[FFWD]

Hey guys. This problem has plagued me for the past week or so...and I have officially hit a brick wall.

I have a multi fixture setup on the table. Hasn't moved for years. Machine is a 1990 Fadal 4020 with CNC88 controller driven by HSMWorks.

I'll use the edgefinder to dial in my fixture X and Y offsets on fixture numbers 1 and 2. Start making the parts. Everything is fine from start to finish.

Power the machine down for the night...come back next day...put new parts in the 1 and 2 fixtures...and start cutting. I find that my Y axis will be off anywhere from .020 to .040 off every single time on both fixtures.

X and Z are always just fine.

However, it makes all the features of the part just fine. Bolt hole spacing, slot widths, and what not are just as they should be. Deadnuts and consistent between parts. Just program zero shifts on me in the Y direction on both these stinkin' fixtures.

This problem only exists on fixture #1 and #2. Anywhere else on the table I have other fixtures, there are no issues.

At the recommendation of my tech (and Fadal USA), I even swapped axis controller cards. Swapped X with Y. Nothing changed. Problem still there.

Another test was to re-set my offsets, turn the machine off, and re-check offsets when I power the machine back up after a few minutes or so. Offsets will be fine and match my previous measurements. Only happens when the machine is left off for the night.

I think at one time, it also changed after only one cycle...I can't remember. I've done so much testing and measuring, I don't know which end is up anymore.

Of course, I did the obvious and made sure nothing wonky was going on with my Y stops on those particular fixtures. Solid as a rock.

Bottomline here....it has a mind of its own. I don't think there is a pattern to this. If there is, I haven't found it.

Any ideas on what's going on here? This is driving me absolute nuts chasing this issue down. I think I've only made two accurate, good parts during the same cycle since this problem reared.

Please save me from jumping off a cliff. If you need more info to help nail this problem, I'll be happy to oblige.

Thanks bunches everyone!

[RCaffin]

Power the machine down for the night...come back next day...put new parts in the 1 and 2 fixtures...and start cutting. I find that my Y axis will be off anywhere from .020 to .040 off every single time on both fixtures.

Known problem, with 2 sources (at least).

Servo systems: as the power goes down (or comes up), a power amp may give a twitch. Not much, just a couple of encoder slots. With AC servoes there could also be some twitch due to poling etc.

Stepper systems with microstep: system stops part way through a full step cycle - ie somewhere in the microstep cycle. When it powers back up it starts at a full step, or at microstep 0.

Solution: Re-zero every time you power up, OR leave the system powered up overnight.

Why only the Y axis? Murphy. Details. Anyhow, how do you KNOW the X & Z axes are not out by a step or two?
Why only overnight? Difference between a hot start and a cold start.
Do I see it myself? Yep. (DC servo w encoder)

Cheers

[FFWD]

I have done more tests on this issue last night.

I can now confirm it's not just leaving the machine off overnight that will bring back an incorrect Y axis...it can also be while the machine is running that the Y axis changes on me. For example, I ran a slotting op within the cycle, paused the program, measure, and it was fine. Finished the part. Started part #2, ran the entire cycle all the way through and the Y will got back "off".

You can also re-zero my fixture, start the same slotting op and it will slot incorrectly on the Y axis. Go back and check the fixture, and it will indeed be off. Crazy! Yes...it can happen that quickly.

So I can't just leave the machine on and/or re-zero every time I power up. This will not rectify the problem or even work around it. Interesting theory though.

How do I KNOW the X and Z aren't out as well? Mics and LOTS of time with an edgefinder. I have spent many hours trying to nail this problem.

Encoders and mini steps? Don't believe so as this problem is only on two fixtures...not on all the others. In other words, isolated...not global.

The only thing left that I can think of at this point is maybe there is a software glitch in the EPROM chips within the axis controller card? I might have swapped the cards, but I also kept the chips with their respective slots.

I still don't know...

