About stepped attenuators (i.e. am i hearing things ?)

[MRC01]

... a friend of mine passed me a series type 20k log mechanical attenuator built using Holco resistors
I put it in a box with RCA ins and outs like a passive preamp and connected it in front of an old Bryston 0.4B preamp that I had around and ... my goodness.
By setting the preamp volume to maximum and controlling the level with the attenuator the sound became much clearer and more detailed
The difference between a knife that doesn't cut well and a scalpel
Question... is it autosuggestion or is there an abyss between attenuators and common potentiometers?

I too noticed an improvement in sound clarity when I replaced an active preamp with a passive attenuator. I wouldn't call it an abyss, but the difference was definitely audible if subtle. I don't think it was just channel balance. Not all attenuators are made the same way. I built mine so that each volume setting was a different pair of resistors, one in series, one in parallel. It did not combine resistors like some other attenuators do. Metal film resistors are among the cleanest, where their total noise is close to the theoretical minimum Nyquist/Johnson noise. Other types of resistors, including those in potentiometers, have additional noise.

The only limitation of a passive preamp is that you want to maintain 10:1 impedance ratios. Its input impedance should be at least 10x higher than the output impedance of the upstream source, and its output impedance should be at least 10x lower than the input impedance of the downstream amp. Most solid state sources have output impedances under 200 ohms, most solid state amps have input impedances > 20 kOhm, so it works out fine. Ladder attenuators present the same input impedance at every volume setting, but have an output impedance that varies. It's highest at -6 dB where both resistors are equal, so the output impedance is 1/4 of the input impedance. That is, a 5 kOhm attenuator has a highest (worst-case) output impedance of 1,250 ohms at -6 dB.

 

[gino1961]

I have made several DC decoupling AC coupling adapters for those car audio high level outputs that float at 1/2 VCC (~7 volts DC.) and adapted to a RCA level termination . They work fantastic and in a sound comparison I could not detect if they where in or out of circuit. Very nice they are.

i am not sure to understand what i mean is that there is a lot of discussion about the best coupling caps but almost never i see lab tests
parts like Mundorf Jensen Audyn Solen etc. I would love to see measurements

 

[Doodski]

i am not sure to understand what i mean is that there is a lot of discussion about the best coupling caps but almost never i see lab tests
parts like Mundorf Jensen Audyn Solen etc. I would love to see measurements

I only used a quality Japanese brand that cost like less than a dollar each. That is all that is required. Capacitive reactance is not better or changed by more expensive snazzy caps.

 

[gino1961]

I too noticed an improvement in sound clarity when I replaced an active preamp with a passive attenuator. I wouldn't call it an abyss, but the difference was definitely audible if subtle. I don't think it was just channel balance. Not all attenuators are made the same way. I built mine so that each volume setting was a different pair of resistors, one in series, one in parallel. It did not combine resistors like some other attenuators do. Metal film resistors are among the cleanest, where their total noise is close to the theoretical minimum Nyquist/Johnson noise. Other types of resistors, including those in potentiometers, have additional noise.

hi thank you very interesting but a little tricky for me I like zero noise a lot Noise is never music
i have to say that the attenuator even if the mechanism was not Suisse watch like was very clean indeed Strings brass percussions exploded in the room
A very sharp sound but not etched Very defined I loved it and the preamp proved to be not that bad after all

 

[gino1961]

I only used a quality Japanese brand that cost like less than a dollar each. That is all that is required. Capacitive reactance is not better or changed by more expensive snazzy caps.

very interesting Speaking of coupling caps some designers use a combination of film and electrolytic caps even in mic preamps where the signal are very low
i have to try

 

[Doodski]

very interesting Speaking of coupling caps some designers use a combination of film and electrolytic caps even in mic preamps where the signal are very low
i have to try

I never was required to try different cap designs because as a techy I was mostly concerned with, "Why is this box broken?"

 

[Ixnay]

I only used a quality Japanese brand that cost like less than a dollar each. That is all that is required. Capacitive reactance is not better or changed by more expensive snazzy caps.

