Hello Guest it is March 28, 2024, 12:31:18 PM

Author Topic: I truely believe it's a bug with many versions:  (Read 7137 times)

0 Members and 1 Guest are viewing this topic.

I truely believe it's a bug with many versions:
« on: December 22, 2009, 08:45:17 AM »
I believe its' a bug, Therefore, I put a bug report in the appropriate place, however, maybe any of you can open my eyes:

I have been running mach occasionally for a year now, all in all it went ok. however, I did mind that I had sometimes I needed to cancel the CV to get a proper corner. So, I went on and changed that on the config, for corners > 44.

The computer is a 2.5GHz, 2GB ram, winxp sp3.It is only running mach, commanding a router.

I used version 3.042.020 for most of the time. however, when I ran into problems with the steppers humming lately, I decided to upgrade to version 032.
Version 032 runs on my machine badly: it doesn't obey the velocity limit, and runs my steppers 1.4 times faster than the 020, and it didn't solve my stepper hum problem. so I downgraded to 029. the 029 obeys velocity limit, but still carries the same problem.

FINALLY: the problem description:
I run various g-codes. one of them is attached here below. the most run without any problems, however, this one caused humming. So I cut the velocity and accel in half, and still it humms. the funny thing is, it starts the humming when in full speed, without any apparent reason. when jogging it gives no hum at all, even though the jog is even faster.
IMPORTANT: I found that when I run the code with different lookahead(tried 1, 2, 5, 20, 50, 200, 400, 500, 1000),  it does the humming on different program lines, but it is always the same line per lookahead value.
This humming causes loss of steps of course, and the programs keeps running from the supposed position, which is already wrong.

I really believe it has a strong relation to the lookahead, please help, I really have no solution, this experimental piece of code that is attached demonstrates all the problems. all it does is v-carve a rectangle, then with a 10 mm flat it makes a bottom, then a 25.4(1``) ball it makes another inner slope. supposes 19mm thick mdf.

G00 G49 G40.1 G17 G80 G50 G64 G90 G58

G21

(V-Carving 3)

M6 T1

M03 S4851

G00 Z60.0000

X96.0010 Y26.0010

G01 Z13.0000  F3000.0

X101.8451  F3500.0

G17

G02X96.0010Y31.8451I0.1549J5.9990

G01 Y26.0010

X95.9010 Y25.9010 Z13.1000

X96.0010 Y26.0010 Z13.0000

Z60.0000  F9000.0

G00

X102.0000 Y273.9010

G01 Z13.1000  F3000.0

X338.0000  F3500.0

G02X343.9010Y268.0000I0.0000J-5.9010

G01 Y32.0000

G02X338.0000Y26.0990I-5.9010J0.0000

G01 X102.0000

G02X96.0990Y32.0000I0.0000J5.9010

G01 Y268.0000

G02X102.0000Y273.9010I5.9010J0.0000

G01 Z60.0000  F9000.0

G00

X95.9010 Y274.0990

G01 Z13.1000  F3000.0

Y25.9010  F3500.0

X344.0990

Y274.0990

X95.9010

X96.0010 Y273.9990 Z13.0000

Y268.1549

X96.0151 Y268.4394

X96.0167 Y268.4555

Z13.0004

X96.0000 Y268.0000 Z13.0010

X96.0161 Y31.5507 Z13.0002

X96.0000 Y268.0000 Z13.0010

X96.0167 Y268.4555 Z13.0004

Z13.0000

G02X101.5445Y273.9833I5.9822J-0.4544

G01 Z13.0004

X102.0000 Y274.0000 Z13.0010

X338.4493 Y273.9839 Z13.0002

X102.0000 Y274.0000 Z13.0010

X101.5445 Y273.9833 Z13.0004

Z13.0000

X101.5606 Y273.9849

X101.8451 Y273.9990

X96.0010

Z60.0000  F9000.0

G00

X343.9990

G01 Z13.0000  F3000.0

X344.0990 Y274.0990 Z13.1000  F3500.0

X343.9990 Y273.9990 Z13.0000

X338.1549

G02X343.9833Y268.4555I-0.1546J-5.9983

G01 Z13.0004

X344.0000 Y268.0000 Z13.0010

X343.9839 Y31.5507 Z13.0002

X344.0000 Y268.0000 Z13.0010

X343.9833 Y268.4555 Z13.0004

Z13.0000

X343.9849 Y268.4394

X343.9990 Y268.1549

Y273.9990

Z60.0000  F9000.0

G00

Y26.0010

G01 Z13.0000  F3000.0

X344.0990 Y25.9010 Z13.1000  F3500.0

X343.9990 Y26.0010 Z13.0000

Y31.8451

G02X338.4555Y26.0167I-5.9983J0.1546

G01 Z13.0004

X338.0000 Y26.0000 Z13.0010

X101.5507 Y26.0161 Z13.0002

X338.0000 Y26.0000 Z13.0010

X338.4555 Y26.0167 Z13.0004

Z13.0000

X338.4394 Y26.0151

X338.1549 Y26.0010

X343.9990

Z60.0000  F9000.0

G00

X90.0010 Y279.9990

G01 Z19.0000  F3000.0

X95.9010 Y274.0990 Z13.1000  F3500.0

Z60.0000  F9000.0

G00

X349.9990 Y279.9990

G01 Z19.0000  F3000.0

X344.0990 Y274.0990 Z13.1000  F3500.0

Z60.0000  F9000.0

G00

X349.9990 Y20.0010

G01 Z19.0000  F3000.0

X344.0990 Y25.9010 Z13.1000  F3500.0

Z60.0000  F9000.0

G00

X90.0010 Y20.0010

G01 Z19.0000  F3000.0

X95.9010 Y25.9010 Z13.1000  F3500.0

G00 Z60.0000

(Engraving)

