Hello Guest it is March 28, 2024, 09:23:14 PM

Author Topic: Bridgeport Knee Mill Conversion?  (Read 270408 times)

0 Members and 1 Guest are viewing this topic.

Offline Davek0974

*
  •  2,606 2,606
    • View Profile
Re: Bridgeport Knee Mill Conversion?
« Reply #470 on: September 05, 2016, 06:07:53 AM »
Your'e a star Stirling, and Hood, and I'm an ass ;)

I'm so glad it was not just me getting the odd results, even though it was my fault.

I have now got a custom screen set from Ger21 with the A axis DRO removed and replaced with UserDRO(1444) so that should solve all issues instantly as it will not be connected to anything internally.

I guess pressing the stop button resets the interpreter?

I will abort further testing and go straight for the new screen-set and test on that, no point in beating the dead horse any more.

Ger21 was right in his reply though as even Hood's suggestion of using the OEMDRO(100) i think, is a Mach DRO for encoder 4 although it seems it works ok for this usage as he reports. Maybe it is more allied to the Mcode calling issue than the actual DRO used as Hood has his code in the buttons??

As for your last paragraph I am still trying to get my head round that conundrum! Sounds like a real hornets nest of trouble there, is there a safe way to not run into that issue or prevent it happening?

Offline Davek0974

*
  •  2,606 2,606
    • View Profile
Re: Bridgeport Knee Mill Conversion?
« Reply #471 on: September 05, 2016, 06:37:40 AM »
Just to nail the lid on this saga, I have modified the custom screen-set on which Ger21 kindly left a neat little gap next to my new user dro, and have managed to add a VB button, so i can now get rid of the double macro call as well as the OEMDRO issue - this should put my system perfectly in line with Hood's proven working version :)

Offline stirling

*
  • *
  •  2,188 2,188
  • UK
    • View Profile
    • www.razordance.co.uk
Re: Bridgeport Knee Mill Conversion?
« Reply #472 on: September 05, 2016, 06:47:32 AM »
EDIT: I posted this to reply to your previous post just as you posted your last reply. I can't be arsed to edit it so I'm pleased you've addressed BOTH issues.  ;)

I have now got a custom screen set from Ger21 with the A axis DRO removed and replaced with UserDRO(1444) so that should solve all issues instantly as it will not be connected to anything internally.

I'm hoping I've convinced you to address the M code as well.

I guess pressing the stop button resets the interpreter?

That would be my guess - though that was one thing I didn't see here. pressing cycle start STILL left the interpreter hung - but then I'm using an old version so maybe that was a "fix" for some related bug introduced sometime later.

I will abort further testing and go straight for the new screen-set and test on that, no point in beating the dead horse any more.

I'm REALLY hoping I've convinced you to address the M code as well.

Maybe it is more allied to the Mcode calling issue than the actual DRO used as Hood has his code in the buttons??

I'm REALLY REALLY hoping I've convinced you It's BOTH that need addressing.

As for your last paragraph I am still trying to get my head round that conundrum! Sounds like a real hornets nest of trouble there, is there a safe way to not run into that issue or prevent it happening?

I guess just being aware is half the battle. Don't have the time or the will to go off down another rabbit hole just now.

Offline Davek0974

*
  •  2,606 2,606
    • View Profile
Re: Bridgeport Knee Mill Conversion?
« Reply #473 on: September 05, 2016, 06:58:18 AM »
Thanks Stirling, yes, i think I have now addressed all issues on this balls-up, i will try and refrain from "thrashing" again but to somewhat mitigate my balls-up, I did test the use of the OEMDRO manually before plunging in and it did seem safe, i was unaware of the M-Code issue as I presumed that as the 2010 screens was working very well, there were no issues in just editing one of the called macros' .

Anyway, I am confident that all the issues will disappear tonight .

Thanks for your patience, apologies for having a higher than average specific density in the brain area :)

Offline stirling

*
  • *
  •  2,188 2,188
  • UK
    • View Profile
    • www.razordance.co.uk
Re: Bridgeport Knee Mill Conversion?
« Reply #474 on: September 05, 2016, 07:37:03 AM »
LOL - no worries - hopefully you'll make some progress.

