Machsupport Forum
Mach Discussion => General Mach Discussion => Topic started by: Davek0974 on March 06, 2014, 10:23:06 AM
-
Ok so I have been eagerly awaiting the delivery of my new plasma torch, while waiting I have been extensively testing and debugging my new DIY cnc table and it was all going very nicely.
The torch came this morning so i moved the table into my test area, connected up the ground clamp and powered the system up.
All went ok, placed a bit of test metal on the table, moved to take a test fire and triggered a quick pierce signal, no cut just a quick pierce.
That went ok so i went to move the torch back to start position for a cut test, the torch would not move so i pressed the estop and restarted, z axis started moving and partial x axis but not Y.
Killed the plasma supply and reloaded Mach.
Now it was jumping about randomly as i turned the MPG knob on the pendant and kept throwing "e-stop requested" errors.
Thinking it might be my cheapo crap pendant, i unplugged it, turned off "run macro-pump" and disabled support for modbus/mpg.
Then i reloaded Mach and tried again, now it sort of moves but the DRO's are moving like 20mm for every mm moved.
Pressing home all, would bring the axes to home but then back off at 1% speed or very slow at least.
It would also still throw the e-stop requested errors.
Now, having spent a lot of effort hardening the table against RF, I cant believe I have fried it with a two second pierce????
No other PC's in the area showed any sign of problem and there was no radio interference.
This was just a quick pierce at half power.
Any suggestions at all?
I'm desperate.
-
Have you got the correct steps per unit set?
Hood
-
Yes, the table has been running a felt tipped pen for a good few weeks, perfectly at various speeds and has not so much as dropped a step!
I even tested it before firing the torch and all was well.
-
You said you had reinstalled Mach and after that a 1mm move showed 20mm, that suggests the steps per unit are incorrect.
Hood
-
My profound apologies, I meant restarted NOT reloaded mach, I have not reinstalled it yet.
Everything was fine and calibrated very nicely.
-
Ok in that case then it sounds like maybe you have damaged something. Hard thing to check if all axes are acting weird as you can not swap out drives or pins on BOB. If you have a scope you could check the output pulses from the port to see if they correspond to the pulses set in motor tuning.
Hood
-
Are these pulses in the kHz range ?
I have a slow scope, not digital, I think its DC to 5mHz possibly.
just been messing about and noticed that there is an LED on the BOB, this is on normally, but after running the axes up and down manually a bit, the LED flashed off then on and Mach threw the e-stop requested error, after that the torch relay will not trigger and the axes start messing about.
I really can't believe I've fried a BOB or the PC surely?
If that is the case then there is little hope of success as I have done almost everything I can to shield this stuff, I have not driven a ground rod yet as its in the factory at work and i doubt the landlord would appreciate that :)
I could understand it if I hadn't connected the plasma ground lead or something else stupid but I even checked it with a meter first!
-
The frequency will depend on the commanded feedrate and the steps per unit set.
For example if you had the steps per unit set to 1000 and you commanded a feedrate of 100mm/min then the frequency of that move would be (1000 x 100)/60 =1666.666667 Hz or 1.6KHz
The reason I say that you may have damaged something is that you say it started just as you fired the plasma for the first time. If you have everything connected as it should be then it is unlikely you have damaged hardware but if you have a ground connected wrong or something like that then it is a possibility.
If you have not fried something then I do not know what the issue could be seeing as it was working fine until you fired the plasma.
You could shut Mach down and run the driver test to see what it looks like. Also have a look on the Diagnostics page in Mach and see if the frequency is steady and close to the kernel you have set, also see what the Time in Int is, again on Diagnostics page.
Hood
-
Diagnostics show the freq as around 28,048 and time in int as +5 to +6 varying.
I need to ring the suppliers to find out what that LED actually means, when that goes off is when the problems arise, I thought it was a power LED, in that case it might be the PSU?
The mach driver test still passes as "Excellent", nice and smooth on the graph.
-
Spoke to the BOB suppliers, seems the BOB is fried 100%. (:
The LED is the output from the onboard regulator so its turning itself off randomly.
They also said it was wired wrongly, I had the BOB supply fed from the Z-axis output of the motor PSU, apparently that is bad and must be a separate regulated supply for the BOB only.
Anyone know anything about this?
-
Here's the current state of wiring...
1 - I have an 11A smps connected to the three motor drives,
2 - a smaller 24v smps connected to the BOB,
3 - The BOB is a cnc4you.co.uk HG07 unit
4 - Another small 24v psu is running the relays for e-stop and limits / torch trigger.
5 - I have used screened cable for all motor leads, this is connected at one end only inside the cabinet to a star ground.
