background noise in circuit in absence of audio

Thread Starter

abc77

Joined May 10, 2017
59
I am getting background noise all the way from output of DAC and hearing it in my headphone.

My connection is as follows:

DAC(PCM1789) --> LPF/ audio amplifier(TPA6138)

How can I eliminate the background noise? Attached is schematic

The noise seems to originate from the DAC and the LPF does not suppress it.

some information,
- the background noise is there at the output of DAC all the way to the headphone, that is, when no audio is played on PC
- when I make PC volume low, the noise is reduced and I can hear clearer audio/music
- when I increase PC volume, the noise is also increased such that I cannot hear audio
- created seperate power supply for LPF/amplifier thinking it may have come from power supply or breadboard
- the DAC to LPF/amplifier is connected using breadboard wire

sch_abc77.gif

Mods edit:
Please post the circuit as 800x600 *.jpg or *.gif file, many of members may not open the pdf to see the contents.
 

Attachments

Last edited:

Alec_t

Joined Sep 17, 2013
10,597
R14/15 give a 100:1 attenuation and put the signal closer to the background noise level. I realise that the inputs of the TPA6138 mustn't be over-driven, but why so much attenuation?
Your LPF component values differ widely from those suggested in the DAC datasheet. In particular, C14/15 seem to be an order of magnitude too low. Why so much difference?
I'm not clear why you don't have the PC audio output muted, since the input to the DAC is surely digital?
 

Thread Starter

abc77

Joined May 10, 2017
59
the attenuation was because the volume was so loud and i came to this values listening to audio. i followed the LPF component from LPF/amplifier TPA6138 datasheet.

the last thing i didn't understand. I am sending audio from PC USB-> microcontroller->DAC-LPF/amplifier-headphone.
 

Thread Starter

abc77

Joined May 10, 2017
59
R14/15 give a 100:1 attenuation and put the signal closer to the background noise level. I realise that the inputs of the TPA6138 mustn't be over-driven, but why so much attenuation?
Your LPF component values differ widely from those suggested in the DAC datasheet. In particular, C14/15 seem to be an order of magnitude too low. Why so much difference?
I'm not clear why you don't have the PC audio output muted, since the input to the DAC is surely digital?
should i be following LPF component value from DAC datasheet? I unknownly was following the LPF/amplifier TPA6138 values.
 

Alec_t

Joined Sep 17, 2013
10,597
I don't know which of the two datasheets would be the better one to follow.

Edit:
Simulation of both LPFs suggests that the one you are using (from the TPA6138 datasheet) actually suppresses the hf noise better.
FilterComparison.PNG
 
Last edited:

Thread Starter

abc77

Joined May 10, 2017
59
i tested the LPF recommendation from the DAC as per your idea and the audio is better

thanks to you

however the sound not quite a quality that I should get-
1) loud, can hear audio from outside without plugging the headphone into ear
2) when i put attenuation i get the noisy audio again
3) the audio sound signal frequency(or phase) seems shifted, for example- like man voice sounding like women(this is just to make understand)
4) there is some little background noise but not so much like earlier, in fact, far better, but even then....

What can I do to solve the above problems?

the LPF recommendation from DAC is better in the sense that,
1) using TPA6138 LPF value produced noisy audio(with accentuation and without)
2) when volume is increased there is no increase in noise level which now doesn't happen(happens only when at at max volume but little)

is the spice model of the DAC and TPA6138 that you used in simulation?
 
Last edited:

Alec_t

Joined Sep 17, 2013
10,597
is the spice model of the DAC and TPA6138 that you used in simulation?
Neither is simulated accurately or fully. The DAC crude model consists of V1,V5, V6, V4. V1 and V5 are simple 1kHz and 40kHz voltage sources of one polarity, whereas V4 and V6 are 1kHz and 40kHz sources of the opposite polarity, representing the DAC differential output. The TPA6138 is represented by an LT1006 opamp.
Reducing both R14 and R15 by a factor of 10 might reduce the noise pickup. If the circuit is on a breadboard, that is a prime candidate as a noise pick-up device.
 

Thread Starter

abc77

Joined May 10, 2017
59
so any suggestion to lower noise floor when i put the attenuation circuit back(to decrease volume)?

why is the sound pitch changed?
 

Thread Starter

abc77

Joined May 10, 2017
59
See above re R14/15 and breadboard use.

I would guess the bass frequencies aren't getting through the filter/amp.
i checked your suggestion on reducing the attenuation, reduced it to 10 ratio, the noise floor is reduced but the problem of loudness remains.

i agree with what you said on both the noise floor and pitch of audio. But how to attenuate the signal at 100 ratio(because this gives me best range on loudness) without increasing the noise floor and how to pass all the frequencies in my case?

edit: reducing volume seems to pass the required frequencies
 

Alec_t

Joined Sep 17, 2013
10,597
I've just run a sim of the frequency responses of the two LPF versions :
FrequencyResponsesCompared.PNG
As you can see, the response (Vout2) for the LPF suggested in the '6138 datasheet is pants: lower audio frequencies are drastically attenuated, as I guessed. The response (Vout1) for the LPF in the '1789 DAC's datasheet is much better, being almost flat across the audio spectrum and providing an extra 6db (compared to the other LPF's response) of attenuation around 192kHz (the sampling frequency of the DAC).

Edit: Oops. Simulation error. Ignore the comments re Vout2.
 
Last edited:

Alec_t

Joined Sep 17, 2013
10,597
Regarding the noise v attenuation problem, I'd try attenuating the signal before the LPF no more than is necessary to avoid clipping, then applying any further attenuation as necessary after the LPF.
 

Thread Starter

abc77

Joined May 10, 2017
59
I've just run a sim of the frequency responses of the two LPF versions :
View attachment 133386
As you can see, the response (Vout2) for the LPF suggested in the '6138 datasheet is pants: lower audio frequencies are drastically attenuated, as I guessed. The response (Vout1) for the LPF in the '1789 DAC's datasheet is much better, being almost flat across the audio spectrum and providing an extra 6db (compared to the other LPF's response) of attenuation around 192kHz (the sampling frequency of the DAC).
But I think I have sampling frequency of 48kHz, could this be making the difference?
 

Alec_t

Joined Sep 17, 2013
10,597
The filter response minimum should preferably be matched to the sampling frequency, so component values will need changing if you are indeed using 48kHz.
 

Thread Starter

abc77

Joined May 10, 2017
59
the DAC datasheet typical specification are rated for 48kHz as i understand. how did you note the 192kHz sampling frequency?

the audio pitch and loudness/noise floor increase problems starts right at the output of DAC.implying the problem is migrated across the LPF/amplifier, attenuation circuit

I directly connected DAC output to the headphone and there is still the pitch problem(some frequencies not available maybe)
 

Thread Starter

abc77

Joined May 10, 2017
59
Regarding the noise v attenuation problem, I'd try attenuating the signal before the LPF no more than is necessary to avoid clipping, then applying any further attenuation as necessary after the LPF.
i liked the idea and tried this but unfortunately same result.
 
Top