You'll probably know enough about my attitude to software by now to know why I can't let this one go without comment:

this should put my system perfectly in line with Hood's proven working version :)

Working - yes - proven - no. It works for Hood on Hood's system - but I've been around Mach3 long enough to know it WILL fail on SOME other systems.

One suspect area is this:

Code "G90" & "G10" & "L1" & "P" & Tool &"Z" & Offset
DoOemButton(121)

You hopefully know enough by now about what the "code" call actually does to know why this can fail.
Does it matter? Not to Hood and not to me - might do to someone who comes across this thread sometime and copies it though.  ;)

Offline Davek0974

*
  •  2,606 2,606
    • View Profile
Re: Bridgeport Knee Mill Conversion?
« Reply #475 on: September 05, 2016, 07:44:06 AM »
Fair-do's :)

Yes i now know what can happen with those two lines of code, and also that the improved version fixes that issue, also not to apply a kludge over an issue even if it fixes it unless the issue cannot be fixed properly ;) Been very educational has this project.

Offline Davek0974

*
  •  2,606 2,606
    • View Profile
Re: Bridgeport Knee Mill Conversion?
« Reply #476 on: September 05, 2016, 03:49:54 PM »
Ok, recent issues being fixed now, moving right along - servo tuning.

Have done nothing on these apart from an auto-tune in CSMIO setup. I thought the results were ok, but having spent a few hours today scouring the 'net it seems they were pretty bad really. The first video here shows one axis after an auto-tune. Now, I have zero experience with servo's being stepper builds up to this build, so I stand to be fully corrected but, i think the graph shows pretty bad over-shoot or oscillation?

https://youtu.be/IOAS4LMJzcg

So, throwing caution to the wind, i dived in and removed all of the "I" and "D" parameters as well as the "ff" or fudge-factor i think its called. That seemed to improve things so i started increasing the gain in steps of 1,000 and the graph seems better to my untrained eye?

The following error is lower, dead-band is reached faster and the dials on the machine no longer overshoot and slowly wind back to a stop - this was a concern as i could see it affecting cut quality etc.

This video was shot after my messing around...

https://youtu.be/Yr4joMqo61Q

Now, please feel free to rain on my parade, but does the last video look happier than the first?
Is there any issue running with zero "I". "D" & "ff" values?
I can't hear any grumbling from the motors, there is no visible 'hunting' at idle, following errors are down and the dials now stop on a dime as shown.

The Z axis would not tolerate as much gain and did go into permanent oscillation once so i backed off a fair bit but still no "I", "D" & "ff" values.

Suggestions?
« Last Edit: September 05, 2016, 03:55:42 PM by Davek0974 »

Offline Hood

*
  •  25,835 25,835
  • Carnoustie, Scotland
    • View Profile
Re: Bridgeport Knee Mill Conversion?
« Reply #477 on: September 05, 2016, 06:23:43 PM »
Not looked at the vids yet but if the following error is low, no overshoot, no hunting then doesn't matter a toss what values you have, every setup will be different.

Hood

Offline Davek0974

*
  •  2,606 2,606
    • View Profile
Re: Bridgeport Knee Mill Conversion?
« Reply #478 on: September 06, 2016, 02:14:39 AM »
Thats pretty much what i was led to believe from my reading, seems permanent oscillation is bad even if it is tiny, slow settling is also bad from a contour point of view, and as you will see in the vids, it certainly looks better than the auto-tune version i think.

Offline Davek0974

*
  •  2,606 2,606
    • View Profile
Re: Bridgeport Knee Mill Conversion?
« Reply #479 on: September 06, 2016, 02:11:02 PM »
Well that was a blast :)

I upped my speed and acceleration a bit, tested the motions again, checked the macro issue and all was good so went for the finish on  my part-done job. All went perfectly, tool-paths were good, CAM was good, CAD was good - this was my 1st part from Fusion360 too :)

Dimensions were spot-on as well, tool-changes were fast and easy as well. Short video of one of the cuts...

https://youtu.be/dzorJYUJpBk

Picture of finished part below.

BUT,

even though all went well the first check-run failed because mach started running the code without starting the spindle!
It would work from the panel button as well as the screen button but NOT from MDI or G-Code.

I have seen this before on the engraver and a Mach restart fixes it as it did this time, but of course you have to re-home. The only message you sometimes get is "Scripter compile error in M3.m1s" but as usual that file only contains DoSpin...

Any ideas??