Hallo Prashant,

you stated, that problems occur or become bigger, when your machine is doing what's intended to do, namely handling the wires. But not only EMC can be the source of problems, but also increased load of stepper motors, when wires are to be handled...

I have seen many times, that situations are becoming worse, BECAUSE of using optos. Why? Because the use of optos often results in a very poor signal routing referring to high frequency interference. An example:

If you do not use optos, then both ends of signal wires are somehow connected to ground. May be the screen is grounded (directly or via some capacitance) or at least you will find some filtering capacitors which help to suppress high frequency interference. But if you use optos, then you will mostly do everything to prevent any connection to ground. So, you will not furtherly use this filtering, shunting high frequency interference to ground. BUT: High frequency interference is still there and you must suppress it somehow. And: Because you omit the shunting to ground at both ends, high frequency interference will even increase!!

Signal routing using optos are very very often very very poor in terms of suppressing high frequency interference. What you can see often is a nasty nasty system of wires and cables, not having any high frequency referencing to ground at both ends. And due to unavoidable inductivity and stray capacitance your whole wiring system is in heavy resonance introducing of all kinds of noise and shit right into signals, although optos are used and designer thinks that everything is ok.

Many people using optos, do NOT use shielded cables. Why should they? They use optos!?! Right? No, absolutely wrong!!! Only the combintaion of using shielded cables, well referenced for high frequency interference to ground at both ends (!!) AND using optos to prevent introduce of noise coupling into the signal can help in demanding situations!!! By the way, I prefer use of symmetrical signal routing, then you always have this ground referencing by design. And you are strongly 'remembered' to use some sort of filtering and shielding.

I recommend you to make intensive meassuring. Use a battery powered scope if you can, because such scope would not introduce noise via mains supply, making everything worse. Check every point of your application.

I assume you will see noise peaks everywhere, only differing in the height of pulses. This is a result of poor signal routing using unshielded cables and neglecting proper high frequency grounding. While doing some fix, you will find, that a capacitor here or there will remove a bit of these spikes, but at different points you will have an increase. Only when having a defined ground referencing for EACH cable you will have a satisfying situation. If you use shielded cables with high frequency grounding at both ends you will end up with a Faraday cage completely surrounding the whole electronics, which is the best panacea against ESD...

Keep in mind, that each interface must have some signal filtering, even if only low frequency signals are transmitted. This is the only way to prevent high frequency noise from spreading arround. It's always wise to introduce a limiting of bandwidth of transmission down to relevant signal bandwidth. This is the only way to prevent that high frequency noise finds a way to enter your application!

Another trap is the use of 'fully isolated' DC/DC converters or 'fully isolated' mains transformers or 'fully isolated' mounted transistors on cooling fins. Forget these phrases. Fully isolation does not exist in real life for high frequency. An example: Very small mains transformer shows a interwinding capacitance between input and output winding of more than 150pF. 150pF gives an impedance at 1MHz of 1 / 2 / pi / f / C = 1.06kOhm. At 10MHz, which is in the range of clock frequency of most micros, 150pF gives an impedance of 106Ohm. Do you think, that your mains transformer provides a 'full isolation' at this frequency? Well, it's just the opposite!!
Assume now, that there IS high frequency interference at mains voltage. How do you prevent it from entering your system? You have a big storage capacitor? Ok, this will not suppress much of this interfence, because of unavoidable equivalent series inductivity. And, think about frequencies above 10MHz. You have a voltage regulator filtering out high frequency interference? Ok, ever had a look into a datasheet? Voltage regulator will not show any relevent suppression of high frequency interference!!! So, high frequency interference has just entered your system!! Although using 'fully isolating' mains transformer, although using big storage capacitor and although using voltage regulator providing 'so much' input voltage ripply rejection.

And I can make the list longer and longer. Finally you will not have any high frequency rejection although using these fanstatic tools all showing 'fully isolation'. No, no. It's on you to prevent high frequency noise from entering your system: By shielding, filtering and grounding!

By the way: ESD events contains frequencies above 1GHz...

Above I stated that inductivity of long cable cause resonances and radiation. Assume you connect 10m long cable to your mains transformer. With a rule of thumb 1m cable has about 1µH inductivity. So, our 10m cable has an impedance at 10MHz impedance of 2 x pi x f x L = 628Ohm. Oops! This is more than impedance of interwinding capacitance of mains transformer!
So, we have a totally different situation now. Mains transformer is NOT fully isolating (high impedance between input and output winding) and cable is NOT showing zero impedance, but vice versa!

Where is the resonance?
150pF and 10µH form a series resonance at 1 / 2 / pi / SQRT (L x C) = 4.1MHz. So, if we do NOT ground the cable at both ends, we will result with a series resonance at 4MHz...

Where is the radiation?
10m cable acts like an antenna. If cable length is in the range of lambda/2 then massive radiation occurs. For 4MHz lambda/2 = c / f / 2 = 37.5m. Well this is a bit away from 10m. But if we assume a smaller stray capacitance of perhaps 10pF, like it can easily be found with DC/DC converters or poorly mounted optos, then resonance occurs at 15.9MHz and lambda/2 is 9.4m. Well, in this case heavy radiation would occur!!!

You can turn it how you want: Using cables in combination with 'fully isolated' parts showing relevant stray capacitance WILL introduce resonances and radiation, unless cables are high frequency referenced to ground AT BOTH ENDS!! This high frequency referencing CAN be a grounding of shield (directly or via capacitance) or CAN be the introduce of filtering. But in any case high frequency interference must be shunted to ground via a low impedant path (mostly capacitance), and this at both ends!


Kai