6 - I have connected the Z-axis to the X-axis carriage and then back to the star point.
7 - I have connected the Y-axis gantry to the star point.
8 - The case of the PC is connected to the star point.
9 - All PC leads are fitted with ferrite chokes.
10 - The mains is now filtered at the plug.
11 - The plasma ground lead is clamped to the cutting bed.
12 - The frame of the table (and therefore the cutting bed) is not connected directly to my star point, I am guessing this is catered for by having the cabinet bolted direct?
The whole lot is mounted in a steel cabinet bolted to the machine frame.
Any obvious issues with that?
-
I have just tested it again with a different 24v PSU connected to the BOB, this one is a PLC supply so pretty good quality.
It worked ok with the torch off, ran a cut program 10 times in succession and all worked ok.
Connected up the plasma and did a quick test pierce and down it goes again, no movement on the motors.
This now confuses me as it seems the BOB is OK still but i have serious issues elsewhere, trouble is i dont where or how to test them????
Following a reboot, normal control is returned.
The PC showed no signs of a crash, no lock-up or BSOD etc.
Not sure which way to go now but clearly i have issues.
-
Its random, tried a reboot again, this time no control is regained.
As soon as i triggered the torch, the power led on the bob went out, the y-axis jumps 5mm and that's it, no more action.
Its also holding the motors locked despite being told to disable the drives, this point indicates the fried BOB but I cant risk another BOB until I know what went wrong or it can get expensive quickly.
-
12 - The frame of the table (and therefore the cutting bed) is not connected directly to my star point, I am guessing this is catered for by having the cabinet bolted direct?
Dave - As you're discovering plasma is a noisy beast. I'd strongly suggest you ground your table direct to ground - if anything this is THE rule for plasma systems. I'd certainly not use your cabinet. Grounding is never as straightforward as perhaps we'd like but I'd be thinking your passing all that noise straight into your cabinet and onto the electronics ground plane.
Just to be clear I'm talking about TABLE GROUND. Your work lead is NOT plasma ground. I know a lot of folks call the clamp plasma ground - but it isn't - it's actually the current source +ve lead.
Also what type of plasma start does your cutter employ?
-
i had a chat with the plasma supplier, told them of my woes and the only suggestion they had was to drive an earth rod, connect the terminal on the cutter to that and then connect the CNC table likewise. Unfortunately I couldn't drill a hole through the factory floor so I just connected the terminal to the table frame.
I tried that and managed three pierce tests this time but then locked up again.
Then I disconnected the earth from star point to PC case and it worked! Passed repeated pierce tests, then I run a simple cut-line test of six 70mm lines at reducing speeds, table was still responding so I ran another test pass followed by a more detailed test and it still responded afterwards!
here's the results...
(http://www.davekearley.co.uk/LinkPics/CNC13.jpg)
numbers are mm/min at 30A.
i like the result at 4000mm/min, it's very fine and about 0.75mm kerf, this is 1.5mm sheet so very thin.
i gather the point where it closes up at the end is due to deceleration?
anyway, it was getting late so I quickly dialled in 4000mm to my next test in sheetcam and compiled the g-code.
(http://www.davekearley.co.uk/LinkPics/CNC14.jpg)
It looked ok, showed up some signs of serious axis wobble but ok. However on turning it over it appears not to be cut very well. ...
(http://www.davekearley.co.uk/LinkPics/CNC15.jpg)
Now, I'm not too worried as I did rush the test and the changes to sheetcam so I may have bodged it up. The main thing for this thread is that it's working and surviving extended cuts.
many thanks to all for the helpful suggestions, much appreciated.
-
Also what type of plasma start does your cutter employ?
This is a 20-50A HF pilot arc torch
When I take the table home, I will definitely have a ground rod, it's easy there.
-
Not sure what you mean by "the terminal". I'm guessing you mean the work-clamp (what you were calling plasma ground). Is that correct? If so it should be connected to your table. Then you should have an earth strap from your table straight to a good ground - (or earth rod) but not to a ground point on your cabinet.
It's hard to tell from your pictures but they look like you don't have a good enough connection to your work with the work-clamp. If you have it connected to your table, try connecting it direct to your work. 30A should cut 1.5mm steel at around 5 to 6000mm/min no problem.
-
The terminal I was referring to is a 14mm bolt on the back of the plasma cutter, this is now connected to the table frame, it will go to ground when I get it all home.
I think the big fix was disconnecting the PC case, another idea bought about by internet dredging, definitely not a good idea it seems though.
I will try direct clamping the lead, but I have a sneaking suspicion that because I was rushing these tests due to being near closing time, I may have inadvertently run them at 20A instead, I will redo them more carefully this week.
