• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required as is 20 years of participation in forums (not all true). Come here to have fun, be ready to be teased and not take online life too seriously. We now measure and review equipment for free! Click here for details.

Mains related hum and noise in power amplifiers

pma

Major Contributor
Forum Donor
Joined
Feb 23, 2019
Messages
2,055
Likes
3,447
Location
Prague
Mains related hum and noise in power amplifiers

We have had some discussion in another thread

https://www.audiosciencereview.com/...they-really-sound-different.18415/post-910720

about the reason of higher residual mains related components hum in the measurement of JC5 amplifier posted by @SIY in post #271

https://www.audiosciencereview.com/...they-really-sound-different.18415/post-910157

Most of the contributors would address the hum to the radiated electromagnetic fields, however I do not think so. Based on decades of years of designing and measuring amplifier circuits and on measurements of electromagnetic interference, I would vote for mistakes in amplifier PCB design and probably insufficient PSRR (power supply ripple rejection) which is related to circuit design. This all is perfectly explained in book [1] by Douglas Self, so I will concentrate to measured plots rather than explaining basics. The study applies to stand-alone class A and class AB amplifiers with unregulated linear power supplies and does not cover ground loops constituted by link cables to DAC or preamp.

Measurements were performed on a functional sample of the Class A power amplifier with idle current of 1.4A, supplied from +/-24Vdc power supply with 230V/2x18V/270VA EI core transformer, center tapped diode rectifier and 2 x 14100uF capacitor bank.

1. Measurements of radiated electromagnetic field near the transformer

Loop sensor probe is used for this measurement which gives output voltage proportional to derivative of magnetic flux according to equations in the image below

loopinducedvoltage.png


The output voltage of the loop sensor is directly proportional to the derivative of intensity of electromagnetic induction dB/dt. The loop sensor was placed side by the transformer as shown in the next image.

tafoloop.JPG


When the transformer is connected only to bridge rectifier with capacitor bank, the loop sensor output voltage looks like this and is dominated by 50Hz, 150Hz and 250Hz spectral lines:

trafoloop2.png


When the amplifier with 1.4A idle current is connect to the output of the capacitor bank, the sensor spectrum shows a lot of high order harmonics due to current impulses that are needed to charge the capacitor bank.

trafoloop1.png


Last, measurement with transformer turned off, showing residual measuring noise
loopnoise.png


This loop sensor is very useful to check the radiated field inside the power amplifier and to try to minimize its possible effect to input circuitry and induction into PCB circuits track loops. We may or not improve the hum components at the power amplifier output by transformer re-positioning and turning, but not always. Often we come to conclusion that output hum components do not depend much on transformer placement or the improvement is not satisfactory. To check if the hum components are a result of transformer and rectifier radiated fields or not we place the transformer and rectifier with capacitor bank outside the amplifier in a distance when intensity of electromagnetic field from the transformer is negligible. 1 – 2m is enough. And we measure mains related components at the amplifier output again. We often find that the difference with power supply inside or outside the amplifier is insignificant so we have to find another sources of hum than the transformer radiated field. Again for further study there is a literature [1]. In my experience the biggest issues regarding hum/buzz components are ground returns, especially ground return of audio signal shared with bypass capacitor returns or dividers connected to supply rails. Several centimeters of shared PCB ground may completely destroy the effort of getting good S/N.

2. Measurements with two grounding schemes

Two grounding schemes were chosen, #1 has about 5cm of shared ground for signal and 47uF bypass capacitors, #2 tries to keep star ground concept as much as possible. Amplifier module under test is class A with idle current of 1.4A, as already stated. Spectrum of 1KHz 5W THD/THD+N is measured and also the spectrum with no input signal. The noise measurement and noise component in THD+N (noise floor) is unfortunately affected by the soundcard used - we are loosing about 6dB in noise.

