My amps have somewhat too much gain/power in my setup for my speakers/room/listening levels which causes me to run my RME ADI-2 DAC at 30-45dB digital attenuation on the +1dBu output level which results in about 75-90dB SNR on the DAC output or so it seems. Is there much to be gained by adding say -20dB passive inline XLR attenuators on the amp inputs so I can get better SNR out of the DAC? My hunch is that any gains will be below the room background noise, plus those attenuators will add some thermal noise of their own even if small amount, hence the question in the subject line.
-
WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!
Inline passive attenuators, should I bother?
- Thread starter gvl
- Start date
jhenderson0107
Active Member
There is no advantage to use of an analog attenuator. Both analog and digital attenuation reduce the theoretical SNR of the preamp output signal. Good, modern amps provide only 100 dB SNR, so using a 20 dB attenuator between the preamp and amp is dominated by the latter.
- Thread Starter
- #3
I'm getting less than 90dB out of my DAC w/out the attenuators, this would seem to dominate and not the amp if the latter can do 100dB SNR. But I get it, at my listening levels way below max power the amp's SNR is probably much worse than 100dB and it will in fact dominate.
Do you have audible noise issues with your current setup? If so analog attenuators may help.
In most cases system noise performance is dominated by noise from your DAC as noise from your DAC will be multiplied by your amplifier gain. However if you don't have audible noise issues with your volume control at your max listening level and nothing playing then an analog attenuator will not audibly improve anything.
What specific amplifiers are you using?
Michael
In most cases system noise performance is dominated by noise from your DAC as noise from your DAC will be multiplied by your amplifier gain. However if you don't have audible noise issues with your volume control at your max listening level and nothing playing then an analog attenuator will not audibly improve anything.
What specific amplifiers are you using?
Michael
- Thread Starter
- #5
No noise I can detect at my listening position, it's Emotiva XPA-1 gen 2s. I realize they are overkill for my current application, but this is what I have at the moment.
Looks like the Emotiva has a 29 dB gain and SNR of 89 dB at 1 W in to 8 ohm. I assume the SNR is actually A weighted which means unweighted is about 3 dB worse but let's give them the benefit of the doubt and say unweighted. Residual noise at the amp output is sqrt(1 x 8) x 10^(-89/20) = 100 uV, not bad but not great.
RME has a SNR of 117 dB at +1 dBu (0.775 V) unweighted. Residual noise at the DAC output is 0.775 x 10^(-117/20) = 1.1 uV, this is very, very good and shows the benefit of the analog attenuation steps built in to the RME volume control.
Multiplying the DAC noise by amplifier gain gives a DAC contribution at the amplifier output of 1.1 x 10^(29/20) = 31 uV. Which means in your case the amplifier noise contribution is actually more than the DAC which is a bit unusual.
Total residual noise at the amplifier output is sqrt(31^2 + 100^2) = 105 uV. This is actually rather good and it is not surprising you do not hear any noise.
Adding -20 dB attenuation at the DAC output would reduce the DAC output noise from 1.1 uV to 1.1 x 10^(-20/20) = 0.1 uV which becomes 0.1 x 10^(29/20) = 3 uV at the amplifier output. Combined noise in this case would be sqrt(3^2 + 100^2) = 100 uV which is barely different from you have now. The only way to get appreciably lower noise would be to get a lower noise amplifier.
Michael
chips666
Active Member
Level mismatch solutions - fixed and variable attenuators (Page 1) — Tips & Tricks — RME User Forum
Level mismatch solutions - fixed and variable attenuators (Page 1) — Tips & Tricks — RME User Forum —
forum.rme-audio.de
This maybe usefull...
- Thread Starter
- #8
Thanks, this makes sense.
avanti1960
Active Member
- Joined
- Dec 31, 2019
- Messages
- 121
- Likes
- 70
- Thread Starter
- #11
It sounds fine, the signal is well attenuated on the DAC so there is no input overload.
My pre-amp had a stupid amount of gain. The main problem was that it was loud at 9 o'clock and dangerously loud at 10 thru a 400wrms/ch@4ohms amp driving low sensitivity speakers. The volume control is not smooth so one step resulted in a big increase in SPL. I was just wondering what you meant by "strained" and "overdriven". I don't think I heard anything bad at moderate volume and I associate those things with an underpowered amp trying to drive speakers of inadequate sensitivity, in which case a 10db attenuator would make matters a whole lot worse.
