Home
JAQForum Ver 24.01
Log In or Join  
Active Topics
Local Time 03:53 26 Nov 2024 Privacy Policy
Jump to

Notice. New forum software under development. It's going to miss a few functions and look a bit ugly for a while, but I'm working on it full time now as the old forum was too unstable. Couple days, all good. If you notice any issues, please contact me.

Forum Index : Electronics : EMI radiated by my inverter - how to reduce

Author Message
poida

Guru

Joined: 02/02/2017
Location: Australia
Posts: 1418
Posted: 12:19am 21 Jan 2024
Copy link to clipboard 
Print this post

Lately I've been playing with my toy spectrum analyser which
is just a STM32F4xx based microcontroller.
It delivers very good signal to noise performance,
nearly as good as the theoretical from it's 12 bit ADC.
When testing and developing the code at work, I get great results.
But when I run it at home, with the inverter powering the house,
it's gone bad. The EMI emitted from the inverter takes away all the fun.

results after the changes (discussed below) can be seen here



The White trace is before mods
Orange is house running on street power, sort of the best I could get
Purple is after the mods.
The Green trace is the same as the White trace but with 10dbV/division

Vertical scale is 20 dbV, and the peak is a 1V p/p sine wave at 2kHz
20dbV is the same as a factor or 10 change in voltage.
So the noise is about 60db down from the signal.
60db down is 1/1000th of the voltage.

The scan width is 0 to 25kHz over those 6 divisions

So it appears I "halved" the EMI since the purple trace is about in the middle
of the other two. It might only be 1/3 less due to the log scale but it's
worthwhile to get.

The inverter is a Madness power board with 4 x 4 FETs, a 70uH choke and a
3kW Toroid. The choke is located on one of the primary winding ouputs
of the power board.
That gave me the White trace.

I change the choke into 2 x 30uH chokes and put one on each of the primary winding
outputs. This gave me the Purple trace.

I then put a mains filter (as found in a 3kW Aerosharp. I used the biggest one.)
but it did very little to the noise level. If you do this, your results
might vary and in fact see it helps significantly.

All the inverter components are inside a metal enclosure. I have two clip on
ferrite chokes on the 240V output cable and these are about 1 foot outside the box.

The above traces are not time averaged. Each is the result on one scan only.

Below are traces where I averaged 8 scans, to smooth over the noise variation
and give a clear indication of the average noise level.

I chose different scan settings
0 to 2,500Hz
1024 frequency bins
and average 8 scans

This is with the inverter off:



and with it on using the modified design



This is not bad at all. I want to compare this with the clean power/room
I enjoy at work, to see what is possible here at home.

The board looks like this:



The code permits:
sample sizes from 256 to 4096
time averaging of the trace 8 and 32 scans
sample rates from  200Hz to 500kHz, resulting is FFT widths of 100 to 250kHz
sample averaging from 1 to 64 samples per recorded data sample (v. good to kill noise)
Linear, 10dbV/division and 20dbV/division vertical scale
A range of windows to choose from, I usually use "Flat Top"
see here to learn what window functions do

According to FFT theory, you can expect a better or worse noise floor
when you change the sample size and or sample averaging.
Also, the ADC introduces more or less noise as you change the ADC frequency
and sample and hold timing.
I can see the effects of this, when playing and learning about signal theory
in the clean power of my work office.
I found this project very enjoyable and educational.

here is the code, if anyone is interested.
You change the various parameters of the FFT and sampling via a serial port
eg "s5" means sample at 5000Hz

stm32f4_cmsis_fft_serial_config.zip
Edited 2024-01-21 10:20 by poida
wronger than a phone book full of wrong phone numbers
 
KeepIS

Guru

Joined: 13/10/2014
Location: Australia
Posts: 1679
Posted: 01:37am 21 Jan 2024
Copy link to clipboard 
Print this post

Seriously that is really nice:

Another nice project to play with, I'm not sure I'll live long enough to try them all.

A while back I carried out some testing with a HIOKI AC line analyzer, I found very little difference in Inverter AC harmonic content with various AC filtering methods.

It was quite good even with a basic filter. I think the only advantage the Mains had was the lower impedance or current sink capability from crap loads causing a DC component into the AC line.
It's all too hard.
Mike.
 
Print this page


To reply to this topic, you need to log in.

© JAQ Software 2024