Hi
I use a DAC connected directly to a power amp to drive my speakers. The DAC has digital attenuator which has to be set to ~-40db for comfortable listening volume. It is said to be a bit too much of digital attenuation (low SNR), and I was suggested to use additional attenuation.
I checked in-line attenuators, which is the cheapest and simplest approach, but some folks here say they're not good. I could also buy a passive preamp, but then again, it adds the distortion of it's own.
What is the recommended approach in my situation?
I use a DAC connected directly to a power amp to drive my speakers. The DAC has digital attenuator which has to be set to ~-40db for comfortable listening volume. It is said to be a bit too much of digital attenuation (low SNR), and I was suggested to use additional attenuation.
I checked in-line attenuators, which is the cheapest and simplest approach, but some folks here say they're not good. I could also buy a passive preamp, but then again, it adds the distortion of it's own.
What is the recommended approach in my situation?
Last edited:
Make a "passive preamp" with a potentiometer, simple. Much better than using the DAC's digital attenuator
I think I'd be just fine to have a fixed attenuation of -20db. And I believe such a device would be simpler and cheaper too, comparing to passive preamps. But then I read folks on this forum saying that in-line attenuators suck. Are they really that bad, it doesn't make sense to look in their direction? If so, what makes passive pre's a better choice, provided I am ok with fixed attenuation? Or are those folks basically saying that all types of passive attenuation suck?
It sounds like you want good quality. Then the best way is no digital attenuation, then the only way is analogue, and that is achieved with a potentiometer. Like I said, simple
you can buy fixed attenuators. They are not cheap.
You can build your own fixed attenuator and it will fit inside the RCA plug.
A -20dB attenuator can be created from a 10k and 1k0 pair of resistors. (not exactly -20dB but probably close enough).
You can build your own fixed attenuator and it will fit inside the RCA plug.
A -20dB attenuator can be created from a 10k and 1k0 pair of resistors. (not exactly -20dB but probably close enough).
No, the advantage would be that you can leave the digital attenuation at zero and adjust your listening level with the potentiometer (volume control) and achieve maximum sound quality at all levels.
Here's what John Westlake (audiolab MDAC designer) says about passive preamps:
Typical passive pots will introduce there OWN thermal noise source – a 50K resistor has a 107dBV noise contribution at 25Deg.C 20K BW… this needs to be added to the systems noise figure (but bare in mind the absolute valve changes depending upon the volume position – thermal noise is determined by system Impedances)… the advantage of direct connection between DAC output and Power amplifier is you don’t introduce additional noise sources… Off the top of my head, MDAC’s EIN (Equivalent Input Noise) is about that of a single 2.5K resistor.
As you can see, there’s no point worrying about Digital verses Analogue gain control, you will ultimately face the same limitations of thermal noise – however its somewhat easier with a Digital preamplifier to reduce the systems total thermal noise levels by maintaining a lower impedance signal path to the amplifier (Higher circuit impedance = higher thermal noise levels).
A Passive pot just attenuates the signal into the system noise floor (while contributing a significant noise source of is own). The Digital system attenuates the signal into the amplifiers noise floor – as the passive pot, but WITHOUT the extra noise contribution of higher circuit impedances a passive pot introduces).
Only by having ultra low impedance levels can a passive pot result in improved SNR - however you would need a power amplifier to drive these circuit impedances without introducing distortion due to the current requirements to drive these impedances. Feasible, but not easy!!! and not with a typical Pot - also, Pots become very non linear at higher currents - you would need to use a stepped attenuator... say 50 Ohm environment, 2V RMS... distortion free is going to require current and result in significant power dissipation... one things for certain a typical 50K or even 10K passive pot is not going to cut the mustard here...
Typical passive pots will introduce there OWN thermal noise source – a 50K resistor has a 107dBV noise contribution at 25Deg.C 20K BW… this needs to be added to the systems noise figure (but bare in mind the absolute valve changes depending upon the volume position – thermal noise is determined by system Impedances)… the advantage of direct connection between DAC output and Power amplifier is you don’t introduce additional noise sources… Off the top of my head, MDAC’s EIN (Equivalent Input Noise) is about that of a single 2.5K resistor.
As you can see, there’s no point worrying about Digital verses Analogue gain control, you will ultimately face the same limitations of thermal noise – however its somewhat easier with a Digital preamplifier to reduce the systems total thermal noise levels by maintaining a lower impedance signal path to the amplifier (Higher circuit impedance = higher thermal noise levels).
A Passive pot just attenuates the signal into the system noise floor (while contributing a significant noise source of is own). The Digital system attenuates the signal into the amplifiers noise floor – as the passive pot, but WITHOUT the extra noise contribution of higher circuit impedances a passive pot introduces).