One possible reason it might make sense:
What would happen if your RME came on at 0dB level - either through fault or because some nutter turns it up.
Would your speakers survive?
What would happen if your RME came on at 0dB level - either through fault or because some nutter turns it up.
Would your speakers survive?
- Thread Starter
- #14
Nope. It is a concern. I’m careful though, but things happen.
DVDdoug
Major Contributor
- Joined
- May 27, 2021
- Messages
- 3,024
- Likes
- 3,979
If everything sounds OK, I wouldn't worry about it. 
48dB is 8-bits so with 45dB of digital attenuation and 24-bit audio you still have more than 16-bits. But with 16-bit audio you are almost down to 8-bits of resolution and you may hear quantization noise. Quantization noise sounds like a "fuzz" on top of the signal and like regular analog noise it's most noticeable with quiet sounds. But unlike analog noise it goes-away completely with digital silence. (You can make an 8-bit file with Audacity. If you try it, turn off Audacity's dither.)
This is not a concern with regular "digital volume control" when you don't have "excess gain" because the quantization noise remains the same low-level (usually inaudible) as you turn-down the volume digitally. The problem comes when you attenuate and then re-amplify, and your high-gain amp is amplifying any quantization noise.
An analog attenuator at the DAC output will (obviously) attenuate the signal and noise together so the S/N ratio remains the same at that point. Without re-amplification, that lowers the absolute level of any noise making it less audible. The high-gain amp will re-amplify that existing noise and add-generate SOME analog noise of its own, but the noise generated by the amp is unavoidable and will (obviously) still exist with digital or analog attenuation.
That's usually my argument, but...
48dB is 8-bits so with 45dB of digital attenuation and 24-bit audio you still have more than 16-bits. But with 16-bit audio you are almost down to 8-bits of resolution and you may hear quantization noise. Quantization noise sounds like a "fuzz" on top of the signal and like regular analog noise it's most noticeable with quiet sounds. But unlike analog noise it goes-away completely with digital silence. (You can make an 8-bit file with Audacity. If you try it, turn off Audacity's dither.)
This is not a concern with regular "digital volume control" when you don't have "excess gain" because the quantization noise remains the same low-level (usually inaudible) as you turn-down the volume digitally. The problem comes when you attenuate and then re-amplify, and your high-gain amp is amplifying any quantization noise.
An analog attenuator at the DAC output will (obviously) attenuate the signal and noise together so the S/N ratio remains the same at that point. Without re-amplification, that lowers the absolute level of any noise making it less audible. The high-gain amp will re-amplify that existing noise and add-generate SOME analog noise of its own, but the noise generated by the amp is unavoidable and will (obviously) still exist with digital or analog attenuation.
Last edited:
- Thread Starter
- #16
This is not how it works. You can’t lose information if it wasn’t there in the first place.
Unlikely. The 16 bit audio will be first upsampled to at least 24 bits. Probably 32 or 64 bit floating point. The DSP/volume control will be done from there. Even if the processing is only done at 24 bits, then the reduction will be back down to 16 bits. (The bottom 8 bits of the 24 bits will only contain the noise floor from the 16 bit recording, so no information is thrown away.)
The question was months ago but I still think it is worth pointing that you should not set the RME ADI-2 at +1 dBu output level and add lots of digital attenuation later. This is defeating the purpose of the device in the first place! Instead of setting +1 dBu output, that is, increasing analog gain, and later attenuating digitally or analog, the correct would be to use lower gain to begin with. And even better forgetting all this and just letting the brillian AutoRef turned on, so it uses minimum gain needed and minimum digital volume attenuation for maximum SNR.
See here.
RME ADI-2 FS Version 2 DAC and Headphone Amp Review
Sure it can. I use the original AEON with it, and unless you want to enter real basshead territory (at least +20 dB gain below 60 Hz plus earth-shaking volume) it is fully ok. Nice. Original AEONS had 92dB/1mW sensitivity. New AEON 2 are quite a bit harder to drive 86 dB/1mW (into 13 ohms...
www.audiosciencereview.com
- Thread Starter
- #20
Actually the +1 dBu is the lowest analog reference level on the RME, using it allows to reduce the amount of digital attenuation if you have too much gain in the amp.
Similar threads
- Poll
