When is a lower voltage DAC better?

[ZolaIII]

Ufff V*2 = 6 dB, 6 dB = 1 bit PCM, +12 dB psy is considered twice as loud. Most DACs have peak performance in - 6 to - 12 dB range from their full output from where it falls relatively to expectations - 6 dB/ V:2. You don't need more than 17 bits and that's listening very loud 82 dB - 102 dB peaks in the most sensitive hearing area (1~4 KHz), even less as highest peeks won't be in that area. As you listen on lower SPL you need less bits. Goal is that beats - SINAD stays above the listening levels SPL/input sensitivity (for headphones/lines) and everything will be fine.

 

[Stereodude]

...Most DACs have peak performance in - 6 to - 12 dB range from their full output from where it falls relatively to expectations...

How are you defining "peak performance"?

What output metrics are better?

 

[ZolaIII]

How are you defining "peak performance"?

What output metrics are better?

Highest SINAD.

 

[Stereodude]

Highest SINAD.

The DACs that do the best in Amir's reviews having the highest SINAD don't have the behavior you are describing.

 

[ZolaIII]

The DACs that do the best in Amir's reviews having the highest SINAD don't have the behavior you are describing.

No they are better autended not to give output above the level they do it best. I stated most including pretty good ones ever measured hire.

 

[jed1]

The value 75 is not the percentage of the maximum level. If it is coded like other SMSL DACs like the D-6s and the DL200, it is -11.5 dBr or 2.5 dBV - around 1.33 V - and not 75% or 4 V.

Also, the internal data path of the DAC is 32 bit so there is no loss of bits before 48 dB attenuation for 24 bit files and 96 dB attenuation for 16 bit files.

You can also just reduce the input gain of the KALIs.

Thanks @JIW . Where can I find documentation or formula for the volume value vs dBr? If I need attenuation > 10dB (it seems like that), I would make inline attenuators and insert them between DAC and powered monitors.

 

[amirm]

Amplifier gain situation is a mess. While many DACs have moved to higher output voltages in the last 10 years, amplifiers are still being built as if their input is from last century. Recently I have tried to push the gain down form 29 to 32 dB which is common to 25 dB when interacting with companies.

For now, as noted, there is nothing wrong with turning down the volume in the DAC. I routinely test in this mode to get the voltage down to 4 volts for example.

 

[Bleib]

For now, as noted, there is nothing wrong with turning down the volume in the DAC. I routinely test in this mode to get the voltage down to 4 volts for example.

I still think this is a problem for a lot of people as they won't know how much they have to lower the volume to be closer to an ideal target.
For non-enthusiasts I always recommend DACs with the normal 2/4v out

 

[JIW]

Thanks @JIW . Where can I find documentation or formula for the volume value vs dBr? If I need attenuation > 10dB (it seems like that), I would make inline attenuators and insert them between DAC and powered monitors.

There is no documentation from SMSL as far as I can tell.

I did some rudimentary testing when I got my DL200 on 23 December.

I got mine today and rudimentarily measured the steps in the volume control using headphones, a 1 kHz tone and an SPL meter app on my phone.

The ESS chip can go from 0 dB to -127.5 dB in 0.5 dB steps for each channel (ES9039Q2M datasheet, page 15). The DL200's volume control has only 100 steps from 0 to 99, however. Thus, it may well differ in some way.

Volume 00 is mute. From 99 down to 25 is in 0.5 dB steps, i.e. 25 is -37.0 dB, 50 is -24.5 dB and 75 is -12.0 dB. 20 is 5 dB lower than 25, i.e. 20 is -42.0 dB. Below 20 is unclear but 01 is about 28 dB below 20, i.e. 01 is about -70 dB.

The line output is closest to 2 V RMS RCA and 4 V RMS TRS at volume 95, i.e. -2.0 dB, giving 1.9858 V RMS and 3.9716 V RMS, respectively.

Maybe @SMSL-Mandy can give the exact values.

On 24 February, @nanook measured the attenuation of the volume control on his D-6s.

Volume Control on the D-6s measured at 1kHz and Full-Scale sinewave on XLR outputs
SN: 33212330199

View attachment 351948

View attachment 351949

Edit: I have double-checked several times, the "missing step" is real and it fortunately applies to both channels - nothing to worry about though.

I confirmed these measurements for my DL200 on 27 February.

I did some similar testing on my DL200 when I first got it (See quote below). I did some more testing today. My results are in line with yours. In particular, the 88 to 89 step does not change the level.

