Dynomotion

Group: DynoMotion Message: 8335 From: cristian_atzori Date: 9/12/2013
Subject: Kanalog with step-dir servo drives. Is it worth?
Hi all.
After about an year of experience on Kflop (you can find my old topic about a conversion of a heidenhain machine) I'm about to build a classic 3 axis cnc from scratch.
It has some mechanical details that persuaded me to give another try with Kflop+kanalog.
There are two ballscrews (with rotating ballnuts) on the longer axis, so I'll use the master-slave option.

I have already some dc motors and drivers that are able to close the loop directly inside from standard RS422 encoders, and can be commanded only with step-dir.

Question:

- can I obtain a good closed loop in Kanalog+Kflop bypassing the internal drive logic, or is it anyway better to search for "stupid" drives without encoder inputs, accepting +10-10V in analog input?

I wish to figure out the proper configuration to arrange to make it work, I really wish to use the coordinated motion capabilities on Kflop, but also I'd like to keep my drives.

I have magnetic linear encoders along all 3 axis, but the long one has two motors, so two magnetic scales, one per side of the beam been moved.

Any suggestion will be useful, even that to absolutely buy new dumb drives.

thank you
Cristian Atzori
Group: DynoMotion Message: 8339 From: Tom Kerekes Date: 9/12/2013
Subject: Re: Kanalog with step-dir servo drives. Is it worth?
Hi Cristian,

It sounds like you will end up with a dual loop type of control with feedback from both the encoders on the motor shafts as well as the linear encoders.

Regarding the magnetic linear encoders I have had bad luck with some of them.  Usually they have magnets at a pitch of 2mm or so compared to an optical encoder with an optical pitch of 20um or so.  100X less basic resolution.  They then make up the different with some sort of electronic or software interpolation.  The result is usually noisy and sometimes delayed and non uniform.  Some Users have reported good results.  I just thought you should be aware of the issue.

KFLOP has a closed loop step and direction capability that would be the simplest  to use.   The drive would provide the motor shaft feedback. 

KFLOP can do dual loop feedback but it is more complicated and requires two Axis channels and two encoder inputs per motor. 

HTH
Regards
TK

Group: DynoMotion Message: 8340 From: cristian_atzori Date: 9/13/2013
Subject: Re: Kanalog with step-dir servo drives. Is it worth?
Thank you for the warning, I was already aware of that risk so I tried to use the most reliable encoders I heard about.
I choose the RLS LM10 heads: http://www.rls.si/en/lm10-linear-magnetic-encoder-system--15857

So this is the way?

- encoder on the shafts => into drives
- linear encoders => into  Kanalog
- step-dir => from Kflop or Kanlog using Mux function

It seems fairly simple.

I wonder what would be the differencies using this other way:
- just linear encoders=> into Kanalog JP1
- +10-10V from Kanalog DAC 0123 outs to the drives (0=X, 1,2=Y in master-slave, 3=Z)

It's difficult for me to imagine what kind of response would be better and more precise between dual and sigle loop.

Encoders on the shafts are rigidly coupled, instead the linear encoders would correct any backlash.

How would then interact the processing power of the drives in conjunction with the super fast Kflop processor?

Is it good to have that kind of relation with drives that can control PID at 300uS intervals?

Do I have to search for an optimized ratio between the linear encoder counts (1um resolution after quadrature) and the encoders resolution?

The rotary encoder resolution is related to the pitch of the screw and not only to the axis of the motor, so maybe I should avoid ratios ending up with fractions.

For instance, if I have a screw with 5mm pitch, considering the 1um resolution of the linear scales, should I prefer a shaft encoder with 1250 cpr (5000 after quadrature) giving to the drive the same resolution of the linear scale or it's not relevant?
Lower values of 128---512 cpr could give better or worse results in this double loop application?


Sorry for this number of questions, I suggest to add an example in the manual pages of the website, this maybe is a FAQ.

Thank you




--- In dynomotion@yahoogroups.com, <tk@...> wrote:

Hi Cristian,

It sounds like you will end up with a dual loop type of control with feedback from both the encoders on the motor shafts as well as the linear encoders.

Regarding the magnetic linear encoders I have had bad luck with some of them.  Usually they have magnets at a pitch of 2mm or so compared to an optical encoder with an optical pitch of 20um or so.  100X less basic resolution.  They then make up the different with some sort of electronic or software interpolation.  The result is usually noisy and sometimes delayed and non uniform.  Some Users have reported good results.  I just thought you should be aware of the issue.

KFLOP has a closed loop step and direction capability that would be the simplest  to use.   The drive would provide the motor shaft feedback. 

KFLOP can do dual loop feedback but it is more complicated and requires two Axis channels and two encoder inputs per motor. 

HTH
Regards
TK

Group: DynoMotion Message: 8347 From: Tom Kerekes Date: 9/13/2013
Subject: Re: Kanalog with step-dir servo drives. Is it worth?
Hi Cristian,

Notice in the data sheet the magnetic encoders are measured from 2mm magnet poles.  Also note things like accuracy +/-40um/m and hysteresis of up to +/-7.5um (depending on ride height).  Hysteresis will show up like backlash.  They do state repeatability (from same direction to eliminate hysteresis) of less than selected resolution (1um?).  That would be quite good unless they base that on a lot of averaging.  Again depending on your application they may work very well.  They are more rugged and reliable than optical scales but just not as accurate.

With linear scales a lot more mechanics are inside and corrected by the feedback loop which is good.  But this makes the control loop much more difficult to control and tune.  For example when motor torque is increased, and the motor stiction is overcome to start moving the motor, this fact will be impossible to detect using the linear scales until after any backlash and compliance in the mechanics actually starts moving the table.  Motor shaft feedback helps provide advance information that the motor is moving so motor torque can be reduced.  With only linear scale feedback it is very difficult to avoid overshoot and oscillation unless the motor torque is increased very gradually - which means the system is slow to correct errors.

There is no need to have any special ratio between the encoders. It is best to have as high of resolution as possible unless it exceeds the count rate of the Encoder input (1MHz for KFLOP).  You can match the encoder counts to the amplifier steps using the InputGain0 parameter.

HTH
Regards
TK