At least it's running now, that's the main thing.
As for grounding, I did notice that once the first cut had passed over a bed slat it was lightly welded down so surely that would ensure a good connection to the plasma return lead from then onwards?
-
I think the big fix was disconnecting the PC case, another idea bought about by internet dredging, definitely not a good idea it seems though.
I'd certainly not use your cabinet. Grounding is never as straightforward as perhaps we'd like but I'd be thinking your passing all that noise straight into your cabinet and onto the electronics ground plane.
::)
-
Well, all is still not well.
I had some more play time today, tried some 30A test cuts and it's messing about, stopping cut motion midway, throwing e-stops, losing co-ordination etc.
It will be even worse if i plug the PC into the same extension as the system, I have to run two extension leads back to the wall socket just to get a partial success.
If it was 100% because i have no ground rod then i would stop testing until I take the machine home where I can fit one, but I'm not totally convinced it is that.
The system has a filtered plug fitted and the pc has an in-line filter fitted.
Not sure where to focus my efforts now????
-
I like differential step/dir signals and also 24v I/O, much more noise immune.
Hood
-
Hmm, that sounds like a complete different control system?
-
Just had a chat with the suppliers and they feel the generator is not switching to full power i.e. I am cutting (or trying to) with pilot arc power which is about 20A on this model
They are sending me a new (tested good) hand-held torch for test purposes and while waiting they have advised me to try upping the air pressure to 75psi just for a test.
But basically at 30A and 1.5mm sheet, it should practically vaporise the damn thing at the slow speeds I have been trying!!
Will know more tomorrow.
-
Hmm, that sounds like a complete different control system?
Ideally yes but you could also do a lot of it yourself if you need to. I used to have to do that on my machines as all that was available was 5v i/o boards and single ended step/dir. I made up line drivers for the step/dir but my drives accepted differential, if yours dont then you would also need to make up receivers as well. 24V I/O can be done different ways but it can get messy with relays etc. The controllers from CS-Lab do it all for you in a small neat package but likely you would only want to consider going that route if you cant get the better of your problems.
Hood
-
I think this BOB is toast really, its the original one and the LED for the relay is off when initialised but after it warms up, the LED is half-on. It still switches the relay as required but the LED goes from full to half power not full to off.
When i started testing I had the BOB powered from the stepper drive PSU, this I now know was bad idea so I am hoping it was this that did the toasting.
I have a new BOB sitting here, just not sure to try it or not.
I now have fitted a small PSU purely for the BOB and its on a filtered supply too, cant do much more than that.
My other issue turned out to be the machine torch, its a duff one, the new test hand-torch they sent me today cuts really well.
-
Well, i'm getting baffled again, the new torch arrived this morning and results seem to be better but something is still wrong.
In this picture...
http://www.landyzone.co.uk/lz/members/davek0974-albums-problems-picture15119-photo-1.jpg (http://www.landyzone.co.uk/lz/members/davek0974-albums-problems-picture15119-photo-1.jpg)
The long cut is made at 20A @ 4000mm/min, its a nice fine cut about 0.5mm kerf with zero dross front and back.
After that success I thought I'd load up a little test file i've been messing around with and try that after editing the sheetcam settings to suit the new parameters...
http://www.landyzone.co.uk/lz/members/davek0974-albums-problems-picture15120-photo-2.jpg (http://www.landyzone.co.uk/lz/members/davek0974-albums-problems-picture15120-photo-2.jpg)
Now, this was the second run as the first one at 4000mm/min was worse so i dropped to 3750mm/min but all i got was a pile of dross and partial cuts.
Both tests were pierced at 4mm (mfg recommendation) and cut at 1.75mm (mfg says 1.5 - 2.0mm)
Now this tells me that something is badly wrong but my lack of experience does not tell me what.
Any pointers here???
Also, I have set a pierce delay of 0.1 sec in sheetcam but mach3 seems to stay for at least a second or so.???
Does Mach3 have a default minimum pierce delay??
-
Well it seems it is lack of current, this was run at 30A...
(http://www.landyzone.co.uk/lz/members/davek0974-albums-problems-picture15121-photo.jpg)
Its pretty good I think, I have axis wobble to fix but the cut is clean there is no dross front or back apart from a small amount where it slowed down for the corners.
Still 100% better than the first go.
Although I am happier now, I still fail to see why 20A will cut an excellent straight line but produce so much dross and crap on a simple shape at similar speeds??
Oh, and I still need to fix the pierce delay issue, I have 0.1s set but its holding so long that the torch starts popping before moving due to lack of metal in front of it.