Grounding scheme #1

gnd1-2_noise.png


gnd1-2_dBv.png


The same in dBr related to 0dBr, which is more usual at ASR, though it is the same thing
gnd1-2.png


Grounding scheme #2

gnd2-2_noise.png


gnd2-2_dBv.png


The same in dBr related to 0dBr, which is usual at ASR though it is the same thing
gnd2-2.png


We can see that even if the transformer radiated field is kept negligible simply by the distance of the power supply from the amplifier, it is the PCB design, namely ground returns, that are responsible for resulting output mains related hum components, together with PSRR of the circuit design. Scheme #2 is much better than #1, but still there is a space for improvements. However it is time consuming and sadly we often see quite unacceptable results in the mass produced amplifiers.

Literature:

[1] Self, D.: Audio Amplifier Power Design Handbook, third edition, Newnes, 2002.
 
Last edited:

SIY

Master Contributor
Technical Expert
Joined
Apr 6, 2018
Messages
6,632
Likes
14,254
Location
Alfred, NY
Mains related hum and noise in power amplifiers

We have had some discussion in another thread

https://www.audiosciencereview.com/...they-really-sound-different.18415/post-910720

about the reason of higher residual mains related components hum in the measurement of JC5 amplifier posted by @SIY in post #271

https://www.audiosciencereview.com/...they-really-sound-different.18415/post-910157

Most of the contributors would address the hum to the radiated electromagnetic fields, however I do not think so. Based on decades of years of designing and measuring amplifier circuits and on measurements of electromagnetic interference, I would vote for mistakes in amplifier PCB design and probably insufficient PSRR (power supply ripple rejection) which is related to circuit design. This all is perfectly explained in book [1] by Douglas Self, so I will concentrate to measured plots rather than explaining basics. The study applies to stand-alone class A and class AB amplifiers with unregulated linear power supplies and does not cover ground loops constituted by link cables to DAC or preamp.

Measurements were performed on a functional sample of the Class A power amplifier with idle current of 1.4A, supplied from +/-24Vdc power supply with 230V/2x18V/270VA EI core transformer, center tapped diode rectifier and 2 x 14100uF capacitor bank.

1. Measurements of radiated electromagnetic field near the transformer

Loop sensor probe is used for this measurement which gives output voltage proportional to derivative of magnetic flux according to equations in the image below

View attachment 153513

The output voltage of the loop sensor is directly proportional to the derivative of intensity of electromagnetic induction dB/dt. The loop sensor was placed side by the transformer as shown in the next image.

View attachment 153514

When the transformer is connected only to bridge rectifier with capacitor bank, the loop sensor output voltage looks like this and is dominated by 50Hz, 150Hz and 250Hz spectral lines:

View attachment 153515

When the amplifier with 1.4A idle current is connect to the output of the capacitor bank, the sensor spectrum shows a lot of high order harmonics due to current impulses that are needed to charge the capacitor bank.

View attachment 153516

Last, measurement with transformer turned off, showing residual measuring noise
View attachment 153524

This loop sensor is very useful to check the radiated field inside the power amplifier and to try to minimize its possible effect to input circuitry and induction into PCB circuits track loops. We may or not improve the hum components at the power amplifier output by transformer re-positioning and turning, but not always. Often we come to conclusion that output hum components do not depend much on transformer placement or the improvement is not satisfactory. To check if the hum components are a result of transformer and rectifier radiated fields or not we place the transformer and rectifier with capacitor bank outside the amplifier in a distance when intensity of electromagnetic field from the transformer is negligible. 1 – 2m is enough. And we measure mains related components at the amplifier output again. We often find that the difference with power supply inside or outside the amplifier is insignificant so we have to find another sources of hum than the transformer radiated field. Again for further study there is a literature [1]. In my experience the biggest issues regarding hum/buzz components are ground returns, especially ground return of audio signal shared with bypass capacitor returns or dividers connected to supply rails. Several centimeters of shared PCB ground may completely destroy the effort of getting good S/N.

