Hello Guest it is May 14, 2021, 06:39:25 PM

Author Topic: Goto work 0  (Read 1257 times)

0 Members and 1 Guest are viewing this topic.

Offline mark4

*
  •  167 167
    • View Profile
Goto work 0
« on: February 05, 2018, 02:15:17 PM »
Hello i have a puzzle. i am running a Ethernet Smooth Stepper with a MB2 and three AC servo drives on a small mill. i have another mill that is almost exactly the same it doesnt have this problem. history 1st i tried to install the latest mach4 release and the latest build on the smooth stepper. this worked until i tried to jog then it wouldnt jog above .01" if you tried i made like it was moving the dro's changed but the machine didnt move. note my setup has a relay set to run the oiler whenever and axis is moved the relay clicked on like it was moving. so after allot of back and fourth i decided to use mach4 install 3233 as this is on the other machine and works fine the problem with the mpg jogging went away. think i am home free think again.
now the mpg jog works in all increments. i can home to the index pulse. but when i try to goto zero the machine moves a small increment and bumps to an abrupt stop, if i hit the button again it will move another small increment and abrupt stop, if i hit the button the right amount of times with abrupt stops sometimes it will continue all the way to zero. note i am not doing z yet so just x and y. i can see the leds on the MB2 and havent seen anything wrong they all light up as expected no flicker except on the index pulse and that is expected. i have my own lua scripts and also was using a pokeys board but for my latest tests i went back to windows 7 mach4 build 3233 ESS build 216 no pokeys and installed a complete virgin mach4 with no script modifications. homing went well goto zero same problem. also i get no errors mach thinks it is doing the right thing. i turned on logging and am attaching file of homing, trying to goto zero, and on the last press actually going all the way to zero
if you have a work around know what i have done wrong or can just define the problem please help
i am desperate
thank you
mark
Re: Goto work 0
« Reply #1 on: February 05, 2018, 03:41:57 PM »
Hi Mark,
that is a very interesting problem...I don't know whats going on but a thought occurred to me that might describe the symptoms and a possible test.

You noted that with a late build of Mach4 that your servos would not move if you attempted to jog with an increment greater than 0.01.
Now you have found that when homing is active that the servos will move in discrete bursts yet in both circumstances the LEDs of the controller
the DROs and the oiler relay would all indicate that Mach had called for motion and passed that to the motion controller but for whatever reason
either the motion controller or servos failed to make the move.

I'm guessing that the motion controller is dumping a whole bunch of pulses to the servo drives and the drives are exceeding the 'following error' window
and stopping. I would expect Mach/ESS to issue pulses at a rate within the acceleration limits of the servo/drive but maybe is not under certain circumstances?

There are a few questions which could resolve whether this is in fact whats happening or not. I don't know what servos you have  or how sophisticated they're
wired. It is common for servos to have an alarm output which could trigger under a number of circumstances including following error. It is also common
that such an alarm would trigger an Estop. Is your machine wired this way? I have a servo as a spindle and haven't been bothered to wire its alarm output
back to my controller, if you have done the same then you may not realize that the servos are faulting 'following error'.

If they could be faulting without you being alerted to the fact then I would suggest either monitoring the alarm output, it might be as simple as an LED and a
resistor. Another way is to widen the following error window, if that is possible with your servos. My Allen Bradley servo allows me to program the following
error window, its default value is 20 encoder counts. That is to say that if the servo lags its input (from the controller) by more than 20 counts (8000 count/rev)
then it will fault. I could reduce that following error but then could expect it to fault more regularly, alternately I could program it to 4000 counts say and it would
not fault until it lagged 180 degrees from where it should be!

What I propose for the purposes of testing that you increase the following error window to some wide value to see whether the fault you've described changes.
What to do about it if it proves to be the case is a different matter.

Craig
My wife left with my best friend...
     and I miss him!

Offline mark4

*
  •  167 167
    • View Profile
Re: Goto work 0
« Reply #2 on: February 05, 2018, 08:08:18 PM »
thanks for the reply and i see what you mean by trouble shooting. fortunately i found the problem the latest ESS plug in is at fault the machines i believed to be almost exactly the same are except for one thing the plug in the one with the problem has the latest ESS plug in the one that is working had v193 i didnt catch this when i was writing paper work. so i tried 193 and it works fine goto zero no problem. for this machine i am going to stick with 193 and move up when i can verify them to work. the important thing is i have a working controller and a functional machine. i use ac servo drives with quadrature signal and home to the index pulse so that takes care of accuracy. my fault output is hooked to an input although that might not have been programmed at the time. the drives themselves have led's to tell you fault. that was one of the problems i had troubleshooting the board led's were working exactly as they should.luck
mark
Re: Goto work 0
« Reply #3 on: February 05, 2018, 08:29:47 PM »
Hi,
kool, you have a solution, but do you have an answer?