People frequently tell you that they use something that is superior, but they won't tell you what it is.
But this place is kinda kooky anyway.

 

[Doodski]

But this place is kinda kooky anyway.

LoL... I always think of KooK as in the surfer interpretation of the term. It's a rough term that way.

 

[MRC01]

hi thank you very interesting but a little tricky for me I like zero noise a lot Noise is never music
i have to say that the attenuator even if the mechanism was not Suisse watch like was very clean indeed Strings brass percussions exploded in the room
A very sharp sound but not etched Very defined I loved it and the preamp proved to be not that bad after all

Zero noise (and distortion) is the unattainable goal. A good passive gets you halfway there, since distortion is zero. But noise can never actually be zero with all those little electrons jiggling around.
More on passives here.

 

[gino1961]

I too noticed an improvement in sound clarity when I replaced an active preamp with a passive attenuator. I wouldn't call it an abyss, but the difference was definitely audible if subtle. I don't think it was just channel balance. Not all attenuators are made the same way. I built mine so that each volume setting was a different pair of resistors, one in series, one in parallel. It did not combine resistors like some other attenuators do. Metal film resistors are among the cleanest, where their total noise is close to the theoretical minimum Nyquist/Johnson noise. Other types of resistors, including those in potentiometers, have additional noise.

i need to start from an off the shelf attenuator pot or stepped I can only solder some wirings My attenuator was of the series type
So the noise of the resistors sum up But the feeling was that noise were minimal anyway Holco resistors are very good quality

The only limitation of a passive preamp is that you want to maintain 10:1 impedance ratios. Its input impedance should be at least 10x higher than the output impedance of the upstream source, and its output impedance should be at least 10x lower than the input impedance of the downstream amp. Most solid state sources have output impedances under 200 ohms, most solid state amps have input impedances > 20 kOhm, so it works out fine. Ladder attenuators present the same input impedance at every volume setting, but have an output impedance that varies. It's highest at -6 dB where both resistors are equal, so the output impedance is 1/4 of the input impedance.
That is, a 5 kOhm attenuator has a highest (worst-case) output impedance of 1,250 ohms at -6 dB.

thank you very much I read about the topic I have seen a popular 10Klog value like in this old good one passive pre

Placette Audio Remote Volume Control passive preamplifier Specifications

Sidebar: Specifications Description: Remote-controlled, 125-step passive volume control. Input impedance: 9k ohms.Dimensions: 7.25" (184mm) W by 1.5" (3.8mm) H by 9.5" (241mm) D. Weight: 4 lbs (2kg).Serial number of unit reviewed: Not noted.Price: $1000. Approximate number of dealers: 1 (sold...

www.stereophile.com

i would go even lower at this point Like 5 or 2,5Kohm considering that i am not use high Zout sources
I see two option with the attenuator before the preamp or between the preamp and the power amp

 

[MRC01]

Thank you very much I read about the topic I have seen a popular 10Klog value like in this old good one passive pre

Placette Audio Remote Volume Control passive preamplifier Specifications

Sidebar: Specifications Description: Remote-controlled, 125-step passive volume control. Input impedance: 9k ohms.Dimensions: 7.25" (184mm) W by 1.5" (3.8mm) H by 9.5" (241mm) D. Weight: 4 lbs (2kg).Serial number of unit reviewed: Not noted.Price: $1000. Approximate number of dealers: 1 (sold...

www.stereophile.com

i would go even lower at this point Like 5 or 2,5Kohm considering that i am not use high Zout sources
I see two option with the attenuator before the preamp or between the preamp and the power amp
I am willing to sacrifice the bryston for the experiment I could extract the pot solder some jumpers and make it a buffer with a Vgain of 3 times

Vishay/Dale are also good quality parts (that's what I used). If your other devices are all solid state, 5k is a good choice. It has the flexibility of compatibility with common input & output impedances, with noise well below audible levels. A 2.5k is requires more strict impedance matching for devices with no real advantage because the 5k noise was already more than low enough.

My attenuator wasn't upstream nor downstream from the preamp, but replaced the preamp entirely. I built it with several selectable inputs.

