Budget passive preamp with stepped attenuator?

[Blumlein 88]

The main thing I wondered is the step size. 256 .5 db steps is 128 db which is more than enough. Even .25 db steps for 64 db would work for almost anyone's system.

I have made my own in several configurations. 1 db steps are as good as a continuous control. So smaller steps are only helpful for matching two different devices precisely in volume. Which normally is not any benefit. One I made had 7 resistors in something like a ladder DAC configuration. 128 steps that were plenty small. I just had 7 switches one for each resistor to ground. I used it enough I could just flip the switches and get the volume I wanted. For someone not familiar it would have been baffling to use. These modern ones take care of all that for you using relays.

 

[sonder]

The main thing I wondered is the step size. 256 .5 db steps is 128 db which is more than enough. Even .25 db steps for 64 db would work for almost anyone's system.

I have made my own in several configurations. 1 db steps are as good as a continuous control. So smaller steps are only helpful for matching two different devices precisely in volume. Which normally is not any benefit. One I made had 7 resistors in something like a ladder DAC configuration. 128 steps that were plenty small. I just had 7 switches one for each resistor to ground. I used it enough I could just flip the switches and get the volume I wanted. For someone not familiar it would have been baffling to use. These modern ones take care of all that for you using relays.

The full PDF datasheet is available, it's a texas instruments ADC0804 board.

N.B. it's not an ADC device though, it's setup so that an analog potentiometer sends an analog voltage to control the 8-bit ADC0804 controller which opens and closes relays on precision resistors .

 

[sonder]

I've personally found that a decently low noise DAC (110 dB DR at 2 V) paired with a NC252MP results in inaudible noise but I don't run very sensitive speakers.

Ran the same calc's you kindly provided earlier for a ~1.5uv DAC (I have 2, one at 1.5 and one at 1.3), a presumably near transparent relay stepped attenuator in the middle, and the nc252mp - does this look about right?

Residual noise from DAC: 1.5uv
Amplified noise via power amp: 1.5 x 10^(25.6/20) = 28.58191uv
Total residual noise from amplified DAC output and amplifier = sqrt (28.58191uv ^2 + 50uv^2) = 57.59275 uV

Voltage at DAC output at -20 dB = 2 x 10^(-20/20) = 0.2 V
Voltage at amplifier output at -20 dB = 0.2 x 10^(25.6/20) = 3.81092 V
Dynamic range at amplifier output at -20 dB = 20 x log(3.81092 x 10^6 / 57.59275) = 96.41324 dB

With stepped relay (assuming it's perfect, looks to be the same resistor setup as a pre90 for example)

Attenuated residual noise from DAC output at -20 dB (analog) = 1.5 x 10^(-20/20) = 0.15 uV
Amplified residual noise from DAC output at -20 dB (analog) = 0.15 x 10^(25.6/20) = 2.85819 uV
Total residual noise at -20 dB (analog) = sqrt (2.85819^2 + 50^2) = 50.0816 uV

Dynamic range at amplifier output at -20 dB (analog) = 20 x log(3.81092 x 10^6 / 50.0816 ) = 97.627 dB

My takeaway here, being that adding a resistor based relay stepped with two signal paths in the mix should perform equivalently at worst, and basically add in a transparent firewall and tactile experience, would you agree?

 

[mdsimon2]

Ran the same calc's you kindly provided earlier for a ~1.5uv DAC (I have 2, one at 1.5 and one at 1.3), a presumably near transparent relay stepped attenuator in the middle, and the nc252mp - does this look about right?

Residual noise from DAC: 1.5uv
Amplified noise via power amp: 1.5 x 10^(25.6/20) = 28.58191uv
Total residual noise from amplified DAC output and amplifier = sqrt (28.58191uv ^2 + 50uv^2) = 57.59275 uV

Voltage at DAC output at -20 dB = 2 x 10^(-20/20) = 0.2 V
Voltage at amplifier output at -20 dB = 0.2 x 10^(25.6/20) = 3.81092 V
Dynamic range at amplifier output at -20 dB = 20 x log(3.81092 x 10^6 / 57.59275) = 96.41324 dB

With stepped relay (assuming it's perfect, looks to be the same resistor setup as a pre90 for example)

Attenuated residual noise from DAC output at -20 dB (analog) = 1.5 x 10^(-20/20) = 0.15 uV
Amplified residual noise from DAC output at -20 dB (analog) = 0.15 x 10^(25.6/20) = 2.85819 uV
Total residual noise at -20 dB (analog) = sqrt (2.85819^2 + 50^2) = 50.0816 uV

Dynamic range at amplifier output at -20 dB (analog) = 20 x log(3.81092 x 10^6 / 50.0816 ) = 97.627 dB

My takeaway here, being that adding a resistor based relay stepped with two signal paths in the mix should perform equivalently at worst, and basically add in a transparent firewall and tactile experience, would you agree?

