Oscope EMI

Discussion in 'General Electronics Chat' started by blazin, Feb 28, 2014.

  1. blazin

    Thread Starter New Member

    Jun 4, 2012
    Need help understanding how to make an oscilloscope measurement.

    I have a DC signal that in normal state is held at logic '0'. During ESD/EMI I see a series of transient pulses swinging from about -3 to 15VDC. I am currently probing with active probe between the signal and reference ground. Probe is 1GHz BW active type. Leads are ~3" (both signal and gnd).

    The problem is a logic '1' will cause a power down or low power event.

    How do I perform this measurement reliably and confirm that the EMI that is being generated near the circuit is also really causing this logic '0' signalling to trigger a power down event through the series of spikes?

    Logic '1' is defined as 3.3VDC, minimum time is not defined, signal is labeled as asynchronous, I assume there is some associated response time, but there are no published specifications.

    Would measuring the ground (null measurement) and the signal then doing a CH1 minus CH2 pseudo-differential provide a more accurate representation? True differential measurement with diff amp/probe between ref gnd and signal? Shorter leads? AC Couple? Some combination of everything?

    Thanks for the help!
  2. MrChips


    Oct 2, 2009
    You have to provide us with some very important details:

    1) What is the make and model of your oscilloscope?

    2) What is the device/instrument you are testing?

    3) What do you mean by "active probe"?

    4) What are your 3" leads?

    5) Can you provide a photo of your setup, scope, probes, instrument?

    6) Can you provide a photo of your scope screen with all VAR knobs in the CAL position?
  3. tindel

    Well-Known Member

    Sep 16, 2012

    If you have access to a differential probe for this measurement this is the type of measurement you want to use it on. Be sure to use your third pin on your probe attached to your common ground to filter out common mode noise. Keep your probe lengths as short as possible. 3" is generally way too long for this type of measurement.

    I would then remove all other test hardware from the circuit that is reasonable to remove (i.e. multimeters, unnecessary loads, any long wires, basically anything that won't be there in the final configuration.

    Finally, I would use an EMI 'sniffer' to trigger the scope. This method is described in one of the appendices of the following app note.
    http://www.linear.com/docs/4159. There's also a good section in this app note on proper noise measurement which is very applicable to your application. Generally speaking - keep your probe lengths as short as possible.

    This test method will help you trigger on the radiated emissions on the EMI source without actually connecting to the EMI source. Which gives you a very strong advantage. If you trigger the scope on a EMI source in this way and you do see ripple or a logic shift on your reset line then you have proven that the EMI source is what is causing the reset.

    After that - there is still more work to do - as you've only proven that the EMI source is causing the problem - not were the problem is actually at. If the two signals are on the same pwb then look at the pwb and see if the reset line is running in parallel with a line of the EMI source. This is often the source of crosstalk which can cause a logic shift in your signal. If it's coming over the air... simply shield your reset line or your EMI source and you're done.