Machsupport Forum

Mach Discussion => General Mach Discussion => Topic started by: Triton on March 15, 2013, 04:25:01 AM

Title: G83 Random Bug
Post by: Triton on March 15, 2013, 04:25:01 AM
Machine :Matchmaker 800 Vertical Mill
Using Machmotion Ultimate screen

Hello,
I am a new user of Mach 3 it has been recently retrofitted to replace an old Ah Ha system.
Everything works fine except:
When using G83 peck drilling cycle IE.
T3M6
G0G90X0.Y-10.A0.S1500M3
G43H3Z2.M8
G99G83Z-29.R2.Q2.F100.
G80

The control will work perfectly for any random number of components and then do this:
Feed approach to feed plane
Rapid to peck depth
Rapid out
Which it then repeats to the full depth of the hole.
This breaks drills.

It is as if the Feed and Rapid get reversed.

The most strange thing is that the frequency of this problem is completely random.
Sometimes it works fine from the start then changes. Other times the problem is there from the first component.

I would appreciate any help or suggestions.
Thank you
Dennis


Title: Re: G83 Random Bug
Post by: Hood on March 15, 2013, 06:58:42 AM
What version of Mach are you using?
If you attach your xml and also the m1083.m1s macro I will take a look. You will probably have to rename or zip them as the forum only accepts a file name once.

Hood
Title: Re: G83 Random Bug
Post by: Triton on March 15, 2013, 07:17:29 AM
Hello Hood,

Version R3.043.022

The m1083.m1s file was in the main mach3 folder not in the profile folder.

Files should be attached.

Many thanks
Dennis
Title: Re: G83 Random Bug
Post by: HimyKabibble on March 15, 2013, 08:46:30 AM
Ha!  I stopped using Mach3 over a year ago because of exactly that bug!  It broke dozens of drills.  I spent hours on the phone with both Brian Barker and Greg Cary over a period of several months trying to isolate it, and we never were able to.  On the rare good days, I could drill hundreds of holes without a problem.  Other days, I was lucky if I could drill three holes before it tried to rapid the drill to China partway through a drilling cycle.  It was completely random, and in my case had nothing to do with G83, as I used hard-coded peck cycles.

Just curious - Are you using a motion controller, or just the parallel port?

Regards,
Ray L.
Title: Re: G83 Random Bug
Post by: Triton on March 15, 2013, 09:06:51 AM
Ray,
Thanks for that. At least we're not alone.

We are using 2 parallel ports via an Apollo breakout board.

We are running Windows 7.

It would be very interesting to compare our specifications to yours

If the error occurs it does it on the first peck and then again all through the cycle.

So it doesn't seem to go wrong once the peck has started.

Each hole is drilled OK if it's first peck is OK.

We think the G00 and the G01 are getting swapped somehow.

We haven't had a problem with G81 yet.

If anyone else has any input on this it will be greatly appreciated.

Dennis
Title: Re: G83 Random Bug
Post by: Hood on March 15, 2013, 10:09:10 AM
Ray,
 Just a shame you declined Brians offer to fly you up with your computer to try and find your issue, didnt realise this was one but good to know.
Hood
Title: Re: G83 Random Bug
Post by: HimyKabibble on March 15, 2013, 10:15:16 AM
Ray,
 Just a shame you declined Brians offer to fly you up with your computer to try and find your issue, didnt realise this was one but good to know.
Hood

Say what?  There was no such offer.  And even if there was, no guarantee it would've led to a solution.  It could sometimes work for days at a time with no problem, other times I couldn't do even one hole.

Regards,
Ray L.
Title: Re: G83 Random Bug
Post by: Hood on March 15, 2013, 10:27:18 AM
Maybe I am dreaming Ray but I was sure Brian mentioned that in a post on the forum.
Hood
Title: Re: G83 Random Bug
Post by: Hood on March 15, 2013, 10:31:59 AM
Well just did a search and here is what I found, looks like my memory is still reasonably good :)

Quote
I am working on Mach3 all the time looking for issues. I spent months working on Rays issue and found some of them (both of us to blame). At the end of the day he will end up with his own little interface for the KFlop and I am happy that he is happy. But I lost months of dev time and I wouldn't change what I did. The fact that he didn't get what he wanted was not a lack of effort. I offered more then one time to fly him out with his PC work work at the office here and to get it working as he would like. I also offered to fly out to work on his machine. I don't know what more I can do other then to build up machine and send it to him. If that is not good support then I just don't know what to do and I give up.

