Dynomotion

Group: DynoMotion Message: 8058 From: Roland Date: 8/2/2013
Subject: what is your typical following error in your system?
This is sort of related to KFLOP as I may switch over to controlling the amps via analog velocity commands using KFLOP instead of pulse step/dir and using the EN-204 amp for tuning.

While waiting for my KFLOP to arrive, I hooked up one of the Adept linear robot modules with 300 watt 4 pole Tamagawa TS4073 brushless motor and 8000 count encoder. Mach3 can only drive this at around 450rpm on my computer (around 60khz) but I figured I could set up the drive tuned as best possible for the KFLOP.

Making a file of back and forth movements from 100IPM to 300IPM and fastening a 15 pound load to the linear stage, I was able to tune the system to have a following error of about 8 steps maximum on the reversal points and about 2 steps average during movements. This stage has a 20mm lead, so 1 step = 2.5um. Thus I have a following error on the worst case of 20um and a continuous path following error of 5um. I had the acceleration set to 250ms from zero to max velocity in Mach3. If I went to 500mS acceleration, I could get the max following error down to around 5 or 6 steps, or 15um.

This sounds pretty good to me but I have no idea what others may be getting. The fiber laser I am experimenting with has a focused beam diameter of 10um with my finest lens, so you can see why I would want the very smallest error in the movements. Of course a lot of cuts would not be such a fast abrupt change in a axis movement, but it seemed like a good way to test.
Group: DynoMotion Message: 8064 From: Tom Kerekes Date: 8/3/2013
Subject: Re: what is your typical following error in your system?
Hi Roland,

I think the errors are dependent on your system and requirements so it is hard to come up with a typical answer.  But servo systems need some amount of error to make corrections so it is unusual to have less than several counts of following error.

One thing I don't understand is your interest in high speeds.  It seems like with your tiny geometries your speeds would be acceleration limited to very low speeds.  If I did the math right it would take ~16mm to get up to 300ipm and then another 16mm to stop.  When cutting a 1mm square and assuming 20in/sec2 acceleration the max speed would be 53IPM.

Regards
TK

Group: DynoMotion Message: 8066 From: Roland Date: 8/3/2013
Subject: Re: what is your typical following error in your system?
Well, true about the speed, but I also want to play with raster engraving over a larger area but using the same setup as micromachining. There also may be some case where I need to rapid from one area to another, but of course I can always wait for the system to settle before triggering the laser and starting a new cut.

I am happy enough with the 5um to 15um following error right now. I did the same test on a 10mm pitch slide and of course the total error was less because there are more steps per inch. I was getting 2.5um of following error on a 100IPM move and less than 10um of error when reversing direction at 200mS acceleration.



--- In DynoMotion@yahoogroups.com, Tom Kerekes <tk@...> wrote:
>
> Hi Roland,
>
> I think the errors are dependent on your system and requirements so it is hard to come up with a typical answer.  But servo systems need some amount of error to make corrections so it is unusual to have less than several counts of following error.
>
> One thing I don't understand is your interest in high speeds.  It seems like with your tiny geometries your speeds would be acceleration limited to very low speeds.  If I did the math right it would take ~16mm to get up to 300ipm and then another 16mm to stop.  When cutting a 1mm square and assuming 20in/sec2 acceleration the max speed would be 53IPM.
>
>
> Regards
> TK
>
>
>
> ________________________________
> From: Roland <lightdoesnotage@...>
> To: DynoMotion@yahoogroups.com
> Sent: Friday, August 2, 2013 4:52 PM
> Subject: [DynoMotion] what is your typical following error in your system?
>
>
>
>  
> This is sort of related to KFLOP as I may switch over to controlling the amps via analog velocity commands using KFLOP instead of pulse step/dir and using the EN-204 amp for tuning.
>
> While waiting for my KFLOP to arrive, I hooked up one of the Adept linear robot modules with 300 watt 4 pole Tamagawa TS4073 brushless motor and 8000 count encoder. Mach3 can only drive this at around 450rpm on my computer (around 60khz) but I figured I could set up the drive tuned as best possible for the KFLOP.
>
> Making a file of back and forth movements from 100IPM to 300IPM and fastening a 15 pound load to the linear stage, I was able to tune the system to have a following error of about 8 steps maximum on the reversal points and about 2 steps average during movements. This stage has a 20mm lead, so 1 step = 2.5um. Thus I have a following error on the worst case of 20um and a continuous path following error of 5um. I had the acceleration set to 250ms from zero to max velocity in Mach3. If I went to 500mS acceleration, I could get the max following error down to around 5 or 6 steps, or 15um.
>
> This sounds pretty good to me but I have no idea what others may be getting. The fiber laser I am experimenting with has a focused beam diameter of 10um with my finest lens, so you can see why I would want the very smallest error in the movements. Of course a lot of cuts would not be such a fast abrupt change in a axis movement, but it seemed like a good way to test.
>