WEBVTT

00:00.000 --> 00:06.154
Electromagnetic noise measurement with spectrum analyzer

00:06.154 --> 00:20.733
Horizontal abscissa: the frequencies, here from 0 to 100 kHz, logarithmic scale

00:20.733 --> 00:25.603
On the ordinate (vertical), the amplitude in dB (decibel)

00:25.603 --> 00:28.992
Currently the average noise is -90 dB

00:28.992 --> 00:33.469
Presence of a noise peak here, we will change the frequency sweep for 1 kHz

00:33.469 --> 00:44.826
15 seconds after ...

00:44.826 --> 00:49.414
The peak is at 50 Hz, the frequency of the distributed network! As well as all harmonics (multiple)

00:49.414 --> 00:55.779
Reduced to 100 kHz, faster display because of shorter period.

00:55.779 --> 01:11.404
Switching on the drive causes significant pollution!

01:11.404 --> 01:15.236
Noise peaks reach -10 dB!

01:15.236 --> 01:20.100
The peak frequency starts at the frequency MLI (PWM) of chopping of the converter (here 10 kHz) as well as harmonics (multiples)

01:20.100 --> 01:31.099
The average noise level has also increased from -90 dB to -60 dB or 30 dB more (1000 times more radiated power!)

01:31.099 --> 01:48.822
We will try to shield the measuring plate

01:48.822 --> 01:55.655
Before bonding to ground: no improvement!

01:55.655 --> 02:04.528
When referenced to ground: the tip is 20 dB lower: radiated power gain of 100!

02:04.528 --> 02:13.267
We gained 10 dB on average noise, a power ratio of 10 ...