• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Dealing with EM noise

Offler

Senior Member
Joined
May 27, 2021
Messages
414
Likes
399
I am looking for source of few EM noise I am measuring. Following FFT graphs are average of 20 scans using 1048576 points.

There is some noise at 50Hz, which is from power grid in EU (220v, 50Hz) and it got much better when I purchased better power cord, so the FFT graph is now much clearer with overall lower noise floor. This is LineIn of a PC soundcard

Interesting is noise at 80Hz and 240-250Hz. I really dont know about any device working on so low frequencies. Any idea?
test56.jpg
test57.jpg
 

Phos

Member
Joined
May 27, 2021
Messages
60
Likes
39
You're measuring using a soundcard? Are you sure it isn't an artifact created by the soundcard?
 

solderdude

Grand Contributor
Joined
Jul 21, 2018
Messages
15,891
Likes
35,912
Location
The Neitherlands
Do you have comparable plots that show differences between power cables ?
 
Last edited:
OP
O

Offler

Senior Member
Joined
May 27, 2021
Messages
414
Likes
399
You're measuring using a soundcard? Are you sure it isn't an artifact created by the soundcard?
Yes and I really don't know.


I used USB card in a different computer (small netbook), however its noise floor was around -110dB and I could not see the spikes whatsoever. I dont know of signal source at such low frequencies in my PC. Everything in PC works in MHz spectrum, except for PS/2 port which works at 16.7KHz.


Do you have plots that show differences between power cables ?
I have one, in lower res. This is before I replaced power cord:
phones.jpg

50Hz was higher by about 10dB.

This is the new power cord:
https://www.se.com/ww/en/product/PM8-FR/apc-essential-surgearrest-8-outlets-230v-france/

Grounding is much better now, but those two-three frequencies in lower spectrum showed up.
 

nerdstrike

Active Member
Joined
Mar 1, 2021
Messages
257
Likes
309
Location
Cambs, UK
One thing to remember about PCs. They are full of fans i.e.. Rotating magnets.

Might not be strong enough to register, but they are there and potentially close to wherever you prop your sound card up. They may also work at different speeds depending on temperature, but I'd expect hundreds of Hz
 

solderdude

Grand Contributor
Joined
Jul 21, 2018
Messages
15,891
Likes
35,912
Location
The Neitherlands
Alas that plot is not comparable.

Can you measure the exact same setup twice with only the mains cable being different ?

Frequencies well above Nyquist can 'fold back' or create interference products in the audible band.
 

Phos

Member
Joined
May 27, 2021
Messages
60
Likes
39
Hmmmmmmmm, fans.... let's say at 300 RPM that's 5 RPS, but the magnets are probably switching multiple times per revolution, but how many polarities would such a simple motor have? Most that I've seen inside have four coils, so I'm going with 2, which would be 10 Hz, but a stepwise signal so it would look like a square wave in the Fourier transform. Since they're more interested in smoothness than torque, it might instead be a four pole motor, which would be 20 Hz but we're no longer analogous to an audio signal. I don't think it's fans.

And yeah, about the power cable, what would the power cable be changing to suppress 50 Hz in the output? Filter 50 Hz on the power input and then the device runs on the harmonics? Hmmm now that has me wondering if you could intentionally tank the power factor of audio equipment to eliminate or reduce mains frequency, suppress the voltage peaks and increase the current...
 

solderdude

Grand Contributor
Joined
Jul 21, 2018
Messages
15,891
Likes
35,912
Location
The Neitherlands

Attachments

  • cord.png
    cord.png
    247.3 KB · Views: 77
Last edited:
OP
O

Offler

Senior Member
Joined
May 27, 2021
Messages
414
Likes
399
Ahh.. a power strip.. that's not what I would call a power cord.
Also this device contains surge arrestors which basically do nothing as long as there are no large spikes on the web.
One thing is certain, those surge arrestors (we don't know how they are connected) also have a capacitance which may be helpful in some circumstances.

powercord:
c13-europe-pc-power-cord-1-8m
Only standard cords, 0,75mm squared thick copper, french to IEC, each 1,5 meter long. I have but one better (1mm square copper).
 
