[derek]

So I'm converting from CNCdrive DG4s drives to DMM gear. I'm having a lot of trouble getting the DYN4 to work in a way that I'm used to. On the DG4S drives you set an encoder count for the following error and when it exceeds that the drive faults and motion stops and the error line signals the BOB to go into estop. From what I’m seeing the “on position range” isn’t doing what I think it should be doing.

I used the auto tune feature and the motion looks pretty close with what I was seeing with the DC system. My goal was to get the drive to estop the system if the following error exceeded a certain amount.

I played with the "on position" range and have pin 6 on the drive switching from high to low when it exceeds the on position setting. The problem is even at the full 127 setting I'm triggering pin 6 on every direction change. I suppose it could be a massive overshoot problem but before I dive down that rabbit hole I thought I would see what others are doing.

Thanks
Derek

[derek]

Well it would seem I'm in the minority of people who use following error. On the CNCDrive DG4S and the Clearpath SD series that I have experience with have a setting based on encoder counts on how far behind command position are you comfortable with before the drive faults. The DYN4 handles it differently. It uses a set amount of shaft rotation which is 90 degrees and a time of 2 seconds before the drive faults. 90 degrees of shaft rotation is .050" on my mill. I have verified this with tech support at DMM.

This is a feature that I feel is a basic requirement for any machine that is depending on the drive to monitor position. I bought two kits and will be returning one. I have been using one on a machine and I will use it in another application that is less critical. I really like the drives and servos but if I had known that this is how they handle following error I would not have purchased them.

[samco]

wow - that is just crazy.. I wonder how many people actually know how well their machine is following the commanded position...

[derek]

wow - that is just crazy.. I wonder how many people actually know how well their machine is following the commanded position...

I know. I keep thinking I have it wrong but I listed several scenarios where a loss of position would occur to make sure I wasn't missing something in my understanding. This is the last scenario I presented just so I had clarity:

Right now with my DG4S drives If I accidentally rapid a milling cutter into the work piece and the following error on the drives is more than .005” ( my setting in the tuning program) it instantly shuts down the drive and estops the system.

My current understanding of Dyn4 is the mill would hit the work piece, Continue pushing until drive overloaded, or 90 degrees of lost phase and after a 2 second delay fault the drive. Is this a correct understanding?



The answer from DMM:

So basically if the drive hits a hard limit, it will fault after the position error is outside 90degrees. Of course within this time, the drive can also throw an over-current error if the current is too high.

These are really great drives and servos for the money but I think they are designed to work with a higher end system and not step and direction.

[DJEkim33]

I can only use my drives in step and direction. I get over current errors when trying to use them with analog control. The drive doesn't report encoder counts fast enough to the control computer. This resulted in the controller trying to drive the motors faster to catch up and resulted in over voltage faults at high speeds. Pretty disappointed if the drives are also not accurately keeping track of their on position. I have to look into what your saying and see if thats the case on my system also.

[Karl_T]

FWIW, I have used the DYN4 drive and servo in +/- 10V analog control using a Galil card for four years now. They work great in this application. I did have a huge learning curve getting them tuned properly.

[DJEkim33]

I'm aware some had success some had had with Galil cards. My understanding is that filters are applied to overcome the slow response of the DMM drives encoder section. It can be done in Linuxcnc but to me that is just masking the problem. Most of the other servo drives out there don't require this bandaid to be used in analog control. The DMM drives are at times a full 7ms behind the commanded speed. Really not acceptable at any price point.

[durundal]

I'm using DMM drives with a KFLOP & Kanalog in torque mode, so the encoder is being read on the KFLOP and +/-10V being output to the drive with tuning parameters and following error being tracked in the KFLOP. What do you mean by the drive being behind the commanded speed? That the DMM drive takes 7ms before applying a new output based on the computed error if it's running in closed loop mode?

[derek]

On the step and direction drives I've used the drive handles the position monitoring and corrects accordingly. The amount of error before the drive faults has, again in my experience, been adjustable. The DYN4 has an out of phase error of 90 degrees for two seconds which on my machine is .050" which is a mile and 2 seconds is an eternity.

I've moved on to a clear path servo and at this point that is the direction I will go every time I need to upgrade a drive or motor. I like the DYN4 and it's a great system but for me it's not a good choice.

[DJEkim33]

I'm using DMM drives with a KFLOP & Kanalog in torque mode, so the encoder is being read on the KFLOP and +/-10V being output to the drive with tuning parameters and following error being tracked in the KFLOP. What do you mean by the drive being behind the commanded speed? That the DMM drive takes 7ms before applying a new output based on the computed error if it's running in closed loop mode?

Looking back through my old notes it was actually 5ms delay. The following picture is a scope of the velocity command (white line) from the control and the reaction of the drives encoder feedback (purple line). The result in velocity mode was the drive attempting to catch up to the commanded speed and over current fault in the process. This was at 20 in/s/s acceleration which is worse than i got out of stepper motors on a different machine.


To confirm these findings I did some testing. I took a 30 year old Heidenhain 500 count encoder and mounted it directly to the servo shaft with a piece of tubing and clamped it in place with some vice grips. pretty much the poorest most misaligned setup you could get for an encoder. The result was the Heidenhain read the movements with less delay then DMM servo drive encoder emulation. You can see below the Heidenhain in white at the bottom of the scope with a 2ms delay. The velocity command from the control in red. The DMM servo drive encoder count in blue at top with a larger delay than the poorly rigged heidenhain.


What it all comes down to is the DMM drives are designed to be easy to setup the drive parameter's. They only have 5 settings that can be adjusted. Inside the drive are internal controls that are filtering inputs and outputs to make for more stable motion. This causes delays in their reaction to commands and the encoder signals coming out of the drive are also delayed. This filtering works against an outside control accurately controlling this drive system. You can see by the second scope that the drive was slow to respond to the command and even slower to report back the movement from that command. I preformed the same test in position mode and got similar results.

Sure you can add filtering in the control to ignore these delays but then you have to question whether the control is accurate. I personally paid for drives that would deliver on the promise of accurate and smooth motion. Considering that the 30 year old motors and drives I removed were capable of accuracy and smooth motion I didn't think it would be too hard for DMM to achieve that but they didn't meet my expectations. The DMM motors couldn't achieve 2/3 of their rated speed without faulting out because of these delays. I wasted a bunch of time figuring this out only to be told by DMM that its not there problem. I currently have the drives in position mode and yes they do move the machine around without faulting out but they aren't moving accurately or when they are commanded too.

[derek]

I did a caveman test before I pulled the DMM stuff. I set the rapid speed well above what I knew the servo could drive it and did a G0 X24.0 command for a 24" move. You could tell by the sound of the servo that it was lagging and if this was a coordinated cut move with the Y axis it would be massively inaccurate.