Hello Guest it is October 15, 2019, 01:30:20 PM

Author Topic: CS Labs CSMIO/P-A  (Read 16659 times)

0 Members and 1 Guest are viewing this topic.

Re: CS Labs CSMIO/P-A
« Reply #80 on: July 19, 2018, 03:32:33 PM »
Hi Craig
I am happy to try and tune differently to what I have already. Will it improve what it's doing now?
Cheers
Mick
Re: CS Labs CSMIO/P-A
« Reply #81 on: July 19, 2018, 04:28:12 PM »
Hi All

Mick, I did think you still had Kp set at 550. I cannot fault the way you have attempted to optimise tuning of the CSMIO, and if you feel that the axis is performing as well as it originally did, and/or are satisfied with it, so be it.

Out of interest, did you get the CSMIO scope to work, or did you just use the figure for following error? The scope would allow  you to see how fast the axis reacts to a step change, and how well damped the resulting transients are.

Like Craig, however, I feel that the velocity loop is weak and this may well be what is preventing the auto-tune completing sensibly as well as rendering your overall manual tune sub-optimal. If you feel inclined to push for better performance, this is the area that I feel needs your attention. You will need a scope, or as I have suggested you could probably use the CSMIO scope if you don't possess one. But if you proceed, do take care not to invoke the position loop as well. You would tune that, repeating the process you've already carried out, after you had optimised the velocity loop. Either use a signal generator to feed the command inputs, or if that is not available, try using Vff (only) with CSMIO as I suggested earlier.

Best of luck if you proceed with it. It might be wise to mark the original position of RV4 just in case you have to revert to the status quo.

Allan
Re: CS Labs CSMIO/P-A
« Reply #82 on: July 19, 2018, 05:43:30 PM »
Hi Mick,
if and how much improvement could be had is at best a wild guess, I would say that you could improve the
underlying servo velocity loop by 50%.

Given that you have followed written instruction only I think you have achieved a good result. To do better
you will need a scope and sig-gen to inform your tuning. While you could use the CSMIO scope and Mach/CSMIO
as a signal source I suspect the confusion that it would generate would make any gains difficult.

If I were doing it I would use my signal generator to produce a step signal, actually a 10V amplitude square wave
of period 10sec, and observe the error with the scope. I would be looking to improve the speed of response
with a 9% overshoot (sometimes called the Gibbs overshoot). RV2 would be left wide open. The tuning would be
principally done with RV1 (tacho gain) and RV4 (stability). Thereafter I would use my signal generator to produce a sine
wave and measure and plot the Bode frequency response and therefore deduce the gain and phase margins.

Does any of that sound familiar to you? If not then I suspect trying to do it by forum post will drive you up the wall!
I think your existing solution is good enough. You might in time come back and revisit it but I would suggest you move on
to other matters and make some chips.....

All in all I think you are to be congratulated on the result you have achieved given your unfamiliarity of feedback
control systems.

Craig
My wife left with my best friend...
     and I miss him!
Re: CS Labs CSMIO/P-A
« Reply #83 on: July 19, 2018, 06:06:04 PM »
Quote
All in all I think you are to be congratulated on the result you have achieved given your unfamiliarity of feedback
control systems.

Absolutely agreed.

Allan
Re: CS Labs CSMIO/P-A
« Reply #84 on: July 20, 2018, 01:16:18 AM »
Hi Craig/ Allan

No, I have not got a signal generator. I used the follow error in the CSMIO. I think I need to get on with the rest of the conversion for now. I can't see any difference now to what they were before, as far as the motor performance. That being said I would like to be in a position to be able to tune the whole system accurately.

What would I need as far as a scope is concerned? I have an Arduino which is capable of becoming a scope or is that ridiculous? Some of what you are saying in the last couple of posts is going over my head a little so please try and explain in laymen terms.

Really pleased where we are now and thank you for all the support.

Cheers
Mick
Re: CS Labs CSMIO/P-A
« Reply #85 on: July 20, 2018, 02:39:42 AM »
Hi Mick,
in recent years entry level  PC based oscilloscopes have become very very much cheaper and yes something like an Arduino could be pressed into
service. The technical requirements for an oscilloscope for tuning servos is fairly modest and you could very easily get a highly useful PC based instrument
for less than 100 pound, such instruments will often have a built in programmable signal generator.

I may do a little searching around and try to make some sensible recommendations for you.

Even with a suitable instrument you will have to learn to use it and then learn to apply the information obtained from its use to tune your servos.
That is not for the faint hearted, to make sense of things you will be required to be absolutely familiar with diagrams of the type I posted earlier.
Not only will you have to understand them but do some basic calculations with the data obtained.

Depending on your inclination it could be a fascinating learning experience which you may enjoy.....well within the parameters of a challenging and rewarding
hobby......or it could put you off for life!!

Craig
My wife left with my best friend...
     and I miss him!
Re: CS Labs CSMIO/P-A
« Reply #86 on: July 20, 2018, 02:43:32 AM »
Hi Craig
Sounds good. Do love a bit of Math so let me know your thoughts on a suitable scope.
Cheers
Mick
Re: CS Labs CSMIO/P-A
« Reply #87 on: July 20, 2018, 03:36:04 AM »
Hi Mick,
just blown away at the range of whats available....this caught my eye....Velleman is a genuine US instrument maker and I used Velleman before
when I worked at an automotive garage. I don't mean to say that there are not other brands and cheaper stuff....its just I trust this one:

https://www.ebay.com/itm/Velleman-PCSU200-USB-PC-Oscilloscope-and-Signal-Generator/112196139443?epid=13011370657&hash=item1a1f6939b3%3Ag%3AKQ8AAOSwKOJYH17u%3Asc%3AShippingMethodStandard%2197203%21US%21-1&_sacat=0&_nkw=usb+oscilloscope+and+signal+generator&_from=R40&rt=nc&LH_TitleDesc=0

Craig
My wife left with my best friend...
     and I miss him!
Re: CS Labs CSMIO/P-A
« Reply #88 on: July 20, 2018, 01:48:26 PM »
Thanks Craig

I will have a look into it but want to complete the conversion first. Will be looking at the spindle and drive next week. I'm sure I'll get stuck again. LOL

Appreciate your time on this.

Cheers
Mick
Re: CS Labs CSMIO/P-A
« Reply #89 on: July 20, 2018, 05:02:09 PM »
Hi,
most of the combined scope/signal generators have a minimum frequency of 1Hz or more which is too fast for testing servos.
This cheap ($20US) and cheerful signal generator might be more useful, it can go down to 0.01Hz:

https://www.ebay.com/itm/0-01Hz-5MHz-DDS-Function-Signal-Generator-Module-Custom-arbitrary-Waveform/292601759815?hash=item44206c8847%3Ag%3Ah78AAOSwgTJbIFCm&_sacat=0&_nkw=dds+signal+generator&_from=R40&rt=nc&_trksid=p2060353.m570.l1311.R1.TR8.TRC1.A0.H1.Xdds+signa.TRS0

If you went this way then you could choose from a multitude of two channel scopes, many for very much less than the Velleman I linked to earlier
on the basis of two scope input channels AND a signal generator. For example:

https://www.ebay.com/itm/Hantek-6022BE-20MHz-48MSa-s-Handheld-2-CH-PC-USB-Digital-Storage-Oscilloscope/122988824398?hash=item1ca2b4874e%3Ag%3ALxgAAOSwaeFak8VY&_sacat=0&_nkw=usb+oscilloscope&_from=R40&rt=nc&LH_TitleDesc=0

Craig
My wife left with my best friend...
     and I miss him!