OP
O

Offler

Senior Member
Joined
May 27, 2021
Messages
414
Likes
399
Ahh.. a power strip.. that's not what I would call a power cord.
Also this device contains surge arrestors which basically do nothing as long as there are no large spikes on the web.
One thing is certain, those surge arrestors (we don't know how they are connected) also have a capacitance which may be helpful in some circumstances.
c13-europe-pc-power-cord-1-8m
So I went through the box full of cables. I identified 2 brands of cables and bunch of unbranded.

My amplifier and computer PSU were both delivered with cords by Linetek.

I found two other branded cables (one was from another computer PSU) by brand I-Sheng, while one of them had 1mm Cu cables and i connected both PC and AMP using those cables.
test100.jpg


test101.jpg


There is still plenty of noise, but the spikes are lower.
 

solderdude

Grand Contributor
Joined
Jul 21, 2018
Messages
15,891
Likes
35,912
Location
The Neitherlands
The 'problem' here is that those kinds of spikes can enter a measurement in lots of ways. Given the level they are well below any audible levels.
With the info at hand I think it is not really possible to determine the exact reasons for this.

Are you experiencing audible issues which prompted you to measure or were you measuring and noticed changes ?
 
OP
O

Offler

Senior Member
Joined
May 27, 2021
Messages
414
Likes
399
The 'problem' here is that those kinds of spikes can enter a measurement in lots of ways. Given the level they are well below any audible levels.
With the info at hand I think it is not really possible to determine the exact reasons for this.

Are you experiencing audible issues which prompted you to measure or were you measuring and noticed changes ?
I was trying to found scientific review of NAD d3020 v2 amplifier, and when I havent found any, I tried to do the measurements myself.

On the way i learned stuff about Windows CAudioLimiter and when I applied workarounds, sound was better - it went from -60dB THD to -80dB. After few tweaks I have found sweetspot of the amplifier volume where THD (without N) went to -90dB. That was audible difference.

So the noise at 1-7KHz was a problem since it affects harmonics in FFT graph, while low-frequency noise at 50, 80 and 240Hz was reducing dynamic range.

Latest graph I posted shows THD at -95dB and noise-free range of 99dB. Its not audible anymore

Since the low-frequency noise does not change when increasing volume on amplifier (connected via toslink) I just assume it was just picked by the cables via induction. Probably it is just an issue with LineIN, or noise on power cords. High frequencies - even after reduction are however still a bit of a problem, even when definitely not audible.
 
OP
O

Offler

Senior Member
Joined
May 27, 2021
Messages
414
Likes
399
There were two effects since the PC power cord got replaced. Noise floor of LineIn went lower, and one rare post-boot crash did not ocurred again.

2 meter, 0.75mm diameter cable was either not enough (850watt PSU, not sure how much power surge it causes when PC gets powered on) or the original cable is simply busted.

Not an audio issue, but it appears the noise indicated bad cable.
 

solderdude

Grand Contributor
Joined
Jul 21, 2018
Messages
15,891
Likes
35,912
Location
The Neitherlands
Crappy cords can cause issues.
Remember, the power cord does not carry a nice sinusoidal current but very short and high peaks at the top and bottom of the mains voltage.
Those peaks can be very high when peak power is drawn.
 

zermak

Senior Member
Joined
Jun 2, 2019
Messages
372
Likes
251
Location
Italy
250Hz is the standard USB frequency rate (and for gaming mouse it can be 1000Hz, you can set it at 250Hz, 500Hz or 1000Hz usually, now there are some 4000Hz too). I am pretty sure if you move your cursor/mouse while measuring, 250Hz frequency will spike.
4000Hz should be ethernet connection working frequency if I recall from some other thread.
Other spikes stuff can be coils whine, usually from discrete high end video cards.
 

solderdude

Grand Contributor
Joined
Jul 21, 2018
Messages
15,891
Likes
35,912
Location
The Neitherlands
Coil whine is an acoustical thing. It won't show up in electrical measurements.
 