Only by having ultra low impedance levels can a passive pot result in improved SNR - however you would need a power amplifier to drive these circuit impedances without introducing distortion due to the current requirements to drive these impedances. Feasible, but not easy!!! and not with a typical Pot - also, Pots become very non linear at higher currents - you would need to use a stepped attenuator... say 50 Ohm environment, 2V RMS... distortion free is going to require current and result in significant power dissipation... one things for certain a typical 50K or even 10K passive pot is not going to cut the mustard here...
What is meant by digital attenuation as he is applying it here? He uses the term "Digital verses Analogue gain control" but we are talking about attenuation not gain.
digital preamps have no gain control per se, so I believe he meant to say volume control (which is in fact the attenuation). The point being (as I get it) - passive preamps have distortion of their own, it's not a remedy, except for some remarkable implementations.
Last edited:
You see, I'm starting to wonder if it is actually a digital attenuator in your Dac or not, that is, is it affecting the pcm digital code, or merely the output level? If it is the former then digital information is being lost, ie, distorted, if the latter it is not a problem
Resistor of suitable value in the RCA jack, calculated (for 20dB attenuation) to be six time the input impedance of your amplifier.
However remember the source impedance looking into the amp is now extremely high, and most amplifiers do NOT like being driven from high impedance. For one, it raises the noise floor to something stupid. For another, it severely limits the current drive ability of the preceding output stage.
For a 32-bit digital volume control, -40dB of attenuation is approximately 7 bits of resolution lost (or 25 still left), which results in a total digital noise floor of below -150dBV. Since the I/V and LPF are much higher, they will still be the limiting factor, not the digital VC.
I would wager no analog system, passive or otherwise, can come close to it, not by a long shot. A potentiometer at mid-point would typically also reduce noise floor in the analog domain, which is real resolution lost (the noise floor of the amplifier is fixed, so moving a potentiometer closer to max attenuation reduces S/N ratio by the exact same quantity as digital attenuation).
You need to stop worrying about inconsequential things.
Of course, if your DAC has a standard 16-bit control, then it's an issue but most DAC have internal volume controls of at least 32 bits of resolution.
Last edited:
I think I am being consequential.... the DAC's output gain is fixed (and the noise level too, for that matter), the only thing that is changing when using volume control on the DAC is signal level (and consequently SNR). That means between -40db and -20db there is 4 times difference in the perceived noise loudness. That's plenty, huh? So theoretically speaking by adding an attenuator I can crank up signal level to up to 0db, thus maximizing SNR, without making my ears bleed. I realize that the attenuator will add it's own noise though. Where am I wrong?
Last edited:
Mmm, I was hoping sangram had answered your question, as he is obviously more familiar with modern digital sources than me, I only have CD player for music, and occasionally stream via a chromecast, not hifi. What is your source exactly?
My DAC is an Audiolab MDAC
my power amp is an Adcom GFA 5500.
The DAC is using a Sabre chip (32bit). Like sangram mentioned, volume control resolution loss is not a problem for this kind of precision. But that is not my concern. My concern is when you use a digital volume control (no matter what implementation), noise level remains static and it is always at it's max. Means the lower your listening level is the worse is the SNR. To get the best result, you should stay at 0db. But in my case, the volume has to stay low (below -40db), so that concerns me a bit as I don't seem to be getting the best out of my DAC.
my power amp is an Adcom GFA 5500.
The DAC is using a Sabre chip (32bit). Like sangram mentioned, volume control resolution loss is not a problem for this kind of precision. But that is not my concern. My concern is when you use a digital volume control (no matter what implementation), noise level remains static and it is always at it's max. Means the lower your listening level is the worse is the SNR. To get the best result, you should stay at 0db. But in my case, the volume has to stay low (below -40db), so that concerns me a bit as I don't seem to be getting the best out of my DAC.
I see your point, this is the age old dilemma of attenuation followed by amplification, there are always going to be compromises. Ideally one wouldn't attenuate at all and have an amplifier with variable gain, it can be done but raises it's own problems, mostly ones of stability. I think under the circumstances you would be better off with a low gain amplifier so you didn't have to attenuate so much, this is the road a lot of audiophiles go down
A total 40dB of analog output attenuation will lower the signal closer to the fixed noise floor by 40dB, whether you achieve that in digital or analog domain.
The two issues are totally separate. The DAC controls volume by decimating the signal but it's an exact, mathematical reconstruction process. It results in a loss of S/N ratio in the exact same way as an analog potentiometer, with the caveat that you lose 1 bit of resolution for every six dB of attenuation. This worsens the S/N ratio by 6dB (16 bit = 6x16 = 96dB, 24 bit = 6x24 = 144dB and so on), which is exactly the same as analog attenuation.
The big issue with digital attenuation is the loss of bit depth, which leads to a loss of fidelity due to the inaccuracy of reconstruction. For example with 7 bits lost from a 16-bit signal, you'd have a very inaccurate representation of any sort of analog signal. With a 32-bit control, that is no longer an issue.
- Status
- Not open for further replies.