Hi,
i had made my own composite Amp with the LM3886.
The Amp runs very stable under all conditions.
Also the THD is ok for me , in a unbalenced ,direct input using of Signal and GND.
But , all if I try , i do not get a better noise, all under FRQ below the Test 1Khz tone is bad...
If I use a regulated PSU , the Noise is also... if I drill +,- and Gnd it becomes better.
The PSU I made is from a "Ringkerntrafo" mit 40.000µF each and a Ripple Eater,Snubber ... so the Vcc to the AMP is very very clear.
So less Ripple, but much noise.... I do not use fast Diodes , I use Bridge Rec. for smooth acting.
Attached , you will find the measuremts diagramm: Input for 10W @ 4Ohm Load
I think the 100 and 200Hz could be a typical Effect, but which one.
I also try with common mode chokes in line of PSU, no bettter result.
I will be lucky for each Tip.
i had made my own composite Amp with the LM3886.
The Amp runs very stable under all conditions.
Also the THD is ok for me , in a unbalenced ,direct input using of Signal and GND.
But , all if I try , i do not get a better noise, all under FRQ below the Test 1Khz tone is bad...
If I use a regulated PSU , the Noise is also... if I drill +,- and Gnd it becomes better.
The PSU I made is from a "Ringkerntrafo" mit 40.000µF each and a Ripple Eater,Snubber ... so the Vcc to the AMP is very very clear.
So less Ripple, but much noise.... I do not use fast Diodes , I use Bridge Rec. for smooth acting.
Attached , you will find the measuremts diagramm: Input for 10W @ 4Ohm Load
I think the 100 and 200Hz could be a typical Effect, but which one.
I also try with common mode chokes in line of PSU, no bettter result.
I will be lucky for each Tip.
Last edited:
Is the noise audible? From your graph the signal to noise ratio looks better than 90db. Hardly ‘bad’ but maybe I am misinterpreting that graph. So it would be useful to know what you perceive. There doesn’t seem to be any problem with your power supply and the fact that between regulated and not makes no difference sort of confirms that. So it comes down to the grounding arrangements. Can you show some pictures if layout. Perhaps someone will spot something
yes you are right... the noise is under 94,5db under drilling the +,-and gnd...
I think i oversized the GND on my 4 layer PCB...
I use the Starground concept.... See the 2th innerlayer.... and Bottom with High Power paths....
The best (17.06.22) I get out for my second Board is:
(my second board is a smaller Version of my first, but i separte all and reduce to AMP for better messasuremets
If I make a noise omly measurement:
50Hz , 150Hz, 250 Hz are the biggest noise sources ....
Thats is the FRQ which comes out from the 230V , i have here 50Hz.
So now is the question how to block mthis noise if i has a snubber and all on my PSU....
Goal of my is noise below -100db
(my second board is a smaller Version of my first, but i separte all and reduce to AMP for better messasuremets
- FRQ Res. + 0,010 / - 0,012
- 95,5 db noise
- 0,0017 THD
- 0,008 THD % Noise
If I make a noise omly measurement:
50Hz , 150Hz, 250 Hz are the biggest noise sources ....
Thats is the FRQ which comes out from the 230V , i have here 50Hz.
So now is the question how to block mthis noise if i has a snubber and all on my PSU....
Goal of my is noise below -100db
Last edited:
Maybe you'll laugh but try with only 4700uF per rail (4700uF on positive and 4700uF on negative).
Just try!
Just try!
i will try it tomorrow, of course, trying is no problem.
On what based your idea to reduce the front Caps ? ?
there is a tread here called Taming the LM3886. A lot of great info on how component placement and layout effects the performance. The location of the transformer, power cables, on off switch, input cables, volume control, type of metal for the case and of course nearby components, Balanced inputs may help.
@Micha2007 - here are a couple of thoughts.
I tested my (much more humble / close to datasheet) LM3886 amp yesterday just to see what was going on. I have a very low spec picoscope so I can't do accurate SNR etc (the AWG on the scope is way too noisy). However, replicating as best I could your first plot (1kHz in and an FFT trace of output) I definitely do not get any noticeable mains harmonics (50,100,150 etc). If they are there they are buried in my noise floor. My best estimate of the SNR I get is that it is in the -90 to -95 range, possibly better. That suggests your much more sophisticated implementation ought to do better than it is and that those low mains-related harmonics are a consequence of layout / design. My power supply is very standard and uses 10,000muF capacitors on each line. The LM3886 has a fantastic PSRR so I seriously doubt there is anything breaking through from your power supply.
Have you tried using a 'near perfect' power supply and seeing what comes out. 2 x 20V high capacity l-ion batteries of the sort you get with serious power tools. That might be revealing.
I tested my (much more humble / close to datasheet) LM3886 amp yesterday just to see what was going on. I have a very low spec picoscope so I can't do accurate SNR etc (the AWG on the scope is way too noisy). However, replicating as best I could your first plot (1kHz in and an FFT trace of output) I definitely do not get any noticeable mains harmonics (50,100,150 etc). If they are there they are buried in my noise floor. My best estimate of the SNR I get is that it is in the -90 to -95 range, possibly better. That suggests your much more sophisticated implementation ought to do better than it is and that those low mains-related harmonics are a consequence of layout / design. My power supply is very standard and uses 10,000muF capacitors on each line. The LM3886 has a fantastic PSRR so I seriously doubt there is anything breaking through from your power supply.
Have you tried using a 'near perfect' power supply and seeing what comes out. 2 x 20V high capacity l-ion batteries of the sort you get with serious power tools. That might be revealing.
Mnay thanks for you Tips ...I tried with two 24V LKW Batteries from my uncle.... then I come under r.a. -102dB noise and the Amp performs now with 0,00083 THD......... so my GND conccept is not good and "swing" on the GND planes under load.... this is what I see.....
Last edited:
... ok I take up you hint, but please say in open Words what is " still ...." ?
And what could I make better form your side of seeing ?
Last edited: