• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required as is 20 years of participation in forums (not all true). Come here to have fun, be ready to be teased and not take online life too seriously. We now measure and review equipment for free! Click here for details.

Reducing the output level from a DAC

Joined
Mar 30, 2019
Messages
19
Likes
21
#1
Hello!

I am currently facing a challenge with my setup and would be grateful if I can get an advice on the issue. The post came out quite long. That's because I tried to describe the situation as accurately as possible. Coming from an engineering background (albeit a software one), I understand that there are no absolute solutions and unless the context of the situation is known, a question usually results in an "it depends" answer ;)

My question is about finding an effective way to lower the output level of my DAC, without sacrificing quality. I did a research on the subject and I think I have found a workable solution, but would be grateful if I can get an advice on whether I am on the right track here.

What I have is the Khadas Tone Board, connected to a JDS Atom amp. Connected to the Atom I have a pair of JBL 305 active monitors (connected to the pre-outs of the Atom) and Alessandro MS-Pro headphones. The MS-Pros are a modified version of the Grado RS-1 and are manufactured by Grado for Alessandro.

Like other Grado headphones, the MS-Pro is really easy to drive. In fact, so easy, that in order to be able to listen to them at all (without damaging my hearing), I usually have to bring down the volume all the way down. This is true on any device, e.g. on my phone, computer, etc.

In this particular case, this means that I have to bring down the volume of the Khadas board (on my PC) to just 20-30% or less and also to have the volume on the Atom at about 7-8 o'clock. It stays basically just a few millimeters above the point at which the device turns on. If I turn the dial even a little bit back, I go to that state of the amp, where the sound of the headphones gets distorted (e.g. you can hear something on the right, but almost nothing on the left).

Listening to the headphones with this setup is quite problematic. On one hand, the volume cannot be finely controlled at all. Even a slight rotation of the dial increases or decreases it way too much. One way to counter it is to set the volume on the computer to a very low value, e.g. 7%. This allows me to operate the Atom volume dial properly. However, at this level, it should be resulting in a significant loss of bit-depth. Whether this is objectively degrading the sound quality in a perceptible way or not, is another question. To me it seems so.

For reference, I had a similar issue with the JBL 305s before purchasing the atom, although not at that extreme level. I had to turn their sensitivity down (they have a switch, which lets me drop it from -10db to +4) and their amp at volume of 4/10, while keeping the software volume of the Khadas at 70% on the computer. Now with the Atom, this problem is solved and I can finally get the Khadas at 100% on the computer and I can also set the monitors, to, say 5/10.

Unless I am missing something here, it seems that to fix the situation, I have to somehow reduce the output from the Khadas Tone Board. I did some research on what can be done and came with three possible solutions:

1. One that was pretty obvious from the start was to look for a way to reduce the volume on the Khadas board itself. There is a thread on the Khadas forums about that. The good news is that controlling the volume in hardware is, in fact, possible. There is (although a very rudimentary) demo of how to do it and back in November they wrote "We will make an accessory for this hardware volume control, which will probably be completed next month." Unfortunately, as of the time of writing of this post, there is still no info on when/if such an accessory will be released. Also, the mod in question seems to be far from being applicable to general daily use. Also, it has not been stated how the volume level is being controlled. E.g. is it a "proper" hardware control, or is it similar to how the Windows volume control works and the bit-depth gets reduced.

2. To use In-Line Attenuators, like this one from Rothwell. However, from what I have found from different reviews, they tend to distort the sound and also they are not cheap. For quite a similar price I can go for the third option, which is:

3. To use a passive attenuating pre-amp. After doing a bit of research, I came down with two options, which seem to be able to solve the problem and both promise to be pretty nearly transparent (I understand that with a passive attenuator some distortion may be expected):

- JDS Labs Ol Switcher;
- JBL M-Patch 2.

So, to sum things up: Am I missing something in the big picture? Is there another, more simple solution, which never occurred to me? Also, if someone has experience with the Ol Switcher or the M-Patch 2, can you share your thoughts on the device?

Thanks in advance for any help provided!
 
Last edited:

RayDunzl

Major Contributor
Central Scrutinizer
Joined
Mar 9, 2016
Messages
8,518
Likes
5,971
Location
Riverview FL
#2
The essential difference between in-line attenuators and a passive (potentiometer based) volume pot is adjustability.

Both put the signal across a pair of resistances, and split the input voltage.

Part of the signal voltage is "lost" across R1, depending on the ratio of the resistance of the resistors.



1553994615905.png


1553994188772.png
1553994780595.png


Signal on pin 1, ground on 3, the output is 2, and moving the wiper splits the resistance, changing the ratio of the resistance of R1 vs R2.

I wouldn't pay too much attention to the "resistors distort the sound" crew.
 
Last edited:

gvl

Major Contributor
Joined
Mar 16, 2018
Messages
1,625
Likes
1,290
Location
SoCal
#4
Turn down volume on the playback software?
+1. Good playback software will allow to a) control volume in the application using 32 bits or more b) control volume on the DAC using 32 bits (ESS volume control), both methods should not cause any truncation way down even with 24 bit material. For PCM at least. If you're listening to DSD then maybe in fact consider passive attenuators, they should not cause any "distortion" and most likely will be transparent. http://www.hlabs.com/products/attenuators/
 

Willem

Major Contributor
Joined
Jan 8, 2019
Messages
1,097
Likes
1,543
#5
I had a similar issue with my desktop system. My source is a desktop PC into an ODAC usb DAC with a 2.1V output. For amplification I use a refurbished Quad 405-2 power amplifier with a 0.5 V input sensitivity. So I wanted two things: a convenient volume control and some attenuation. Both are provided by an Emotiva Control Freak. This is/was what is called a passive pre amp, but of course it is nothing more than a variable resistor. In the end I added some inline attenuators from Parts Express as well, to reduce the output to a level where I could no longer accidentally feed a dangerously loud signal into my amplifier and blow up my speakers (that almost happened). This also avoids using the Control Freak near maximum attenuation because such variable resistors tend to suffer from channel imbalance at low levels. Of course the latter could also have been achieved in software, but I decided that a hard wired solution was safer to avoid accidental high output. Sound quality of the set up is excellent, with Harbeth P3ESR speakers. In your case, the JDS Labs switcher looks like the most convenient solution, perhapos with added inline attenuators.
 

solderdude

Major Contributor
Joined
Jul 21, 2018
Messages
4,297
Likes
7,981
Location
The Neverlands
#6
The essential difference between in-line attenuators and a passive (potentiometer based) volume pot is adjustability.

