Solved problem and a severe disappointment

[lloydsp]

I've had a problem on my newer 5x9 bed router for about seven months.  One or the other of the two Y-axis servo drivers will fault during a run.  Today, I spent an hour on the phone with the manufacturer, and they stepped me through a 'realignment' of the gantry in Y.  Since there are two y servos/screws, and two homing switches, if anything gets skew between the two, you get an over-current on the drivers as they attempt to maintain positions or accelerate.

I loosened all the gantry elements, homed twice, then secured the gantry again.  It worked fine.  About two hours later, it failed the same way.  But it occurred to me that I had done several 'home' operations during that time.  So this time, I repeated the exercise, only doing ONE home to accomplish the alignment, then never doing another.

I ran my exerciser routine for hours, and the current never increased a bit.  Then I did a home.  It jumped-up dramatically.  I did another, and it went past 100% allowable, and faulted.

I've been complaining for seven months that my home position was varying by up to 30-thousanths on EVERY homing.  They denied that could happen (Their switches 'never fail').  Now I have proof that the home switches are at fault.  When one triggers before it did that last time you did a home, the gantry ends up skewed in Y across X.

The disappointment is when I had a bad cable connection, I noticed that a Z-axis motion was 'occurring', logging the steps on the screen, but never moving the axis!  With SERVOS with true position encoders, they're still using 'open-loop' positioning!  They send the pulses, and just pray they move the axis!  DAMN!

Lloyd

[Bob La Londe]

Since you didn't build the machine it may be harder, but many "dead reckoning" servo controllers have a programmable following error, and you can set it to throw a fault output which can be wired into the machine controller in the event of a fault.  Even closed loop steppers have fault outputs than can be wired into machine controller inputs. 

The Crappodrive... I mean Geckodrive servos aren't so good, but they do have "some fault" output capability.  Whale drives (Dugong) have programmable following error.  I am sure most higher end servo drivers have some programmability and fault output pins. 

Now whether or not the machine builder uses them is another story. 

[lloydsp]

These servo amplifiers (Mits', and good ones) actually have real-time position feedback outputs.  They're just not monitoring them.  They just ASS U ME that they're working always.

I can't modify their software; it's not even theirs.  But I can report it to the software vendor and see if there's a way to accept the feedback from those drives, and use it to 'justify' what the screen is showing.

Lloyd

[Bob La Londe]

I am 100% aware of what you already said.  Now you are sounding like somebody else on the forum.  LOL. 

I am saying even on servo drivers that are not closed loop to the machine control computer you can have and use fault outputs to at least stop the machine. 

You could even do some limited machine control on a fault by using a low trigger relay and cutting into the stop circuit.  There is always a work around.  I don't like work arounds, but it was this very forum and the software for which it is dedicated that beat the work around thought process into me against my will.  Well, that and growing up having to fix it myself or walk. 

[lloydsp]

Bob,
Please don't insult me for stating the conditions, even if I restated some things.  You're not the only person reading this thread, and some may not know the background.

I also don't think you understand, at all.  The software DOES react to faults -- by stopping DEAD.

There is no method to actually 'handle' a fault in their software.  It stops and won't allow you to continue.  It requires a full hardware reset of the amplifier box (remote to the computer). There's no 'graceful' fault recovery method, either.  You can restart the code at approximately the point before it failed, but only after having performed another homing operation, and it always causes you to re-cut features that were already cut, with the tooling off by several thousandths due to the inaccurate homing.

Only the authors of the software can fix those things.  But they can, if they will.

Lloyd

[Bob La Londe]

 So... output a new partial code file.  I do it all the time. 

As to the rest I quit.  You are the expert in what you want. 

[Bob La Londe]

I thought about this for a little while. I have to ask, why aren't you rezeroing when you know that's a problem? 

Even an ordinary edge finder can get you within a couple of thousands.  I know you may be running a high-speed spindle, but even my 24000 RPM spindles are capable of having enough torque at 1500 RPM for an edge finder.  If the configuration of the machine won't allow you to use it slow enough to use an edge finder. You can still use an electronic probe even if the machine doesn't have a probing input.  You can connect the probe to an led & a power supply. 

I know I know you. You want the machine manufacturer and the software programmer to fix all your problems, but in the meantime, you still have to make parts.

[lloydsp]

Bob,
I never said what I did or did not do during one of those episodes.  I've had just a little experience doing this stuff for four decades.  All I said was what would happen if I restarted a job after homing.

OF COURSE I re-zero, or the rest of my job would be ruined.  I even peck a precision index mark in the work at the beginning of a job to check my own re-zeroing against a known location.

Addressing your earlier remonstrations:  If I had to re-edit the g-code every time this happened, I'd spend half my days just doing that.  I know how to edit code -- I do it all the time.  The cleaner solution, until my new limit switches come in, is to simply not home the machine during the day.  It stays dead-to-nuts on, until I do a homing operation.

Lloyd

[Bob La Londe]

So sorry.  My attempts to help  are obviously grossly inadequate and unappreciated.  I apologize.  I'll try not to do it again.  How does that ignore list work?  Wouldn't want to accidently point out the obvious by accident and offend your four decades of experience, since by implication I couldn't possibly have anything to add you don't already know, and know it better. 

[lloydsp]

C'mon, Bob.  This isn't you.  I don't know what's going on, but it's not the Bob we know.

L

[Bob La Londe]

Just responding to that bug up your a** Shouting out your mouth. 

[lloydsp]

I apologize, if I said anything to personally offend you.
L

[dh42]

Hey guys

Please be cool, a beer, a joint and you'll feel better!

++
David

[Bubba]

Hey guys

Please be cool, a beer, a joint and you'll feel better!

++
David

+1