When Does One Draw The Line On EMI?

So where's the problem?

Fix what's broken and filter what's remaining.

The only time I draw a line on repairs is when the cost of repairs for a old unit exceeds the purchase and install cost of an equivalent or superior new unit.

If only I had that option.....

This place requires 2 signatures for me to buy a box of paper clips. I don't know why and I really don't care because our people want to know in advance how much money they are going to save first. That is how a lot of good ideas die.

It sounds like ignorant been counters run the show and not competent individuals who know what it cost to keep a business running.

I've only been in that position a few times and the last time my answer to their 'how much will it cost?' question came down to questioning them on how much it will cost to replace me and train in someone else who can then start from scratch and eventually get to the point of knowing what I already know now.

That answer right there has way too many variables that all end in too large numbers for any been counter to ever not agree to your demands.

I would like to say my counter argument worked but to be honest that ended up being the second job in my life I ever walked out the door on and left 'those who knew more than me about my value to the company' hanging on the end of what quickly became a very expensive rope.

When you're a small company and you only have overworked 3 technicians one leaving is a pretty hard hit. Especially when you have to negotiate with the other two or watch your business sink like someone just installed a screen door in the bottom of their rowboat.

Just use Flexseal.

That was the first thing that came to my mind. Damn late night ads! BTW, I tried it on my cracked bird bath, and it's not nearly as good as they advertise.

There goes my last illusion . . .

200 volts peak to peak on a +/- 12 volt supply????

That is the noise level on all three lines. Anything connected sees the ±12 VDC but it would be like driving your card over big potholes. It still works, but it is probably not good for it.

200 volts p-p relative to what? If you measured 3 DC lines relative to an AC line, you would see something like that. If you are measuring with respect to ground, are you sure the ground is really ground?

Sounds like you either need to service the old equipment to try and fix the problem, replace the old equipment or add additional filtering to your test setup to prevent damage.

Can't you turn the old equipment causing the problem off during testing, or do the testing somewhere else?

It seems that the DC variable speed drives (5 or 6 of them per machine) take their power straight from the 3 phase full wave bridge. Filters? Ain't no stinking filters, or capacitors, etc. The manufacturer wanted $500,000 for about a full set of spare modules or about $6000 each. No guarantees that the modules made with new parts would work. They didn't think they had any stock of 40 year old components that might have a chance of working. I recommended a major overhaul which was met with one of those "deer in the headlights" stare.

Some modules had undocumented field modifications. Modules built from the prints only ran about $200 each. It looks like each module will need major modification to withstand the noise.

Then its probably going to be cheaper to just protect and isolate any new circuits (within reason), going so far as to use older more robust technologies (like transformer-supplied linear regulators.

If you cannot fix the problem or protect modern electronics effectively perhaps an old-school resilient power supply solution is best.

I used a differential strip chart recorder. Any connection to any point on this machine sends the differential amplifier off the scale. I confirmed an approach with a manufacturer of Common Mode Chokes. They come in SMT and you probably have one your cell phone. I'm also going to tame the noise at the source by adding a 3phase line reactor (choke) on the AC line that feeds the unfiltered full wave bridge from which the pulse width modulated DC motor drive takes it's power.

The emf was so powerful that my Bluetooth connection (phone app) to my hearing aids would not work until I was 10 meters outside the room in which the equipment was installed. I wonder if nearby aircraft have noticed the problem?

Do you know the bandwidth on your strip chart recorder? If you are trying to sleuth down the source of the EMI problem, the strip chart recorder may not have enough bandwidth to identify the true problem.

The reason your strip chart recorder goes off the scale is likely due to ground loop currents coupling into your chart recorder. I understand it goes off the scale. That very well may be the test set up causing the crazy high readings.

Common-mode chokes can be very effective. The best way to determine if common-mode chokes will be effective is to use an oscilloscope with at least 50 MHz bandwidth and a wide-band clamp-on current probe like the one below, and see what frequency range your common-mode currents are. You clamp these over the whole bundle of cable(s) to see what the common-mode currents really are.

Of course the vendors of the common-mode chokes are going to tell you their stuff is great. Until you can take valid measurements, you are just shooting in the dark. Take enough shots and you will get lucky, eventually. You might hit your target on the first shot. I would have no way of knowing this from here. Common-mode chokes will have different core materials and construction to address the frequency range of the common-mode noise. Choose the wrong core type and you will get very little improvement.

PWM motor drives are typically IGBT-based inverters and you will probably get ringing at device turn-off in the range of one to ten megahertz depending on the PCB layout and package parasitics. RC snubbers are typically used across the 'switches' to suppress some of the ringing. Usually a film cap and a series resistor.

You really need to take valid measurements, or you're just guessing. Addressing EMI issues is a major part of my job. Below is spectrum analyzer results for conducted emissions test CE102 of MIL-STD-461F. And you can see from the results below this system is not completely compliant (at this time).

Cheers!

have you tried using EMI suppression cores on the wiring? they are cheap and may be all you need.

example