[sdiy] Poor man's EMC testing?

[Magnus Danielson]

Joe Grisso wrote:
> Hey gang,
> 
>      While this is somewhat off-topic, I was wondering if anyone had
> insight into economical test methods for performing EMC analysis
> (FCC/CISPR) before going to certification? Most of the solutions I've
> found involve purchasing a $5K+ Spectrum Analyzer and equally
> expensive antenna for measurement. Are there any other methods you
> guys know about?

Proper spectrum analysers require odd bandwidths and odd detectors.
You can do without a proper room if you go outdoors, which may work 
fairly well unless you are in urban areas full of transmitters.

The spectrum analyser could be an older beast but at least covering up 
to 800 MHz is recommended. The normal bandwidth steps and detectors work 
well enough unless you want real qualification (if you are 3 dB away 
from limit, you are fine). A sine-spike will not really change with 
bandwidth and detector anyway, but for noisy stuff it may make a 
meaningful difference. HP and Tektronix made good specs.

For antennas you should recall that qualification is made for broad-band 
measurement at 3 or 10 m from DUT. 30 MHz to 1 GHz. The antenna gain, 
amplifier gain etc is all calibrated for. Making qualitative measures is 
a beast. Having E and H field probes and a small amplifier together with 
lots of time collecting experience could give you good hunches of things 
to fix. A very nice thing is the near-field probes that cancel out 
fields in the distance, such that only local field-gradients gets any 
attention is a great too for pin-pointing where a noise-source is.

In the end, what is good Signal Integrity usually goes hand in hand with 
good EMC. Recall that sharp edges can be just as dangerous as high 
frequence.

Building simple E and H-field probes yourself is simple. The near-field 
probe is tricky but the capable and handy can pull it off.

Find the spec, that will remain the (physically) largest investment in 
all this.

Cheers,
Magnus