Kit's Machine

[Kitwn]

Kit, your last response does concern me - if you remove the "electromechanical galvanic isolation module" from the Z-limit input to the bob, and simply connect this input to the local 0V, is it still noisy? If so you're down to BoB/Cable/PC.

I haven't actually removed the EGIM but it does strap the input to ground. I have made some interesting measurements though. Some screen grabs from the oscilloscope are very telling...

trying to accurately the voltage on the earthed case of the machine is a bit difficult since earth is meant to be the refference voltage. this is measured on the case of the LinuxCNC PC with the scope earth floating. 10mS/div. 4V/div.


Slightly less scary, this is one of the parallel port pins pins referenced to the PC case. 1mS/div. 0.2V/div. None of the motors were running so there's not controll data on the pin.


Most scary is another floating measurement of the metal frame of the CNC router itself. 1mS/div. 4V/div.


So my theory that that the intereference is going out on the mains cable of the VFD is probably wrong, it's more likely coming from the spindle itself.
The spindle is connected using screened 3-core cable with the screen connected to the mains earth inside the VFD and to the body of the spindle and hence to the frame of the machine.
So why does the mains filter on the VFD do some good? Dunno. Why does plugging the VFD into another mains socket stop all trips occuring and allow me to use the machine as normal? Dunno.

I have improved the earth bonding of the Z axis and gantry to the base frame but the above measurements are after that was done. For the present I'm just going to get on and use the machine as it is for it's intended purpose and hope for the best but, as I've said before, if this hobby turns commercial I shall want to upgrade the controller to something less flakey.

Kit

[JAZZCNC]

Throw the scope and the bloody BOB in the bin and buy an AXBB-E you tight bugger, all your problems solved.!. . . Simplizzzz

[JAZZCNC]

The spindle is connected using screened 3-core cable with the screen connected to the mains earth inside the VFD and to the body of the spindle and hence to the frame of the machine.

If you have earthed your frame then surely you have created an earth loop by connecting the VFD cable shield at the spindle end.? I don't earth the spindle end and never have noise problems.

[phill05]

We had a problem at one stage that sounds close to what you have, we had the screen connected at both ends, to remedy took off screen at spindle end now left with 3 +Earth, ( 4-core CY cable) connected spindle to VFD, at the VFD end the screen was stripped back and finished at star point in cabinet, VFD was fitted in same cabinet on the same power line in.
Been running for four years no problems.

Phill

[Kitwn]

Throw the scope and the bloody BOB in the bin and buy an AXBB-E you tight bugger, all your problems solved.!. . . Simplizzzz

It's just possible you may be right.

[Kitwn]

If you have earthed your frame then surely you have created an earth loop by connecting the VFD cable shield at the spindle end.? I don't earth the spindle end and never have noise problems.

Connecting the spindle cable braid to the spindle body has a lot to do with safety, there being plenty of volts available to spoil your day, though making sure the machine body is well earthed should solve that. Interesting point made earlier about grounding emissive cable screens at both ends and receptive ones only at one end. The principle of reciprocity may not support this however.

I have already got UCCNC running in demo mode on my laptop and have confirmed that delivery to Australia will not take too long so I can see his happening soon. How does UCCNC compare to LinuxCNC? How easy is it in practice to set up auto-squaring for a two-motor axis?

[Kitwn]

We had a problem at one stage that sounds close to what you have, we had the screen connected at both ends, to remedy took off screen at spindle end now left with 3 +Earth, ( 4-core CY cable) connected spindle to VFD, at the VFD end the screen was stripped back and finished at star point in cabinet, VFD was fitted in same cabinet on the same power line in.
Been running for four years no problems.

Phill

Thanks Phill. I might try this, though at the moment I just want to start making some bits with a working machine, which is what I now seem to have. I expect that I will soon have to bite the bullet and upgrade to an AXXB-E and UCCNC to make things more reliable. It will also be intereststing to see what difference this makes to the maximum speed of the machine and the surface finish as ONE OF THE USEFUL THINGS I'VE LEARNED FROM MY OSCILLOSCOPE (subtle hint there you notice ) is that the pulse timing from LinuxCNC and it's parallel port are far from perfect. Long ago I set up a test and measured the step pulse interval at 125uS with frequent (approx 1 in 8) pulses at either 100 or 150uS (always +- 25uS). That's a pretty dramatic error and cannot be good for best performance.
I have now got the mechanical side of the machine working quite well, as far as my limited experience can tell, so an upgrade to the control side is worthwhile. I will keep yours and Dean's comments in mind when that rebuild gets underway.

Kit