Re: running servo's

Jan wrote:

> Hello all,
>
> I' m busy to enhance my EMC knowledge with servo's but unfortunately my brain can't figure out what is going wrong.
>
> I have:
> Working Deckel-machine with servo's
> STG-board
> RedHat 5,2 with kernel 2,0,36 Rt-patch and EMC 16/9/99
> Working and properly connected encoders (tried this out when working with original controller and EMC but leaving DAC cable out and so determined INPUT SCALE 1000 0 )
>
> My problem is that when I disconnect the original DAC cable from the amp's, the servo's are drifting the moment I switch on the machine. Can this come from the fact there is to much gain on the amp's? (I don't like to turn this pot-meter as I still want to work with the original controller and this value is yet tuned for the machine.)

This is the way EMC works. You have to hit F1 to get out of Estop, then wait a moment for everything to
settle, and then hit F2 to close the positioning loop. Between those two steps, the machine is drifting at whatever
velocity develops due to the slight offsets in the servo amps. My setup with EMC works the same way, although
the drift is quite slow, about .0001"/sec on the worst channel. You want your servo amp gain to be high, so that the
following error is small when the machine is not moving.

> Then with this fault and connecting to the DAC of the STG-board, starting EMC; out of E-stop and put the machine-on, I can be impressed to see that I have a speedy machine, still it would be great when he waited for a command to move.

After you press F2, movement should be sensed by the encoder counters, and EMC should develop a velocity
command to move the machine back to whatever location it was at when you hit F2.

> So I thought that this movement came from the fact that I reversed the DAC cable, so one movement could compensate the other, tried that out, the result was even worse as I had a servo flipping one side to another for just a couple of degrees (shaking is maybe a better word).

Yes, that sounds like the command phase is backwards.

> Tried also with other PID value but no improvement was seen.
>
> So what knowledge am I missing ?
>
> and what is BIAS? I do not find a description of this parameter.

Bias is used for axes that have constant loads, like gravity acting on the Z axis.

Jon