G53 X150 Y1740

M6 T20

M03 S4851

X102.0000 Y32.0000

G01 Z13.1000  F3000.0

X220.0000  F6500.0

X338.0000

Y150.0000

Y268.0000

X220.0000

X102.0000

Y150.0000

Y32.0000

Z13.0000

Y150.0000

Y268.0000

X220.0000

X338.0000

Y150.0000

Y32.0000

X220.0000

X102.0000

G00 Z60.0000

(Engraving)

G53 X150 Y1740

M6 T3

M03 S4851

X107.0000 Y37.0000

G01 Z13.0000  F3000.0

X220.0000  F6500.0

X333.0000

Y150.0000

Y263.0000

X220.0000

X107.0000

Y150.0000

Y37.0000

G00 Z60.0000

G00 Z155

M5 M9

M30
Re: I truely believe it's a bug with many versions:
« Reply #1 on: December 22, 2009, 08:49:17 AM »
one more thing: the missed stepper steps always happen on a line that has a z move line following. ???

Offline Tweakie.CNC

*
  • *
  •  9,196 9,196
  • Super Kitty
    • View Profile
Re: I truely believe it's a bug with many versions:
« Reply #2 on: December 22, 2009, 01:26:54 PM »
Hi Omri

Don't know if this is any help but I just ran your code without issue other than it scared the pants off me because I don't usually use feedrates that high.
I only cut air, with the spindle motor switched off, but there were no unusual stepper noises and as far as I could tell no missed steps.

My PC is running Pentium4 at 2.4 Gig with WinXP Sp1(I had mega problems with Sp2 onwards) and Mach R3.042.032.

Just out of curiosity - Can you set your motors (vel. and accel.) to a high rate which works OK in the motor tuning page but then they stall when jogging in the main Mach window ?.

Tweakie.
PEACE
Re: I truely believe it's a bug with many versions:
« Reply #3 on: December 22, 2009, 01:51:09 PM »
Hi Tweakie,

Well, I tried all sorts of configurations, and was able to cause missed steps while jogging. it happened when I tried a high frequency (45KHz) which enabled me a higher speed on the steppers, and since I use a gas spring to balance Z, the very fast up move caused missed steps midway up. it was too fast for the spring.

I really think this is a mach issue with my configuration, how can you explain otherwise that the missed steps happen in different portions of the code for different lookahead value, and that the missed steps are perfectly repeatable per lookahead value??

Thanks for your kind help. we now know for sure that the code is 100%. I''ll appreciate any idea, since I'm out of new ideas for finding this one.
Re: I truely believe it's a bug with many versions:
« Reply #4 on: December 22, 2009, 02:53:51 PM »
I too have found that velocity behavior is overfeeding with version 0.032.

Please fix this.
Re: I truely believe it's a bug with many versions:
« Reply #5 on: December 22, 2009, 09:33:10 PM »
I see that you are getting the error in a jog... that may be a problem with phase jitter.

Also I see you say you are getting an over volocity.. Could you please tell me if you get the error in Exact stop? I know where there is a problem with CV and I have added some code to check for this but there are times the CV will enter the move a bit to fast..

Thanks
Brian
Fixing problems one post at a time ;)

www.newfangledsolutions.com
www.machsupport.com

Offline RICH

*
  • *
  •  7,427 7,427
    • View Profile
Re: I truely believe it's a bug with many versions:
« Reply #6 on: December 22, 2009, 09:36:24 PM »
Brian,
Off thread but while your here .....
check out the COMPLEX THREADING WIZARD post...
http://www.machsupport.com/forum/index.php/topic,13498.0.html
RICH

modified / added link
« Last Edit: December 22, 2009, 09:39:41 PM by RICH »
Re: I truely believe it's a bug with many versions:
« Reply #7 on: December 23, 2009, 06:56:35 AM »
Hi Brian, thanks for joining in.

If you'll look at the top of the Gcode from my first post here, you'll see the G64, changed that and ran again with G61. the exact same behavior. I'll tell you more than that:
The behavior is 100% repeatable. no matter what g64/1, the only thing that matters is the lookahead value. the machine starts the move, gets to full speed, and somewhere along the way starts loosing steps. if I linger with the E-Stop, it even causes my Geckos 203V light up the error led.

About the over velocity - only happened on version 032, I now run version 029, and get none of the over velocity.

What say you? can it be fixed? I have examples I promised to deliver during next week (mdf doors), Should I delay them?

One more thing that may help: in my configuration, running on 500 lookahead will cause loosing steps on the first jog to the work. this is about 1.4 meters jog to the G59.

Thanks in advance, any help appreciated,

Omri.
Re: I truely believe it's a bug with many versions:
« Reply #8 on: December 24, 2009, 08:54:48 AM »
Another thing worth mentioning is that feeding the commands one by one on the MDI screen (second screen) causes no trouble at all. The moves are performed without a problem.
Re: I truely believe it's a bug with many versions:
« Reply #9 on: December 24, 2009, 10:18:28 AM »
I would like to have your XML to test. You can upload it here in a zip file
Fixing problems one post at a time ;)

www.newfangledsolutions.com
www.machsupport.com