I agree with your numbers. I wouldn't think a stepped volume control would cause any noise issues vs a digital volume control but I don't think it will make an audible difference either. At this point you will almost be certainly be limited by the noise floor of the recording.

A tactile knob is a nice feature and I understand the peace of mind of having an analog volume control upstream of the amplifier.

Michael

 

[Apesbrain]

Another option:

 

[solderdude]

What'll happen if I mix in a 50kA impedance pot instead of a 10k? (flow E30II in pure dac -> pot -> NC256MP)

edit: from what I can tell that's 2ohms output from DAC, 10k ohms input to amp - struggling to find any resources to calculate the optimal potentiometer k value

context:

• 10K Jenson https://www.ebay.com/itm/273591977617
• 50K ALP 09 4 channel https://www.ebay.com/itm/253646372699
• 50K ALPS 27 https://www.ebay.com/itm/253818627052

edit 2: screw it, I asked gpt4, result below:



specific calculations:

Simplifying:
Vout = 2 VRMS x (10k ohm / 10,002 ohm)
Vout = 1.9996 VRMS (approximately)

Simplifying:
Vout = 2 VRMS x (50k ohm / 50,002 ohm)
Vout = 1.9998 VRMS (approximately)

Yeah, this is totally irrelevant.
What is relevant is the source resistance the amp will see.
This will vary between 2ohm and 5kohm (between 0dB and -6dB) and then gradually become lower again the more attenuation is used.
The load the DAC will see will vary a little but always be irrelevant.

The background noise the amp will have may be lowest at -max attenuation and 0dB attenuation and may be a bit worse at other settings.

For the 50k attenuator it will vary between 2ohm and 25k.

So the 10k is a much better choice and perhaps even a 10k or 5k attenuator will be technically superior.
Also make sure the box the attenuator is mounted in is screened and connected to common(ground).
The higher the attenuator resistance the higher the sensitivity is to pick up nasties.

 

[sonder]

Yeah, this is totally irrelevant.
What is relevant is the source resistance the amp will see.
This will vary between 2ohm and 5kohm (between 0dB and -6dB) and then gradually become lower again the more attenuation is used.
The load the DAC will see will vary a little but always be irrelevant.

The background noise the amp will have may be lowest at -max attenuation and 0dB attenuation and may be a bit worse at other settings.

For the 50k attenuator it will vary between 2ohm and 25k.

So the 10k is a much better choice and perhaps even a 10k or 5k attenuator will be technically superior.

Cheers, I wound up getting them all, but I'm most looking forward to the relay resistor stepped attenuator, it'll click, but should offer similar performance to the la4 or pre90 (or some 4-5 figure devices) for a fraction of the cost.

Minor collation note, they've swapped the pot in the little bear mc2 to a 20k now, if you get a new one from source.

 

[solderdude]

It'll offer the same performance as an RK27 in practice
But you can create presets, ensure max. volume limiting, add remote control etc. Use up-down buttons or choose something with a nice tactile feel etc.
Better channel balance in the lowest volume control settings also of course.

Better SQ or 'numbers' alas will not happen.

 

[egellings]

Is there any? I just want volume control for RCA cables without channel imbalance.

Thoe things should not be called preamps, since there are no amplifying devices in them. Coin some other word to call 'em.

 

[Deleted member 58865]

Thoe things should not be called preamps, since there are no amplifying devices in them. Coin some other word to call 'em.

The correct term is a potentiometer or a stepped attenuator.

 

[Deleted member 58865]

@solderdude

Will https://www.ebay.com/itm/133922947794 be practically transparent?

I want to put this 10k-ohm stepped attenuator between Topping D10s with 20-ohm output impedance and AIYIMA A07 that has unknown input impedance.

Where should the volume pot of AIYIMA A07 be for minimal noise? 9'o clock? 12'o clock? 3'o clock? Perhaps, the position of the pot doesn't matter much as long as there is no perceptible channel imbalance?

 

[sonder]

Yes, that smd dact type at 10k is great, or the nobsound if you want to go unstepped Alps rk27, the rest / casing / RCA are all the same.

 

[Deleted member 58865]

Also make sure the box the attenuator is mounted in is screened and connected to common(ground).
The higher the attenuator resistance the higher the sensitivity is to pick up nasties.

