0
\$\begingroup\$

I'm planning to buy [this flat ferrite choke][1] for ribbon cable to suppress suspected EMI on a data acquisition's multiplexer.

I have no experience on signal integrity. Bu the system setup is basically like this(On the right side the system is single-ended earth grounded):

Instead of adding chokes to each BNC I thought it would be easier to use just one flat core for the ribbon cable since it conveys the signals to ADC insdie the PC.

Which would be a better option? To apply chokes to each BNC cables or to use one flat ferrite core around the ribbon cable? I also couldnt find any passive chokes for 50 Ohm coaxial BNC cables in the market until now, neither info how to install them.

I'm having random glitches sometimes. By random I mean 2 or 3 times in 30 minutes acquisition. And this is not every time. By glitch I mean at some point suddenly all the channels are affected. So it is almost impossible to observe this by scope because I don't know when the noise will hit.

\$\endgroup\$
12
  • \$\begingroup\$ It's not that easy. There's a lot of different types of noises, and which one is relevant for your signal can't be said just by looking at your system from this afar. \$\endgroup\$ Commented Nov 26, 2016 at 16:58
  • \$\begingroup\$ @MarcusMüller Assume the noise is EMI interference. I want to check of the effect of using ferrites since it is impossible to see this noise by scope since it is extremely random no one knows when it will hit. Is there a typical solution for EMI related noise? I dont know how to connect a choke to a BNC and how big should the toroid be. Couldnt find any info \$\endgroup\$
    – floppy380
    Commented Nov 26, 2016 at 17:02
  • \$\begingroup\$ "EMI interference" tells nothing about how the noise relates to your signal, what kind of amplitude and frequency distribution it follows, how or where it comes into your system and what it does to your sampling … sorry, can't help you on this level of information :( \$\endgroup\$ Commented Nov 26, 2016 at 17:09
  • \$\begingroup\$ @MarcusMüller I added more info for you and others. Please see edit. Maybe it helps to clarify my question. \$\endgroup\$
    – floppy380
    Commented Nov 26, 2016 at 17:14
  • \$\begingroup\$ definitely helps! \$\endgroup\$ Commented Nov 26, 2016 at 17:18

1 Answer 1

2
\$\begingroup\$

All channels simultaneously get a value of 0 – that's most definitely not EMI, or any other type of what I'd call "noise".

That's a problem with your sampling hardware. Now, this might be anything – from a driver bug to a hardware fault, to samples not being fetched before some buffer runs over to some problems with a on-board power supply, or, and that's my current guess: a faulty connector somewhere that randomly makes the "BNC box" power off.

Generally, if you really just need 6 channels, your bandwidths are this low and you don't calibrated voltage readings, but just relative values: Try replacing the PCI DAQ card with a simple multichannel sound card.

A single 16bit ADC that gets multiplexed across a few 4kS/s channels? That's really not that great – you can actually build that with something as cheap as an arduino. Still, it's probably more time- and cost-efficient to not build something like that yourself, but try with widely available DACs – of which soundcards are the cheapest ones, and perform surprisingly well.

\$\endgroup\$
2
  • \$\begingroup\$ BNC box is not power supplied. Do you mean a connection problem between the ribbon and the BNC box? The thing is it is not possible to change the daq board for some reason, all the software written for this and libraries ect. But is there a way to log this data in parallel(using a data logger or arduino something easy to read at least one channel and log the data for 30 minutes) and prove if it is a software bug or hardware problem? thanks \$\endgroup\$
    – floppy380
    Commented Nov 26, 2016 at 17:43
  • 1
    \$\begingroup\$ how should I know? I didn't write your software! \$\endgroup\$ Commented Nov 26, 2016 at 17:44

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.