Curiously, in my original raw measurements, the level difference between 80 and 90 is exactly 4.5 dB instead of the expected 5.0 dB for 10 0.5 dB steps. From 80 to 70, from 70 to 60, from 60 to 50, from 50 to 40 and from 40 to 30 are all exactly 5.0 dB as expected. However, I did not think too much about it and thought it due to measurement uncertainty.

In summary:
Volume in 0.5 dB steps down to 25 except from 89 to 88 where there is no change.
Volume in 1.0 dB steps from 25 to 19
Volume in 1.5 steps from 19 to 01
Volume 00 is Mute.

About the need for attenuation, the benefit of external attenuation is a reduction in noise. Since the DO100 Pro has a claimed SNR of 131 dB, it will be no lower than 121 dB when the output is reduced by 10 dB. This is probably a good bit below that of the speakers. You can easily test this by checking if turning on the DAC increases the noise from the speakers. If not, there is no benefit to external attenuation.

 

[JIW]

Amplifier gain situation is a mess. While many DACs have moved to higher output voltages in the last 10 years, amplifiers are still being built as if their input is from last century. Recently I have tried to push the gain down form 29 to 32 dB which is common to 25 dB when interacting with companies.

For now, as noted, there is nothing wrong with turning down the volume in the DAC. I routinely test in this mode to get the voltage down to 4 volts for example.

The Fosi actually has a gain of 26 dB and reaches its maximum power for an input of around 1.33 V RMS which is a little more than 3 dB below 2 V. However it also has the same gain for the XLR connection which is equivalent to 32 dB for RCA since the voltage differential is twice as large.

For an amplifier capable of 2x200 W at 4 Ohm and 2x100 W at 8 Ohm, i.e. sqrt(800) V or 29 dBV, 26 dB is reasonable for a 2 V source leaving 3 dB for headroom. If it is twice as powerful, gain should be 3 dB higher to maintain the headroom and if it is half as powerful, gain should be 3 dB lower. For a 4 V source, gain should be 6 dB lower.

 

[jed1]

There is no documentation from SMSL as far as I can tell.

I did some rudimentary testing when I got my DL200 on 23 December.


On 24 February, @nanook measured the attenuation of the volume control on his D-6s.


I confirmed these measurements for my DL200 on 27 February.


In summary:
Volume in 0.5 dB steps down to 25 except from 89 to 88 where there is no change.
Volume in 1.0 dB steps from 25 to 19
Volume in 1.5 steps from 19 to 01
Volume 00 is Mute.

About the need for attenuation, the benefit of external attenuation is a reduction in noise. Since the DO100 Pro has a claimed SNR of 131 dB, it will be no lower than 121 dB when the output is reduced by 10 dB. This is probably a good bit below that of the speakers. You can easily test this by checking if turning on the DAC increases the noise from the speakers. If not, there is no benefit to external attenuation.

I would go with balanced L-pad design as below;

https://audiosciencereview.com/foru...der-for-active-diy-speaker.28709/post-1000093

If I use a same resistance value for all 4 resistors, then I will get attenuation of 9.5dB so I would call it very close to my need.

 

[KSTR]

If I use a same resistance value for all 4 resistors, then I will get attenuation of 9.5dB so I would call it very close to my need.

I'm getting at -6dB (0.5x), as each leg is a 2:1 voltage divider (assuming sufficiently low source and high load impedances).

For a factor of 0.33x you need half the R value in the ground leg resistors, or two in parallel. Like 5 pcs of 2k resistors, 1% or better, metal film. That gives a 3k load per leg which most any output stage should handle fine.
Output impedance is not super low, and thus comes with a slight noise penalty which may or may not be relevant, of course.

 

[JIW]

I would go with balanced L-pad design as below;

https://audiosciencereview.com/foru...der-for-active-diy-speaker.28709/post-1000093

If I use a same resistance value for all 4 resistors, then I will get attenuation of 9.5dB so I would call it very close to my need.

I don't know much about that but as I told you, you can also turn down the speakers so that they attenuate the input by about 10 dB. This is the simplest solution and what it is for and I don't know why you would not utilize it.

 

[jed1]

I don't know much about that but as I told you, you can also turn down the speakers so that they attenuate the input by about 10 dB. This is the simplest solution and what it is for and I don't know why you would not utilize it.

