You mention using shielded cables. Did you ground only one end of the cable shield ? I have seen installations with both ends of the cable shield connected to ground, which makes ground loops a concern.
Having built and serviced CNC plasma cutting machines, proper shielding of cables does wonders. The comercial products we used all had shielding built into the harness to keep the noise out. And the shields were always grounded at 1 end only., and in some cases where the design was older and more sensitive they actually had 2 shields, insulated from each other and each one grounded at opposite ends of the cable.
Also - effort should be made to seperate the motor and encoder cables. Often just having them shielded and a few inches apart is sufficient.
While it doesn't explicitly state it in the article, it appears that the encoders in question have single ended outputs. It appears that the extenders convert the signals to differential mode, which is much more noise tolerant. Shielding is a good idea, but with single ended signals I doubt it would have been enough if there is serious high frequency switching noise all around.
Just wondered if you considered using a 4-20-mA current loop in place of the voltage signals. Current loops are not susceptible to electrical noise that can scramble voltage signals. Avago Technologies has 8-pin DIP optically coupled 20-mA transmitters and receivers that simplify current-loop circuits and provide isolation, too. Texas Instruments has at least one 20-mA transmitter.
I have driven a teletypewriter (circa 1978) with a 20-mA current loop over almost a quarter of a mile of wire. Granted, at 110 baud. But in short-distance wires, you can get a high data rate.
Single ended encoders will work just fine. Plasma machines with 5V single ended encoders and proper shielding and machine grounding will work just fine. Because of the arc involved in Plasma cutting poor machine grounding can induce pulses in the various electrical circuits and cause damage. And grounding is at times a black art... Machines that run fine in one location with minimal attention to grounding practice won't function in another area without special grounding practice.
If you connect a shield at both ends it becomes a ground wire and can cause even more problems if and when there are gound currents. The purpose of a shield is to block and blead off any external electromagnetic garbage, not to act as a possible conductor. There should be a dedicated ground wire in the motor harness.
Another important thing to consider is GROUND LOOPS. If you have multiple ground paths on a machine you have the potential for different currents to flow in the various ground wires/paths. Then your grounds are no longer at the same reference and wierd and wonderful things can happen like solid state components turning on and off seemingly at random.
I should have been more specific - I was talking about the power cable for a motor. Quoting from one drive manufacturer's installation manual:
The Shield terminal provides a grounding point for the motor cable shield. It must be connected to an earth ground by a separate continuous lead. The motor cable shield should be connected to this terminal on the drive (drive end) and the motor frame (motor end). Use a shield terminating or EMI clamp to connect shield to this terminal.
Motor power cables are the only time I would do this; I would be curious if this might have eliminated the positioning motor and the spindle motor radiated noise.
Not knowing the particular construction of the motor you refer to - I would assume that the shield connection at the motor end is just continuing the shield connection through to parts on the motor not directly connected to electrical ground. This would keep the noise (pwm is really noisy..) from the motor cables from getting to the encoder/resolver wires in a parallel cable.
For industrial control applications, or even a simple assembly line, that machine can go almost 24/7 without a break. But what happens when the task is a little more complex? That’s where the “smart” machine would come in. The smart machine is one that has some simple (or complex in some cases) processing capability to be able to adapt to changing conditions. Such machines are suited for a host of applications, including automotive, aerospace, defense, medical, computers and electronics, telecommunications, consumer goods, and so on. This radio show will show what’s possible with smart machines, and what tradeoffs need to be made to implement such a solution.