May I suggest putting a LED or LEDs on screen and attach the input signals which are associated with your alarm outputs. It may give you a valuable insight
into the behavior of your machine but also the nature of the firmware fault.

If you are using wx4.set then the machine diagnostics page has six LEDs for inputs INPUT#0 to INPUT#5 so would be a fairly simple exercise.

What servos are you using? I have been spoiled by Allen Bradley, they have software, which I had to buy much to my disgust, that lives on my PC which I can
program all the parameters and features of the servo, simulate how it runs and when I'm happy download the whole shooting match to the EPROM memory
on the drive. I can unplug the drive from the PC and use it as a standalone drive. I thought it was a bit over the top when I first looked at it and trying to
justify spending a couple of hundred bucks on it but I'm glad I did.

Of course Allen Bradley are far from the only good servos out there but given I've got this software and how easy they are to use I'll be buying more of them,
they can be reasonbly priced second hand off EBay.

Amongst the things you can program is an analogue output that reflects the current following error, fascinating watching the error loop close when you offer up
a motion profile.

Craig
My wife left with my best friend...
     and I miss him!

Offline mark4

*
  •  167 167
    • View Profile
Re: Goto work 0
« Reply #4 on: February 05, 2018, 08:51:09 PM »
hi
i use DMM ac servo drives they are very easy to program and the only thing i had to buy to program was a cable. there tech support has been very knowledgeable and helpful. the can configure step/dir, quadrature, analog and i think something else. and they have configuration software. if thats not enough they will output a full encoder output. i use them in 1/2 of the machines i rebuild.
Re: Goto work 0
« Reply #5 on: February 05, 2018, 09:22:30 PM »
Hi,
I looked (actually drooled) at DMM when I buying a servo for a spindle. In the end the 1.8kW Allen Bradley servo and drive turned up on Trade-me, a New Zealand
auction site and I bought pretty well, about $700USD including shipping. Shipping to New Zealand for something like a servo and drive can be over $100USD in
itself. So despite being second hand I'm very happy with it. It also explains why I had to buy the tuning software, had I bought new the software would be included.

Its the first 'modern' servo I've really had anything to do with and man they are great!. As you say all the flexibility of pulse/analogue modes, torque/velocity/position
modes, all the monitoring functions and the list goes on.
What really impresses me is that this thing can maintain angular position right through to full revs and rated and more torque with a following error of only
(20 / 8000)*360=0.9 degree and a zero position error of (4 / 8000)*360= 0.18 degrees. Way, way, WAY overkill for a spindle but really impressed by the technology.
This particular servo is ten years old and the design is older still. If its as capable as this then truly modern servos are surely better....

Craig
My wife left with my best friend...
     and I miss him!

Offline mark4

*
  •  167 167
    • View Profile
Re: Goto work 0
« Reply #6 on: February 07, 2018, 09:37:40 PM »
hello i like the new with autotune features thats huge i have been doing this for long enough that tuning pid loop you need a weji board and spiritual assistance. and even then i am never sure i am right or way off. i know there are people out there that can do the math and are great pid tuners but i am not one of them lol. autotune takes five minutes the drives do it themselves all i do is click and i can go onto other problems such as goto work zero. i do want to give a shoutout to ESS teck support they are great to work with. luck
mark
Re: Goto work 0
« Reply #7 on: February 08, 2018, 01:58:40 AM »
Hi Mark,
I agree, the software I use has all the Allen Bradley servos in a database with electrical, magnetic, thermal and physical parameters. It was not necessary for me to
tune anything. I could edit the tuning but I doubt I could do better than the manufacturer.

Funnily enough I quite enjoy fiddling with PID loops, a leftover from my University days when we were instructed/ commanded to fiddle with them...and tested on the
results! Truth is that you can usually find much better things to do with your time though.

I second your positive comment about Warp9 Tech support. I'm guessing that Andy is pretty much the chief cook and bottle washer which explains why development
is so slow/erratic. There are just too many support calls for him to concentrate on development. Nonetheless I am impressed by the support.

Craig
My wife left with my best friend...
     and I miss him!