And your reply

Quote
Just to backup what Brian said, his support has been truly outstanding,


Post is here if you want to read it http://www.machsupport.com/forum/index.php/topic,20546.msg143035.html#msg143035

Hood
Title: Re: G83 Random Bug
Post by: HimyKabibble on March 15, 2013, 10:37:22 AM
Hood,

I don't recall Brian ever making such an offer to me, and I don't how I could've done it anyway.  The "PC" is an integral part of the entire controller on that machine, which is a 24x24x12" steel box jam-packed with electronics, weighing probably 150#.  Moving it is simply not practical.  He did do everything he possibly could remotely, as did Greg, but it seemed every time the thing was "instrumented", it would behave.  As soon as I tried to get any real work done, it would act up again.  We found and fixed several other problems, but not this one.

Regards,
Ray L.
Title: Re: G83 Random Bug
Post by: Triton on March 15, 2013, 11:24:34 AM
Ray,

Did you always work with the 1024 screen?

Dennis
Title: Re: G83 Random Bug
Post by: HimyKabibble on March 15, 2013, 11:38:24 AM
Ray,

Did you always work with the 1024 screen?

Dennis


Dennis,

No, I used a completely custom screen set.  I also used a SmoothStepper - both USB and Ethernet.  We always suspected what I saw was some obscure communications issue between Mach3 and the SmoothStepper.  You're the only person besides me I've ever heard of who's had this problem.

Regards,
Ray L.
Title: Re: G83 Random Bug
Post by: Triton on March 15, 2013, 12:05:35 PM
Ray,
I just ran the cycle using the 1024 screen set and it misbehaved just the same.

My electronics man says we are unlikely to find the problem and we should just try and work around it.

However you can't beat a try.

There must be something in common between our two set ups.

Or should I just wait for Mach 4?

Dennis
Title: Re: G83 Random Bug
Post by: Hood on March 15, 2013, 02:38:43 PM
Dennis,
I know you say its random but how often does it happen? If Brett could get it to repeat on his computer I think he can hook up to Brians computer with the source code on it and Brian can see whats going on. I will see if Brett will chime in and hopefully something can be found.
Hood
Title: Re: G83 Random Bug
Post by: Triton on March 15, 2013, 02:54:54 PM
Hood,
Thank your for your interest.
We reckon over the last couple of days it's about 1 in 4 times misbehaving.
However it is random so don't hold your breath whilst waiting.
I think I must have something in common with Ray's setup.
Is there anything else I can do to assist.
Ive just been reading about the Mach4 situation and realized that it isn't an option.

Regards
Dennis
Title: Re: G83 Random Bug
Post by: Hood on March 15, 2013, 03:02:45 PM
I will get Brett to look in and  if he can get it to do it on his computer then I am sure Brian will find it.

Mach4 is not out yet, its getting closer but..... well everything gets closer ;D

Hood
Title: Re: G83 Random Bug
Post by: Triton on March 15, 2013, 03:17:47 PM
Thank You
Title: Re: G83 Random Bug
Post by: Chaoticone on March 15, 2013, 03:33:57 PM
Dennis, do me a favor and post you xml and the g code your running when it shows its head.  On second thought, copy and zip your entire mach 3 folder and send to me via email please.   I'm redoing the controls on my mill right now but I will do some testing as soon as I get it back together.  I hate you or anyone else is having this problem.  The random ones are the toughest to dig out.  What version of mach are you running?

Brett 
Title: Re: G83 Random Bug
Post by: Triton on March 15, 2013, 06:13:35 PM
Brett,
Thanks for your interest.

It's a bit late now.
I will send it tomorrow.

Regards
Dennis
Title: Re: G83 Random Bug
Post by: Chaoticone on March 15, 2013, 06:19:25 PM
Thanks Dennis,  just whenever you get a chance.