Thanks. I lowered the speaker volumes so that I don't hear any self-noise at tweeters. I believe that's more or less optimum gain structure.

 

[rcstevensonaz]

Thanks. I lowered the speaker volumes so that I don't hear any self-noise at tweeters. I believe that's more or less optimum gain structure.

You may want to consider putting your speakers at their lowest gain level and at their maximum dB attenuation (if you have both), which may be well below when you stop hearing a hiss.

Assuming an audio chain of [DAC >>> Preamp >>> Active Speakers], my interpretation of this article from Sweetwater on Gain Staging Like a Pro is that your configuration approach would try to start your signal (at DAC) at the highest level (before clipping) and the amplification/attenuation at the Active Speakers (i.e., the final amp) as low as possible; and then make adjustments as needed so that you gain the maximum amount of volume knob movement on the device you are using to control volume.

Simplified steps:

• Set DAC initially at its highest amplification setting (leaving it in DAC mode for "100% volume")
• Set Speakers at lowest amplification and attenuation setting
• Test range of motion on your volume knob

If the range of volume is still limited (i.e., too loud too quickly), then use the following steps:

1. If the DAC has multiple gain level settings, lower the the gain level and retest.
2. If still too little range of motion, then either of the following:
   • Put the DAC into PreAmp mode, and use its digital volume control to bring down the output level as necessary
   • Keep the DAC at high gain and add an attenuation pad (e.g., JTS MA-12 or Shure A15AS) just before the speakers

However, if the range of motion on volume is no longer limited (i.e., not getting to maximum volume), then:

1. Start backing off on the attenuation setting on the speaker
2. If still not getting maximum desired volume, then increase to next higher gain at maximum attention (and go back to step 1 above reducing attenuation)

 

[olds1959special]

The Fosi ZA3 has the same gain for XLR and RCA input - 26 dB or 20x - but XLR levels tend to be twice as high. This is not the most clever design. It reaches maximum output at an input of around 1.4 V RMS or 3 dBV. The RCA output of SU-1 is 2 V RMS or 6 dBV but the XLR output of the DL200 is 5 V RMS or 14 dBV.

Unless you have a noise issue with the RCA connection, use the DL200 with RCA connection and set the volume control to 88 which is -5 dBr or 3 dBV output. Then set the volume control on the amplifier so that is as loud as you would ever want it or you hear distortion. After that, control volume via the DL200. If you have a noise issue with RCA, use XLR but set volume to 74 which is -11 dBr instead and do as above.

Thanks. I set my dac at 74 and adjust the volume with the amp. I'm using the XLR inputs. All is good now.

 

[jed1]

I'm getting at -6dB (0.5x), as each leg is a 2:1 voltage divider (assuming sufficiently low source and high load impedances).

For a factor of 0.33x you need half the R value in the ground leg resistors, or two in parallel. Like 5 pcs of 2k resistors, 1% or better, metal film. That gives a 3k load per leg which most any output stage should handle fine.
Output impedance is not super low, and thus comes with a slight noise penalty which may or may not be relevant, of course.

What would you recommend for -15dB attenuation with balanced L-pad configuration or anything better one? DO100 Pro's output impedance is 100 ohm. I cannot find input impedance value for Kali, though.

 

[terryforsythe]

A DAC with lower output voltage probably has lower gain, which should mean less noise all else being equal. However, if your amp has selectable gain (mine do), and you end up needing higher gain on the amp to compensate for the lower output voltage of the DAC, you will get more noise out of the amplifier, which defeats the purpose of using a low voltage DAC.

 

[jed1]

One factor that may be relevant to some cases including mine, input signal level needs to be optimum (matching) for a powered speaker with built-in ADC / DSP. Unless an active speaker has digital inputs like (spdif, AES), analog input needs to be digitized at the front end with ADC. When I was reading instructions for Kali, input gain is recommended around 0dB (neutral position) for good performance. When gains are very different from DAC output and powered DSP speaker input, it will be challenging to optimize gain structure.

I like how WiiM provides output level options - 500mV rms to 2V rms (though I don't use WiiM's DAC functions).

 

[KSTR]

What would you recommend for -15dB attenuation with balanced L-pad configuration or anything better one? DO100 Pro's output impedance is 100 ohm. I cannot find input impedance value for Kali, though.

-15dB means a 5.6-to-1 divider, so something like 2k and 470R would work as good as anything else which gives a 2k..3k load and the approximate attenuation. Channel matching is the only thing that is important.