[fizzissist]

Wow. This hurts.
Good basic machine, and there's a lot of 'em out there still.. so not likely an unheard of problem with the firmware.

Is Neil still around in the Fadal forum? I'd check to see if you can get his attention. Great guy, knowledgable and very helpful.

[FFWD]

Wow. This hurts.
Good basic machine, and there's a lot of 'em out there still.. so not likely an unheard of problem with the firmware.

Is Neil still around in the Fadal forum? I'd check to see if you can get his attention. Great guy, knowledgable and very helpful.

Thanks buddy. I'm at absolute wits end. I've drained my knowledge base!

I don't know Neil. I'm very new to this forum.

[fizzissist]

Thanks buddy. I'm at absolute wits end. I've drained my knowledge base!

Let's give you a bigger base...

http://www.cnczone.com/forums/fadal/...r_fadal-3.html

[FFWD]

Let's give you a bigger base...

http://www.cnczone.com/forums/fadal/...r_fadal-3.html

Absolute awesomeness buddy. I will give Yoda a call. He sounds really helpful and accommodating as you had said. Hopefully he has some answers (or at least a better track for me to follow).

Appreciate ya!

[RCaffin]

You can also re-zero my fixture, start the same slotting op and it will slot incorrectly on the Y axis. Go back and check the fixture, and it will indeed be off. Crazy! Yes...it can happen that quickly.(

Oh .... Well, that's a new one.

OK, so it has nothing to do with the overnight power cycling. The problem seems to be confined to one fixture and program? Other fixtures and programs don't have this problem? That suggests the problem is linked to either the fixture or some step in this particular program.

The obvious question is whether the problem happens EVERY time or just sometimes. Your earlier posting suggested that the answer is 'sometimes'.

Another question is whether the shift is always in one direction, or whether it can be in either direction. I am not sure yet what either anwer would mean though.

If the problem can happen while the program is running, and near the start of the program as you suggest, but it does not happen every time, then
* either the fixture itself or some part of the Y-axis HW can move when it shouldn't, OR
* some step in the program is very close to the limits of what the system can handle without losing steps.
Granted, neither seems super-likely.

As a thought - what happens if you slow the machine down by 20% for the whole program?

Cheers

[FFWD]

The problem seems to be confined to one fixture and program? Other fixtures and programs don't have this problem?

Yes. Confined to two fixtures I have that are close together. Other fixtures of different styles are not affected that are roughly 6 or so inches away.

The obvious question is whether the problem happens EVERY time or just sometimes.

Yep...it's still "sometimes" but when it works, it only works for one part. Definitely more "off" than "on".

Another question is whether the shift is always in one direction, or whether it can be in either direction.

One direction only. On the average, about .022" off...but it has been as much as .060".

What I've learned is that this machine will start cutting the part throughout the cycle just fine...it's just that part program zero to the machine is not MY part program zero. For example, the program will call up E1, machine will go there, cut the part, and end. Inspect the part, and find that everything it did has been shifted .020-.060 on the Y axis. I'll find that my fixture offsets for E1 are indeed off for the Y.

This has what is hard to swallow thinking it has anything to do with gibs, turcite, and/or coupler slop for the ballscrew and/or encoder.

As a thought - what happens if you slow the machine down by 20% for the whole program?

Haven't tried that one yet. But it seems like this has started happening when implementing HSMWorks for the CAM. Coincidental? That's what I have been thinking...but you really never know. I'm open to anything right now.

I'm about to the point of going through MDI and sending the machine to E1 and see what happens...as another test.

[RCaffin]

But it seems like this has started happening when implementing HSMWorks for the CAM. Coincidental? That's what I have been thinking...but you really never know. I'm open to anything right now.

I had better explain that I do not know anything about HSMWorks or Fadal systems. I use Mach3 and I write my own G-code programs. So I am working at a somewhat 'lower' level.
But going back to basics and checking everything is what I would do. If the problem only comes in at a higher SW level, well ...

Cheers