OP
O

Offler

Senior Member
Joined
May 27, 2021
Messages
414
Likes
399
250Hz is the standard USB frequency rate (and for gaming mouse it can be 1000Hz, you can set it at 250Hz, 500Hz or 1000Hz usually, now there are some 4000Hz too). I am pretty sure if you move your cursor/mouse while measuring, 250Hz frequency will spike.
4000Hz should be ethernet connection working frequency if I recall from some other thread.
Other spikes stuff can be coils whine, usually from discrete high end video cards.
There is a difference between USB frequency and mouse polling rate.

In short, mouse sends data up to set polling rate. Since it high-priority ISR call, its better not to send data all the time, but only when there is an update.

If you know LatencyMon (people on Audio forum should), you might disconnect all USB devices, keep only mouse and check WDF0100.sys in Drivers tab. When you run the test and move cursor, there will be much more ISR calls as when you dont.

Coil whine is an acoustical thing. It won't show up in electrical measurements.
Its very individual.

In my specific case, the coil whine got to the analog outputs and LineIn of my soundcard (onboard ALC1220) and to the USB as well.

I took notebook (on battery), installed AudioTester, and ran stress test for GPU. When coil whine was audible, it definitely leaked to the AUX input on my d3020v2, but it did not got through OPT input (optical SPDIF). The coilwhine was not just measurable on Analog, but also audible.

It does not leak through the power cord or power strip (which should filter everything except 50Hz).
 

solderdude

Grand Contributor
Joined
Jul 21, 2018
Messages
15,891
Likes
35,912
Location
The Neitherlands
The power strip will not filter anything unless there is also a filter inside. If there is a filter inside it will start to filter well above the audible range.
If it would pass through 50Hz to heaters and older lamp bulbs but the PC would not power on as the current peaks (which are much higher than 50Hz in frequency but repetition in a 100Hz rate) would not pass through the filter.

In my specific case, the coil whine got to the analog outputs and LineIn of my soundcard (onboard ALC1220) and to the USB as well.

If the same frequencies that were audible coming from the device itself are also visible in the analog output signal there is some really poor designing being done or you have a ground loop where groundloop currents through the screen of the interlink, or poor PCB design in either the source or receiving end, were creating unwanted signals into the amplifier's path.
Could also be we have a different definition of coil whine.
To be coil whine is caused by coil wires vibrating caused by currents flowing through it and causing air to vibrate (makes it audible)
The fact that it can whine is caused by the frequency changing because of feedback loops and varying loads.
 
Last edited:
OP
O

Offler

Senior Member
Joined
May 27, 2021
Messages
414
Likes
399
The power strip will not filter anything unless there is also a filter inside. If there is a filter inside it will start to filter well above the audible range.
If it would pass through 50Hz to heaters and older lamp bulbs but the PC would not power on as the current peaks (which are much higher than 50Hz in frequency but repetition in a 100Hz rate) would not pass through the filter.
Depends on what manufacturer of power strip means by the following statement:

Surge Protection and Filtering
Response time 1 ns common mode
Common mode rejection 60 dB
Peak current 48 kA common mode
60 kA per phase normal mode
Surge energy rate 2690 J

I made the purchase by the Surge energy rate as the primary purpose I expected is to protect the connected devices, and values above 2000Joules were recommended to me by an electrical engineer.

Powerstrip is rated for 240v and 10A, while I have to note that when I turn on the switch on it, and PC powers on at the same time, 'Overload' led blinks momentarily. Thats why I suspect a surge, which could be worse with lowgrate cables.
 
Top Bottom