If a stepped attenuator box is grounded by RCA cables, is it connected to the common ground?

 

[Deleted member 58865]

if you want to go unstepped Alps rk27, the rest / casing / RCA

I already suffer from channel imbalance of the volume pot on AIYIMA A07. I don't want another piece of equipment that is subject to channel imbalance.
Channel imbalance annoys the hell out of me.

 

[Blumlein 88]

@solderdude

Will https://www.ebay.com/itm/133922947794 be practically transparent?

I want to put this 10k-ohm stepped attenuator between Topping D10s with 20-ohm output impedance and AIYIMA A07 that has unknown input impedance.

Where should the volume pot of AIYIMA A07 be for minimal noise? 9'o clock? 12'o clock? 3'o clock? Perhaps, the position of the pot doesn't matter much as long as there is no perceptible channel imbalance?

Your questions indicate some basic misunderstandings I'm afraid. -6 db or usually 12 o'clock position will give highest output impedance of the pot. It will be one fourth the pot value. 2500 ohms for a 10,000 ohm pot for instance. All other positions have less output impedance.

Minimal noise? Honestly that just doesn't enter into using these attenuators. For instance 2500 ohms will cause a thermal noise level of -121 db referenced to 1 volt. That is the worst it will be. Amplifier gain will amplify that, but remember at all other volume settings it will be lower than this.

Putting them as close to the amp input as possible is good to prevent treble roll off. All this has been covered one way or another.

 

[Deleted member 58865]

@Blumlein 88 Are you saying I should just stay away from 12'o clock to minimize noise from the volume pot? Or, are you saying the volume pot noise is negligible?

Someone else said I should turn the amp volume pot to 3'o clock to maximize amp's SNR and reduce the volume in a (stepped) attenuator between DAC and amp. I don't know how accurate the advice is. As far as I know, if I turn the amp volume pot to 3'o clock and reduce the volume in a stepped attenuator between DAC and amp, SNR still decreases.

 

[Deleted member 58865]

Does any volume potentiometer have 0dB gain at 3'o clock? It seems Schiit Sys potentiometer does. I want to use a stepped attenuator instead of the amp volume pot which has channel imbalance. It seems Schiit Sys has minimal channel imbalance at 3'o clock.

 

[Blumlein 88]

@Blumlein 88 Are you saying I should just stay away from 12'o clock to minimize noise from the volume pot? Or, are you saying the volume pot noise is negligible?

Someone else said I should turn the amp volume pot to 3'o clock to maximize amp's SNR and reduce the volume in a (stepped) attenuator between DAC and amp. I don't know how accurate the advice is. As far as I know, if I turn the amp volume pot to 3'o clock and reduce the volume in a stepped attenuator between DAC and amp, SNR still decreases.

Mainly that volume pot noise is negligible. The rest of your advice all depends upon the particulars of the gear in use. And for the most part, don't use too long a cable leaving the attenuator and there isn't much to worry about. Your worrying about very tiny levels of noise that aren't a big deal. And the best possible result will be dependent upon your particular string of gear. Chances are even less than optimal it won't change what you hear, but it all depends upon various details.

 

[Blumlein 88]

Does any volume potentiometer have 0dB gain at 3'o clock? It seems Schiit Sys potentiometer does. I want to use a stepped attenuator instead of the amp volume pot which has channel imbalance. It seems Schiit Sys has minimal channel imbalance at 3'o clock.

A passive pot or switched resistors only has 0 db gain at max volume. It has no active volume and can only reduce the signal. Stepped/switched units should have nigh on perfect channel balance. However some pots are so close it hardly matters. Some aren't.

You also cannot say the SYS has minimal imbalance at 3 o'clock. The one Amir measured did, the next one may have it somewhere else or have nearly none. Pots will vary a bit in that respect.

 

[solderdude]

@solderdude

Will https://www.ebay.com/itm/133922947794 be practically transparent?

I want to put this 10k-ohm stepped attenuator between Topping D10s with 20-ohm output impedance and AIYIMA A07 that has unknown input impedance.

Where should the volume pot of AIYIMA A07 be for minimal noise? 9'o clock? 12'o clock? 3'o clock? Perhaps, the position of the pot doesn't matter much as long as there is no perceptible channel imbalance?

3dB steps are kind of 'coarse' as a volume control but it will be fine for attenuating to a certain attenuation level.

Audible it will be just as 'transparent' as an expensive one. Dunno about longevity.
Don't worry about noise. It will be below any audible thresholds.

If a stepped attenuator box is grounded by RCA cables, is it connected to the common ground?

Most likely ... yes but have no idea how this is done here.