Brett
Title: Re: G83 Random Bug
Post by: BR549 on March 15, 2013, 08:07:34 PM
The random misbehaving of the Z axis in drill cyles is a VERY old bug That has come and gone several times over the years in various verisons.

Just a thought, (;-) TP
Title: Re: G83 Random Bug
Post by: Triton on March 16, 2013, 02:58:59 AM
Hello TP

Thanks for the input.
If you could expand on your information I would appreciate it.
Especially the "gone" part.

Dennis
Title: Re: G83 Random Bug
Post by: BR549 on March 16, 2013, 11:24:30 AM
You may want to try older versions of MACh3 as some seem to not be effected.

Just a thought, (;-) TP
Title: Re: G83 Random Bug
Post by: Triton on March 16, 2013, 01:27:52 PM
TP
Thank you,
That is a constructive suggestion.
Dennis
Title: Re: G83 Random Bug
Post by: TOTALLYRC on March 18, 2013, 05:45:19 AM
The random misbehaving of the Z axis in drill cyles is a VERY old bug That has come and gone several times over the years in various verisons.

Just a thought, (;-) TP

Just for information,
I have not had this problem on either my DSPMC driven mill which has done 1ks of g83 holes nor my PPort driven router which has about the same number of holes in G83 and lot more G81.
Is this know to only occur on the smoothstepper or has it been documented on any other motion controller?

I have also run several different versions of Mach3 and windows from w2k to win 7.

Mike
Title: Re: G83 Random Bug
Post by: Hood on March 18, 2013, 06:24:38 AM
The OP is using the PP as far as I know.
Hood
Title: Re: G83 Random Bug
Post by: Triton on March 18, 2013, 07:10:39 AM
We are using 2 parallel ports through a Apollo board provided by mach motion
Title: Re: G83 Random Bug
Post by: Triton on March 18, 2013, 08:25:29 AM
We don't really understand  the concept of smoothstepper.
So we don't know if we have it, or if we are using it.
If we are using it, can we turn it off somewhere?

Thanks
Title: Re: G83 Random Bug
Post by: Hood on March 18, 2013, 08:27:42 AM
You are connected to the parallel ports so you are not using a SmoothStepper. The Smoothstepper is an external motion controller which takes the place of the parallel ports.
Hood
Title: Re: G83 Random Bug
Post by: Triton on March 18, 2013, 08:31:55 AM
Thanks Hood.
We can eliminate that then.
I have zipped up my entire Mach3 folder and sent it with the troublesome programme to Brett.
Dennis
Title: Re: G83 Random Bug
Post by: Hood on March 18, 2013, 08:36:36 AM
Hopefully Brett will have the test machine up and running soon. In the mean time have you tried a different version of Mach? Recommend you try 056 or 057 as I have found them to be very stable on my machines. Others have said up to 062 is also fine but please dont use 066 or 067 as they have issues for a lot of people and in my opinion 066 should not be on the website as the lockdown version.

You will find the other versions on the ftp site
ftp://anonymous@machsupport.com/Mach/ (ftp://anonymous@machsupport.com/Mach/)

Hood
Title: Re: G83 Random Bug
Post by: Triton on March 18, 2013, 08:39:03 AM
Thanks Hood,
We will have a look this afternnon.
Dennis
Title: Re: G83 Random Bug
Post by: Triton on March 19, 2013, 08:47:06 AM
We have tried Mach3 version 57 and left it on the machine at the moment.

Allowing for the fact that this is a random bug these are the results.

A. After start up of the Mach3 software the G83 programme misbehaves as before but only for the fist run of the programme.

After that it seems to behave properly until the Mach3 software is restarted when it seems to perform as line A.

So we can run the programme in air for the first run while it sorts itself out and then put tools in the spindle and put on a pair of brown corduroy trousers to run the job.








Title: Re: G83 Random Bug
Post by: Chaoticone on March 19, 2013, 10:00:24 AM
Dennis, in the test files you sent there is no G83 in test2.txt  There is a G83 on line 9 in test.txt.  If you would, mail me 2 complete files or post them here.  One known good and one known bad.  The closer they match one another the better but make sure they are known to be either fine or problematic.

Brett
Title: Re: G83 Random Bug
Post by: Triton on March 19, 2013, 10:24:07 AM
Brett.

