[RFI] Coupling methods for RFI emission testing on SPS

[Larry Benko]

Jeff,

Your method sounds a bit too "Rube Goldberg" for me.  Assuming the RF 
interference is via common mode current radiating from external wires 
(reasonable assumption) the FCC limits for a part B device are 56dBuV 
quasi-peak/46dBuV average for freq. from 0.5 to 5 MHz and 60dBuV 
quasi-peak/50dBuV average for freq. from 5 to 30 MHz (9kHz msmt. 
bandwidth).  These measurements are made into a 50 ohm / 50uH LISN which 
is easy to build but not all wiring is easily measured with a LISN.  
Unless you have a spectrum anaylzer with an EMI option you will be stuck 
with a 10KHz average measurement which will need the limits raised by 
0.4dB to account for the BW increase.

Doing a bit of math, 50dBuV average corresponds to 316uV avg. and 6.32uA 
average common mode current.

If you don't want to go the LISN route take a current probe and measure 
the common mode current along all wires leaving the device you are 
measuring.  Without a screen room the wires will pick up a lot of 
external radiators (AM broadcast) but with a little care they can be 
ignored.  I use a 0dB-ohm shielded current probe which means it has 50 
turns but anything over 10 turns is ok.  However you will need to 
compensate for the current increase.

For me over the years I have found that if the _10KHz average common 
mode current is 1uA or less it is nearly impossible to hear_ it unless 
you are only a few feet from the antenna. 1uA corresponds to 15.6dB 
better than the part B limit which is damn quiet.  For a 0dB-ohm current 
probe 1uA is -107dBm so you might need a preamp on your spectrum 
analyzer.  If you have a 10:1 current probe 1uA will be -93dBm.

If you can't reduce the common mode current to <= 1uA then you need to 
decide if what you have is good enough.

73,

Larry, W0QE


On 8/8/2016 10:46 PM, Jeff AC0C wrote:
> There are a few applications where I have needed to use a SPS and 
> worried about RFI.  My solution has been to compare the powered and 
> unpowered spectrum on a SA, looking for areas where the floor is 
> raised.  Based on what I see on the SA, filtering is added to whack 
> the noise.  I also look (again with a wire loop moved over the 
> circuitry) what the radiated level seems to be.
>
> I generally couple the SA using a few turns of wire looped around the 
> input wires, and later the output wires of the SA.  If I can't see a 
> lift in the noise floor, I will try a few more turns, or move the 
> pickup wire along the wire a bit.
>
> This test setup lets me judge the before/after effect of various 
> combinations of filtering to determine which one actually serves best 
> as the silver bullet.
>
> It seems to work fine but what I don't know is "how much is too much" 
> with respect to the raised noise floor?
>
> I worry this coupling method is frequency dependent and in an absolute 
> sense, I really don't know if what I am seeing is a bit insensitive 
> compared to "real" antennas.  Or if it is complete overkill (meaning 
> the test setup is too sensitive).
>
> Generally I will let a bit of a change in the nose floor (say 10 dB or 
> so) slide, with the thought that a signal 10 dB above the floor sensed 
> with a SA probe separated only a couple mm from the radiating 
> conductor will be in the mud given real-life separate distances 
> between the SPS location and shack antennas of a couple hundred feet, 
> against a background of atmospheric noise.  But I don't have any test 
> data to say this is a solid assumption, or that 30 dB raise would be 
> the acceptable limit, or that 1 dB is...  The test criteria is too 
> subjective.
>
> Hoping one of you EMI/RFI pros out there may have some comments on how 
> the big boys do it.  Or maybe some other suggestions based on your 
> experience.
>
> Generally I try to use linear supplies whenever I can.  But in a 
> growing number of cases, SPS cannot be eliminated so I'm left to clean 
> up the noise profile as best as I can and with the significant cost of 
> ferrites I want to make sure I'm being efficient in what the target 
> noise level needs to be.  A guy can spend a LOT of money on ferrite 
> and never know if it's adequate, inadequate or what.
>
> 73/jeff/ac0c
> www.ac0c.com
> alpha-charlie-zero-charlie
>
> _______________________________________________
> RFI mailing list
> RFI at contesting.com
> http://lists.contesting.com/mailman/listinfo/rfi
>