Both put the signal across a pair of resistances, and split the input voltage.

Part of the signal voltage is "lost" across R1, depending on the ratio of the resistance of the resistors.



View attachment 24324

View attachment 24323 View attachment 24325

Signal on pin 1, ground on 3, the output is 2, and moving the wiper splits the resistance, changing the ratio of the resistance of R1 vs R2.

I wouldn't pay too much attention to the "resistors distort the sound" crew.
When pin 1 is the left pin on the potmeter shown above the volume control will be maximum on the CCW setting and minimum on the fully clockwise setting.
1 and 3 numbers in the schematic should be switched in this case.
For the potmeter: 3 is input, 2 = output and 1 is ground.
 

daftcombo

Major Contributor
Forum Donor
Joined
Feb 5, 2019
Messages
1,833
Likes
1,538
#7
Listening to the headphones with this setup is quite problematic. On one hand, the volume cannot be finely controlled at all. Even a slight rotation of the dial increases or decreases it way too much. One way to counter it is to set the volume on the computer to a very low value, e.g. 7%. This allows me to operate the Atom volume dial properly. However, at this level, it should be resulting in a significant loss of bit-depth. Whether this is objectively degrading the sound quality in a perceptible way or not, is another question. To me it seems so.
Hi,
One thing is not clear: do you want to listen to music in a particular application, like Foobar2000 or JRiver? Or do you want to listen to all the sounds & music coming from the computer?
In the first case, you could probably just use the internal volume control (32bit or 64bit) of those applications.
 
Joined
Mar 30, 2019
Messages
19
Likes
21
#8
Thank you for the replies!
Hi,
One thing is not clear: do you want to listen to music in a particular application, like Foobar2000 or JRiver? Or do you want to listen to all the sounds & music coming from the computer?
In the first case, you could probably just use the internal volume control (32bit or 64bit) of those applications.
I mainly use Foobar, but sometimes I also use other sources, like YouTube/Spotify.

It may be that I am overestimating the distortion caused from loss of bit-depth. I was under the impression that even if the DAC supports 24 or 32-bits, reducing the volume digitally (e.g. via Foobar or the Windows volume control) by a large amount (e.g. setting it to 15%), results in a significant compression of the dynamic range, which leads to worsened quality. When reducing it from both Foobar and the volume control, I will be applying such compression twice, using different algorithms. Unfortunately, I cannot measure the effects of that and cannot say for sure what the loss in quality is.
 

RayDunzl

Major Contributor
Central Scrutinizer
Joined
Mar 9, 2016
Messages
8,518
Likes
5,971
Location
Riverview FL
#9
It may be that I am overestimating the distortion caused from loss of bit-depth. I was under the impression that even if the DAC supports 24 or 32-bits, reducing the volume digitally (e.g. via Foobar or the Windows volume control) by a large amount (e.g. setting it to 15%), results in a significant compression of the dynamic range, which leads to worsened quality.
15% = -27.7dB on my Win10 and the on-board Realtek sound chip...

-27.7dB is between 4 and 5 bits.

24bit attenuation leaves you with 19 or 20 bits, 32bit attenuation leaves you with 27 or 28 bits.

Right click the slider to swap between % and dB

1554067664033.png
1554067702686.png
 
Last edited:

RayDunzl

Major Contributor
Central Scrutinizer
Joined
Mar 9, 2016
Messages
8,518
Likes
5,971
Location
Riverview FL
#10
When reducing it from both Foobar and the volume control, I will be applying such compression twice, using different algorithms.
The basic digital attenuation algorithm is to multiply every sample by the same suitable fraction.

-6dB = multiply every sample value by 0.5, for example.

For the attenuation in the prior post, -27.7dB, multiply each sample value by 0.04121

1554068122916.png


Various dither types may be injected, but other than that, I don't know what kind of fancy math might be involved.
 
Joined
Mar 30, 2019
Messages
19
Likes
21
#11
15% = -27.7dB on my Win10 and the on-board Realtek sound chip...

-27.7dB is between 4 and 5 bits.

24bit attenuation leaves you with 19 or 20 bits, 32bit attenuation leaves you with 27 or 28.

Right click the slider to swap between % and dB

View attachment 24360 View attachment 24361
Wow, thanks a lot! This is a very useful feature. I think this pretty much answers my question.
 

RayDunzl

Major Contributor
Central Scrutinizer
Joined
Mar 9, 2016
Messages
8,518
Likes
5,971
Location
Riverview FL
#12
I wouldn't worry about it if you don't hear a problem.

1554069009357.png
 

HammerSandwich

Addicted to Fun and Learning
Forum Donor
Joined
Nov 22, 2018
Messages
636
Likes
677
#13
The most likely reason to want analog attenuation is to reduce the source's noise. This happens when the combination of the source's output level + the amp's gain + headphone's/speaker's sensitivity is too high. IOW, don't worry if you're not hearing hiss.
 

Similar threads

Top Bottom