2. Measurements with two grounding schemes

Two grounding schemes were chosen, #1 has about 5cm of shared ground for signal and 47uF bypass capacitors, #2 tries to keep star ground concept as much as possible. Amplifier module under test is class A with idle current of 1.4A, as already stated. Spectrum of 1KHz 5W THD/THD+N is measured and also the spectrum with no input signal. The noise measurement and noise component in THD+N (noise floor) is unfortunately affected by the soundcard used - we are loosing about 6dB in noise.

Grounding scheme #1

View attachment 153517

View attachment 153518

The same in dBr related to 0dBr, which is more usual at ASR, though it is the same thing
View attachment 153522

Grounding scheme #2

View attachment 153519

View attachment 153520

The same in dBr related to 0dBr, which is usual at ASR though it is the same thing
View attachment 153521

We can see that even if the transformer radiated field is kept negligible simply by the distance of the power supply from the amplifier, it is the PCB design, namely ground returns, that are responsible for resulting output mains related hum components, together with PSRR of the circuit design. Scheme #2 is much better than #1, but still there is a space for improvements. However it is time consuming and sadly we often see quite unacceptable results in the mass produced amplifiers.

Literature:

[1] Self, D.: Audio Amplifier Power Design Handbook, third edition, Newnes, 2002.
This is indeed well described in Self's books and articles. But... it is not the only source for this noise. This is especially true for amps with large filter caps and hence high peak charging currents. Besides the radiated noise from the current-carrying conductors, the transformers often get close to or even past where they saturate and really splatter around the noise.

The only way to know the mechanism in a specific case is to dive into the specific case. One size fits all is not a valid approach.
 

MakeMineVinyl

Major Contributor
Manufacturer
Joined
Jun 5, 2020
Messages
1,984
Likes
2,937
Location
Santa Fe, NM
This is indeed well described in Self's books and articles. But... it is not the only source for this noise. This is especially true for amps with large filter caps and hence high peak charging currents. Besides the radiated noise from the current-carrying conductors, the transformers often get close to or even past where they saturate and really splatter around the noise.

The only way to know the mechanism in a specific case is to dive into the specific case. One size fits all is not a valid approach.
The thing I see most with toroidal power transformers is that the part of the transformer with the most leakage is where the leads are brought out. Care needs to be taken to position this as far away as possible from sensitive input circuitry.
 

SIY

Master Contributor
Technical Expert
Joined
Apr 6, 2018
Messages
6,632
Likes
14,254
Location
Alfred, NY
The thing I see most with toroidal power transformers is that the part of the transformer with the most leakage is where the leads are brought out. Care needs to be taken to position this as far away as possible from sensitive input circuitry.

My experience as well.
 
OP
pma

pma

Major Contributor
Forum Donor
Joined
Feb 23, 2019
Messages
2,055
Likes
3,447
Location
Prague
OP
pma

pma

Major Contributor
Forum Donor
Joined
Feb 23, 2019
Messages
2,055
Likes
3,447
Location
Prague
This is indeed well described in Self's books and articles. But... it is not the only source for this noise. This is especially true for amps with large filter caps and hence high peak charging currents. Besides the radiated noise from the current-carrying conductors, the transformers often get close to or even past where they saturate and really splatter around the noise.

My experience is following - if the power amplifier is dual-mono design with 2 separated transformers and power supplies, radiated EMF from transformers and rectifiers is not an issue, in case of adequate power amp PCB design. It is also a non-issue if at least a common transformer with separate secondary windings for two channels is used.
However, in case of just one common power supply for both channels, the issues are inevitable, as there is a common power ground for both channels and also signal grounds of both channels are interconnected, thus there is always a loop to which the radiated fields are coupled. This loop can be made small, but cannot be avoided. It cannot be minimized in such way as in case of PCB tracks for a separated power supplies amplifier design. As JC5 is a dual mono design, I suspect the PCB design.
The issue of radiated fields is IMO the reason why complex multichannel power amplifiers like AVR are worse in SINAD than stereo amplifiers. The wiring topology is too complex and it is difficult or impossible to prevent induction of interference voltages.

