Encoder support in stepper systems.

Hi,
I'm planning on using a pair of step/direction drives in a special machine I'm building. It will be of the Mitsubishi/Panasonic/Yaskawa sort of universal drive which can accept either step/direction or velocity inputs.
They typically have encoder outputs which you don't need to implement in a classic step/direction setup. They are great though in that they have the simplicity of steppers, but are actually a pretty sophisticated servo. If you stall out one of these drives, they will clear any pending step inputs and flag an error.
So with encoder available it is possible to maintain position through Estop, and the Pico documentation alludes to it. I'm unclear how it's being implemented though. Is encoder used at all times as a check on follower error, or only loaded in an Estop condition? Or is it actually being used in the position loop? I would think trying to use it in a position loop would lead to all sorts of conflicts. Maintaining position during times of servo-off would be sufficient for my purposes.
Is this a driver function, or just a Hal function? Do the Mesa boards also support step with encoder use? I guess if it's in Hal, the answer should be yes.
Thanks for any enlightenment.
lm

It is possible to do both systems with EMC, that is, just use the encoders to monitor following error or to run a full EMC controlled position loop feeding a step/dir drive.

Running in a EMC controlled position loop is really not a lot different than a standard analog velocity mode servo, the main difference being that the velocity command to the drive is now a step rate (and dir) rather than an analog +-10V signal. The drive would have to "forget" any buffered steps on ESTOP or when not enabled to make this possible.

Both Mesa and Pico systems cards can run in these modes (as can a bare parallel port up to some maximum step rate/encoder counts/s)

I did it one year ago.
You can see:
www.linuxcnc.org/index.php/english/compo...atid=30&id=5487#5489
All is in working condition now.

sheetmetal wrote:

So with encoder available it is possible to maintain position through Estop, and the Pico documentation alludes to it. I'm unclear how it's being implemented though. Is encoder used at all times as a check on follower error, or only loaded in an Estop condition? Or is it actually being used in the position loop? I would think trying to use it in a position loop would lead to all sorts of conflicts. Maintaining position during times of servo-off would be sufficient for my purposes.

lm

OK, there may be two ways to do this. The normal way is to have EMC read position from the drive's
encoder output at all times, and set following error limits to suit your requirements. This would allow
EMC to follow the machine's actual position even when in E-stop, also. Surprisingly, it does NOT
cause a conflict. If the drive has tight position control, then you set the gain of the PID loop very
low, and EMC/driver/USC board mostly acts as an open-loop step pulse generator, but final
tallying of movement is made from the encoder reading. Switching between closed-loop and
open-loop is fraught with all sorts of edge conditions, it is much better to just run in the
closed-loop mode at all times.

Another way would be to use the drives fault output to signal EMC of that condition and halt all axes.

Jon

Thanks for the replies.
Yes, switching between closed loop with servo off, and open loop with servo on seemed to have difficulties.
Closed loop all the time seems much more straightforward. I guess conceptually it's no different than if I took a standard analog interface board and ran it into a voltage to frequency circuit with pulse outputs. A Panasonic drive is basically like an analog drive pretending to be a stepper after all.

aike. Thanks for the examples. I love real world examples. I just wish I could understand Russian. Looks like you have a nice installation there.
One thing that I noticed in your ini file is that the PID gains are 50,0,0. Jon said PID gains could be low, but I expected a bit of integral gain to bring the axis into position. Were these values determined by experimentation, or by recommendation?

I'll be using the fault outputs from the drives for sure. I want the axis drives to be minimally powered. This system will, I know, be stalled fairly often in the course of operation, and I don't want the drives to be able to damage the hardware. Open loop is unacceptable in such a system, but I like the simplicity of the step/direction hardware.
lm