everything working but always wonder about everyone else's settings.

Mine seems to be much smoother with a P value around 150. I always thought that a higher P value was better, but might do some testing with your and Forests settings. I think I had to back off my acceleration settings or I would get some overshoot.

And just for those of you reading along.   I also added a BT30 spindle off AliExpress.   I use the same VFD.   The little green button lights up when ready for a tool change.   I might add a rack mount tool changer but am a bit reluctant as the built in chip vac works really well for me and it's not like milling where I go through 10 different tools for a single job. 

 

The key to getting good servo performance on the Gerber (low following error, low oscillation) is to realize that their brushed DC servo drives are operating open loop in torque / current mode (closing current loop, but nothing more), not voltage mode (semi-velocity mode but load dependent) nor true velocity mode (tachometer or encoder feedback to the drives).

PNCConf generates a PID loop that by default operates off position error and modulates the velocity command, and if you instead hook that to torque, it doesn't work very well (it is at least difficult to tune).

It is much better to do two levels of PID, one "position loop" PID that modulates velocity, and a second "velocity loop" that modulates torque. The drive itself manages the "torque loop" by regulating current through modulating PWM duty cycle (voltage). Separating these loops makes the job of each loop much more simple, and easier to tune. The position loop is operating in proper engineering units, and while the velocity loop isn't, it's still happier operating off velocity error rather than position error. The reason it doesn't have engineering units that make sense is because command/feedback/error are in units of linear distance per second, but the output is torque/current in units of newton meters or inch pounds / amperes, and there isn't really a direct conversion since it's dependent on acceleration/load which isn't fully known.

Attached is a screenshot of HalScope showing the performance of the X axis on my Gerber Sabre 408, you can see that the actual velocity trace very closely matches the ideal, and that the following error is within 0.030mm or 0.0012" when cruising at 590 IPM, and within 0.080mm or 0.0032" during acceleration/deceleration. The following error of course gets lower if you are operating at lower speeds, this is a rapid move for stress testing.

My machine was quite heavily abused when I bought it. But despite the worn lead screws the machine still did some amazing routing. The person who I sold it to was made fully aware of the condition of the screws, but even he is using the machine for his daily production without any problems. I advised him to have the screws either refurbished (there are companies who can apply super durable industrial teflon coatings for approx. 1/4 of the price of a new lead screw) or have them replaced, but he is happy with the machine as it is.
As for higher P values.... vibrations are a weird thing. When I increased the P values, my machine started screaming like a pig.

Hi Forest, thank you for your additional clarification on the operating mode of the Gerber servo system. It all makes much more sense to me now. From the documentation of the Elmo drives I somewhat understood how things worked, but never fully grasped the way in which the servo drives were working on these machines and was even doubting my own mental capacity I still wonder why they designed it in the way they did.....

Hey Forest, I sent you an email about your replacement mobo.

I have recently found some manuals about the Gerber original servo motors (Parker RX320ER1026)
Original Encoders (Renco 324599-03).
www.stabe.se/pub_docs/files/RTS/192_1244...X_Axem_Catalogue.pdf