Only the test file has a G83.

Test 2 was written out long hand to check that the machine would work normally and it does.

We have not managed to make a G83 file that will work without the bug.

Please see my last post where I tried an alternative version of Mach3.
As you will see the bug is only there on the first run.

Please let me know if I can send you anything else.

Dennis

 
Title: Re: G83 Random Bug
Post by: Chaoticone on March 19, 2013, 10:30:22 AM
OK, I'm sorry, I misunderstood.  I thought it was a G83 random bug but your saying it is a bug with G83 anytime it is used?  I hope so, that should make it easier to pin down.

Yeah, I saw that after one air cut it was fine in an earlier version.

Brett
Title: Re: G83 Random Bug
Post by: Triton on March 19, 2013, 11:52:31 AM
Brett,
Yes that's it.
Whenever we use the G83 we can nearly always produce the bug that seems to swap the feed speed with the rapid approach speed.
We can make a video if that helps.
Dennis

Title: Re: G83 Random Bug
Post by: Chaoticone on March 19, 2013, 11:55:47 AM
Thanks Dennis, that wont be necessary right now.  Let me get my machine back together and do some testing.  I appreciate your help with this.

brett
Title: Re: G83 Random Bug
Post by: rcaffin on March 22, 2013, 05:47:55 AM
Quote
He did do everything he possibly could remotely, as did Greg, but it seemed every time the thing was "instrumented", it would behave.  As soon as I tried to get any real work done, it would act up again. 
I read this to mean that while the version of Mach3 had debug diagnostics in it the program would run correctlly, but when the diagnostics were removed it would fail.

IF (!) this is correct, then what you probably have is a race conditions in the code. They are well known elsewhere for creating random behaviour which does not appear when in debug mode. The reason is that the debug instructions alter the internal timing so the race condition goes away. The very devil to sort out! Bugs fixes in various releases will also alter the timing, affecting the frequency of the crash.

Me, I am running 022 and probably staying there until Mach4 is released. I can recognise certain well-known bugs in subroutine layering (Hi Ray) and just program around them.

Cheers
Title: Re: G83 Random Bug
Post by: rufan on August 09, 2013, 04:41:26 PM
Dear,

Just recently finish retrofit on a Brother TC-225.
I just noticed that I also have problems with this random G83.
It seems to happend only when drilling the firts hole of a part.
I am using the 66 version. Tomorrow I will downgrade the version and hope this helps.

Thanks for this topic otherwise I was troubleshooting for a long time.
Title: Re: G83 Random Bug
Post by: rcaffin on August 09, 2013, 05:51:11 PM
Hi Rufan

I found it advisable to always include the X & Y coordinates in the g83 instruction even if I had previously moved to the right XY position. Sometimes Mach seemed to forget where it was if many subroutine layers were involved. Of course, sometimes the g83 worked fine without the X & y parameters ...  I do a lot of air cutting first.

Cheers
Title: Re: G83 Random Bug
Post by: Triton on August 10, 2013, 04:29:08 AM
Rufan,
We never found an answer to the G83 problem.
We have to work around the problem by using another machine.
The only answer I can think of is to wait for Mach4 or write your
drilling programme long hand.
Dennis
Title: Re: G83 Random Bug
Post by: rcaffin on August 10, 2013, 04:36:34 AM
Quote
The only answer I can think of is to wait for Mach4 or write your
drilling programme long hand.
Yup, custom drilling programs work very well.

Cheers
Title: Re: G83 Random Bug
Post by: BR549 on August 10, 2013, 01:37:35 PM
Actually  some of the SUBs I used were better than the Canned cycles. For example a deep drilling sub did a rapid BACK to just above the drilled depth so you did NOT have to feed all the way back down as in the Mach canned cycles. I also put in a deep first pass to get to depth pecking.

Just a pain sometimes when you cannot use a stock canned cycle. Never had much trouble back on a 1mhz PC it only got qweird when I moved up to a 2.2GHZ machine. I believe a race condition existed at that point that was caused by software

(;-) TP
Title: Re: G83 Random Bug
Post by: Triton on August 10, 2013, 02:58:44 PM
TP,
Thanks for that, constructive comments are appreciated.
Dennis