just using digital attenuation results in noise

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Building a headphone amplifier I want to control the volume just from the computer, not with an analog attenuator so to speak. This because of lack of space in the enclosure.

It may sound obvious (and probably is) but I hear more 'digital' noise than using both a variable resistor and the volume control on the computer.

I put a resistor of 50K from input to ground (that seemed to be necessary anyway) but it does not help.

Is this just the effect of digital attenuation?
 
Last edited:
Administrator
Joined 2007
Paid Member
Its an effect of the attenuation system used on your computer. Digital volume controls vary enormously on how well they are implemented and what the resolution of the digital control is. The audio level is reduced but the inherent noise of the DAC tends to remain the same.

An analogue control after the DAC improves things because you are turning the DAC level up higher to compensate for the additional analogue control that is in circuit.

If the headphone amp has lots of gain then adding a fixed resistive attenuator at the amp input might help. Arrange for the attenuator to be just enough so that you get maximum desired volume when the PC volume is on maximum. Try something like a 10k and 3k3 divider for starters.
 
Its an effect of the attenuation system used on your computer. Digital volume controls vary enormously on how well they are implemented and what the resolution of the digital control is. The audio level is reduced but the inherent noise of the DAC tends to remain the same.

An analogue control after the DAC improves things because you are turning the DAC level up higher to compensate for the additional analogue control that is in circuit.

If the headphone amp has lots of gain then adding a fixed resistive attenuator at the amp input might help. Arrange for the attenuator to be just enough so that you get maximum desired volume when the PC volume is on maximum. Try something like a 10k and 3k3 divider for starters.
Thank you for your clear explaination! That must be the cause.

With these high temperatures in Holland I used earbuds and not my Beyerdynamics DT770 pro 80 ohm headphones. The latter can only be used in winter :D

But when I tried the Beyerdynamics I could hardly hear any noise. Otherwise I will try you suggestion.
 
Bit off topic, but this is wat I built:



It is a max4410 with a DC-DC converter to lower the supply voltage (an external switched regulator) to 5 Volts. Inside the box are two ferrite toroids.

The enclosure is a brushed aluminium Hammond box, 3x5x6 centimeters.

The coin on the front obscures the hole where the potentiometer was supposed to be.

Although considered mediocre, the first impressions of the max4410 are not too bad.
 
Is this just the effect of digital attenuation?


Probably - to attenuate well in the digital domain you need to work at a higher resolution than the signal source (ie 24 bit for CD quality, not 16 bit). A good digital subsystem upsamples/oversamples to high resolution internally for all the processing, then downsamples/decimates for the actual DAC at the last stage, possibly using noise-shaping if appropriate. This aims to avoid adding another chunk of quantization noise onto the already quantized signal, and push quantization noise above the audio band.
 
Probably - to attenuate well in the digital domain you need to work at a higher resolution than the signal source (ie 24 bit for CD quality, not 16 bit). A good digital subsystem upsamples/oversamples to high resolution internally for all the processing, then downsamples/decimates for the actual DAC at the last stage, possibly using noise-shaping if appropriate. This aims to avoid adding another chunk of quantization noise onto the already quantized signal, and push quantization noise above the audio band.
I never used upsampling, it must be better but for me it was not necessary. However I will test it if it is possible.

The way I use the amp without analog attenuation is not much different from the way I used it with analog attenuation. The potentiometer was always fixed half way and I changed the volume on the computer.
With this new amp the gain is less and the 'analog volume' is at maximum so to speak. So I was surprised to hear noise.

After further experimenting I settled for the music player Deadbeef with the ALSA audio system, bypassing the PulseAudio audio system.
I now have near 'bit perfect' audio without the noise that I had before, at least I cannot hear it anymore :)
 
You say the gain is less, but how much is it? Even if it's just 0...+6 dB, this may still be far too much for sensitive headphones. If you can plainly hear 16 bit digital noise, you need a lot less... probably at least 20 dB. Where is your digital attenuation set now? I would want this to be at around -20 dB at normal listening volume for a decent amount of headroom.

I guess you may have underestimated just how much attenuation a volume pot actually provides... the 12 o'clock position tends to be at around -26 dB if memory serves. Even when followed by typical headphone amp gain of +8 to +16 dB, total gain remains substantially negative.

Once you have figured out how much attenuation you need, there is little keeping you from implementing a resistor divider to replace the pot. Most modern soundcards tend to be pretty good load drivers, so you may very well get along with ~3 kohms total. (The specifics would depend on your hardware.)

If you say you have an "external switched regulator", is that a wall-wart style power supply? I'm asking because you want some galvanic isolation there. I guess it's present since you say the noise is gone now.
 
You say the gain is less, but how much is it? Even if it's just 0...+6 dB, this may still be far too much for sensitive headphones. If you can plainly hear 16 bit digital noise, you need a lot less... probably at least 20 dB. Where is your digital attenuation set now? I would want this to be at around -20 dB at normal listening volume for a decent amount of headroom.

I guess you may have underestimated just how much attenuation a volume pot actually provides... the 12 o'clock position tends to be at around -26 dB if memory serves. Even when followed by typical headphone amp gain of +8 to +16 dB, total gain remains substantially negative.

Once you have figured out how much attenuation you need, there is little keeping you from implementing a resistor divider to replace the pot. Most modern soundcards tend to be pretty good load drivers, so you may very well get along with ~3 kohms total. (The specifics would depend on your hardware.)

If you say you have an "external switched regulator", is that a wall-wart style power supply? I'm asking because you want some galvanic isolation there. I guess it's present since you say the noise is gone now.

I tested the resistance between the 'input/mains' and 'output' of the switched power supply before using it and the resistance was too high to measure.

With input disconnected I hear no noise at all. So the noise I hear originates from the pc, not the switched power supply (which I got from an old laptop). And I only hear noise using the earbuds, not the Beyerdynamics.

My conclusion is that the earbuds are much more sensitive than I expected just like you said. I barely opened the volume when using these.

I normaly use the Beyerdynamics but with this warm weather of the past weeks these headphones soon started to look like a swimming pool... That's why I switched to the earbuds.
 
Last edited:
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.