[I_V_A_R]

So I have had my machine running with a Gecko g540 + ESS + Mach4 settup for a while now.
The last ting I did to ensure stability, was to get a CAT6 isolated ethernet cable, with isolated adaptors, I have an isolated cable for the spindel signal, and I made a closed aluminum compartment for the ESS. I think this improved stability quite a bit, but you should be able to trust a CNC controller more then that.

Still doing a large v-carving project, I had the 0 origin jumping, and I think it had to do with the screen saver.

Now I set the screen saver and sleep mode to Never. I have done so in only one way in windows 7.

It stil feels like I can never fully trust my system to complete a large job without errors, but hopefully this is it, and the machine will run in stable maner from now.


Is there anything else I should think of to ensure stability?

Best regards, and thanx for any additional tips.

[RaderSidetrack]

So I have had my machine running with a Gecko g540 + ESS + Mach4 settup for a while now.

With an ESS (motion controller) in the chain, I would be surprised if the stability issue you experience is related to the Windows. As I understand it, the ESS is handling control of the machine at that point.

Still doing a large v-carving project, I had the 0 origin jumping

Are you sure this is not a version of 'mechanical'? Perhaps trying to move the router/spindle too quickly through the material, resulting in the stepper missing steps?

[harryn]

Just throwing out the general idea that RF noise is not completely easy to eliminate. Make sure that all of your system is tied together to a good ground using braided cable vs solid wires.

In the US, most electrical grounds are solid wire to a solid copper stake hammered into the ground. If you use this as an RF ground for your system, it might not be sufficient. ( as in likely not )

We are in a high solar activity period, and this is more or less RF noise.

Probably other things are also factors, that is just what struck in my mind.

[joeavaerage]

Hi,
I've been using Mach4 and an Ethernet SmoothStepper for 8.5 years....and I've never had the problem you describe.

The only recurring and pesky noise issue I've ever had was electrical noise from the VFD. I now have a line reactor and a two stage EMI filter on the primary of the VFD, that was
seven years ago, and never a problem since.

Craig

[joeavaerage]

Hi,
most hobbyists do not understand electrical noise at all well, and as a consequence they do things like opto-isolated this that and the next thing, imagining somehow that
it magically confers noise immunity. More often than not they make it worse rather than better.

First thing to understand is that there are two ways electrical noise can get into a sensivtive signal line.

1) Electromagnetic radiation. As the name suggests one part of a machine is generating noise in the for of electric and magnetic fields ant that couples into another circuit
but WITHOUT any electrical connection between them. The principles of electromagnetic induction apply. Most understand this but overlook some basic ideas. More anon.

2) Electrically conducted noise. Lets say the VFD, being the usual and prime suspect, is generating electrical noise. In particular switching noise and harmonic noise on the AC supply.
VFD's are notorious for that. That harmonic noise can couple into the PC power supply, the stepper power supply and the motion controller power supply. Thus all these supplies are
'polluted' by harmonics from the VFD and are gaining ingress into sensitive circuit by conduction via the common AC supply voltage. This is remarkably common but very poorly understood and widely
overlooked.

The example of me solving my noise problem is a perfect example. I put a line reactor and a two stage EMI filter on the incoming AC side of the VFD. That prevented the VFD (largely) from
'polluting' my power supply and therefore all the other DC power supplies were (largely) free of noise. I also have another two stage EMI filter protecting the AC supply to my PC and motion control
supplies. Thus any (conducted) noise from the VFD has to 'get out' past the line reactor and an EMI filter, and then 'get in' via another EMI filter.

It is not always east to tell if a noise problem is 'conducted' or 'radiated' in nature. A useful and easy test is to run a tool path but with the spindle off. If the noise problem abates then at least you know
what's causing it.

When it comes time to try to stop noise there are two basic approaches
1) Using screened cables, introducing filters and inductors, opto-isolation and the list goes on. This is the first approach most people take...but should in fact be the last plan of attack.
2) Reducing or stopping the device from emanating the noise in the first place. The line reactor I fitted very VERY effectively suppresses harmonic distortion, and thereby reduces the amount
of 'noise pollution' the VFD generates.

Another very often overlooked and misunderstood principle is 'effective noise impedance' This applies to radiated noise. Radiated noise is an energy transfer process. It would be a fairly
extreme noise environment that would cause 1 milli Watt of electrical energy to be 'transmitted' from one circuit to another (without electrical connection between them).
If that energy is incident on a very high impedance circuit, then it will produce a large voltage disturbance, which could affect the signal quality of the circuit.

As a numerical example: lets say we have a high impedance signal circuit, say 50kOhm, to an amplifier controlling a stepper signal. P=V²/R

V= (0.001 x 50,000)^0.5
=7.07V

So the noise voltage would be 7V....which is very high. If this were a 5V signal then the noise would swamp it....no good at all.
Lets say you reduced the impedance down to 1kOhm:
V= (0.001 x 1000)^0.5
=1V

Now for the same signal, ie 5V the noise voltage is 1/5th of the signal and so the signal dominates and is unlikely to cause a fault.
You can of course go too far the other way and reduce the impedance to zero, but then you have opened the door for current noise. In short there is an optimum impedance that
means the signal circuit is LEAST affected by noise. What you are trying to do here is make the signal circuit less prone to noise, without yet going to shielding and all that other stuff.

As an example I made my own breakout board so it had all the features I wanted and nothing else. In particular the many (31) inputs are all 24V, and they all have an impedance of 4.7kOhm,
by my design. This means that It would take 12V noise impulse to cause an unintended signal transition and that would require an electromagnetic pulse of 30mW!!!!
None of my input cables a screened or anything special but unless I lie them next to the (unscreened) VFD cable I get no noise faults. By choosing and designing for a known noise
impedance I have made my breakout board (largely) immune.

Thereafter you use shielding, earthing, filtering etc.

If you are having noise issues the don't start shielding and all that, start by asking 'is the noise conducted or radiated'. If its radiated have you appropriate impedance levels
to reduce the susceptiblity to noise?. THEN consider shieling etc. This last lot is 'trying to shut the gate after the horse has already bolted'.

Craig