1631777781539.png


https://parasound.com/product-images/jc5_interior.jpg

https://www.soundstagenetwork.com/i...er&catid=97:amplifier-measurements&Itemid=154

from @SIY measurement with Y-axis calibration added: (0dBV = 1V, -60dBV = 1mV, -80dBV = 100uV, -100dBV = 10uV)

Figure 4 Parasound Halo JC5 Bal vs. Unbal Inputs dBV.png
 
Last edited:

dualazmak

Major Contributor
Forum Donor
Joined
Feb 29, 2020
Messages
1,003
Likes
1,056
Location
Ichihara City, Chiba Prefecture, Japan
....
The issue of radiated fields is IMO the reason why complex multichannel power amplifiers like AVR are worse in SINAD than stereo amplifiers. The wiring topology is too complex and it is difficult or impossible to prevent induction of interference voltages.
....

Hello pma,

Your point in the above is very much suggestive for me.

In June last year, I tested/evaluated Yamaha MX-A5200 (11-Ch multichannel AVR power amplifier) in my multichannel multi-driver multi-amplifier project, and I concluded as;
Even though MX-A5200 functionally works perfectly in my project, from Total Sound Quality (SQ) point of view, I found that MX-A5200 is a little bit inferior, with my ears, to my Reference Sound system (EKIO - DAC8PRO - ACCUPHASE E460 - renovated LC-Network - NS-1000+YST-SW1000+T925A); I decided, consequently, not to use/choose YAMAHA MX-A5200 in my project.

That was actually the start of my long exploration journey of amplifiers in the project, and recently I finally decided to use the four separate HiFi integrated amplifiers as shared in my post here.

I feel that your point would support and validate my selection of "separate/individual well designed HiFi integrated amplifiers" to dedicatedly and directly drive the SP drivers (eliminating LC network) in sensitive and efficient multichannel multi-driver system.
 
  • Like
Reactions: pma
OP
pma

pma

Major Contributor
Forum Donor
Joined
Feb 23, 2019
Messages
2,055
Likes
3,447
Location
Prague
I feel that your point would support and validate my selection of "separate/individual well designed HiFi integrated amplifiers" to dedicatedly and directly drive the SP drivers (eliminating LC network) in sensitive and efficient multichannel multi-driver system.

Konnichiwa @dualazmak , I agree. Use the best amplifiers (re S/N) for the multi-driver project .
 
OP
pma

pma

Major Contributor
Forum Donor
Joined
Feb 23, 2019
Messages
2,055
Likes
3,447
Location
Prague
Anyone of those audible?

Yes, grounding scheme #1 is audible even with medium sensitivity speaker like 86dB/W/m. The reason is spectral forest that spreads into middle frequencies where ear is very sensitive. It is audible rather as a buzz than a hum.
 

DualTriode

Senior Member
Joined
Oct 24, 2019
Messages
481
Likes
274
Hello All,

Power line 60Hz or 50Hz plus harmonics are the worst.

If I recall from the Aiyima threads the High Frequency switching supplies tend to not have this type of problem.

Thanks DT
 
OP
pma

pma

Major Contributor
Forum Donor
Joined
Feb 23, 2019
Messages
2,055
Likes
3,447
Location
Prague
Hello All,

Power line 60Hz or 50Hz plus harmonics are the worst.

If I recall from the Aiyima threads the High Frequency switching supplies tend to not have this type of problem.

Thanks DT

Yes if the SMPS leakage current is eliminated. I have to measure AIYIMA + Meanwell SMPS with the SMPS supplied from isolation 1:1 transformer (low stray capacitance) to get spectra that are completely free from mains line components. I guess you understand that the PSU becomes an Oxymoron then. The best results (without need of isolation transformer and other band aids) are with the linear regulated PSU. AIYIMA itself has quite poor PSRR.
 
Top Bottom