 

[gino1961]

Vishay/Dale are also good quality parts (that's what I used). If your other devices are all solid state, 5k is a good choice. It has the flexibility of compatibility with common input & output impedances, with noise well below audible levels. A 2.5k is requires more strict impedance matching for devices with no real advantage because the 5k noise was already more than low enough.

thank you very much again and no more question on pot impedance

My attenuator wasn't upstream nor downstream from the preamp, but replaced the preamp entirely. I built it with several selectable inputs.

i see I will try soon the 5k option Mine was 20K a little high value for a passive
i will try with a cheap pot and then if it works with a much better one

 

[somebodyelse]

i am not sure to understand what i mean is that there is a lot of discussion about the best coupling caps but almost never i see lab tests
parts like Mundorf Jensen Audyn Solen etc. I would love to see measurements

https://www.audiosciencereview.com/forum/index.php?threads/capacitor-distortion.40917/

 

[gino1961]

https://www.audiosciencereview.com/forum/index.php?threads/capacitor-distortion.40917/

thank you for the link I dont know why square wave test is not more popular I like it a lot

 

[LSPhil]

If your preamplifier has a noise of 3mV and your music has 30mV then you have an S/N=20dB, your signal increases to 1.4V resulting in S/N=53.4dB.

You simply improved your S/N by setting your preamp to max. Not more than that.

 

[MRC01]

There's no benefit to a passive at full/max volume, but nobody actually listens at full volume. When we turn the volume down to the low to mid settings actually used for listening, a passive can have lower noise / higher SNR than an active preamp. The site linked above calculates specific examples, but the overall point is that the SNR of active preamps typically drops 1:1 as you reduce the volume. But passives don't behave that way, and can give higher SNR at the low to mid settings actually used for listening.

For just example, consider a 2V signal being attenuated 30 dB for listening, so the signal is reduced to 63 mV. An active preamp will have an SNR about 30 dB lower than its measurement at max volume. So if it measures 110 dB SNR at max, it will be about 80 dB at this volume setting.

A 10k passive attenuator at -30 dB will have 2 resistors of 316 and 9,684 ohms. The output impdance is these in parallel, which is about 306 ohms. Thermal/Nquist noise at 306 ohms at room temp and 20 k bandwidth is 3.15e-7 Volts, with a signal of 63 mV, the SNR is 106 dB. The SNR is more than 20 dB better than the active preamp.

 

[AuricGoldfinger]

...My attenuator was of the series type
So the noise of the resistors sum up But the feeling was that noise were minimal anyway Holco resistors are very good quality

Mine are series-types in 5k Ohms, as in the circuit shown in the Stepped Volume Control paper by W.Marshall Leach Jr. I used Susumu RR, 0.5% 25ppm, surface mount 0805 packages, because they are practical. At low volumes, the noise is the equivalent of a 4950 Ohm resistor in the signal path. This value reduces as the volume control is rotated clockwise. It’s not a problem. It is transparent and absolutely silent.

In fact, I couldn’t distinguish it from the noise floor of the analyzer, ie 5-digits right of the decimal.

About the “noise of the resistors sum up”:
Even if a non-series type stepped attenuator is used, it still has the same noise level if it’s the same value. This is because it has to use the same resistance values for equivalent attenuation. Whether it be one or 50 resistors makes no difference in this regard. Many high voltage test probes use resistors in series rather than a single, and this does not degrade them.

 

[MRC01]

Mine are series-types in 5k Ohms, as in the circuit shown in the Stepped Volume Control paper by W.Marshall Leach Jr. I used Susumu RR, 0.5% 25ppm, surface mount 0805 packages, because they are practical. At low volumes, the noise is the equivalent of a 4950 Ohm resistor in the signal path. This value reduces as the volume control is rotated clockwise

There are different designs for passive attenuators, so I'll clarify that with a ladder attenuator (voltage divider) the effective impedance for computing noise is its output impedance. It has the same input impedance (5k, 10k etc.) at every setting but its output impedance varies with volume position. Worst case highest is 1/4 the input impedance so 1.25k for a 5k attenuator, and lower at other settings.

 

[AnalogSteph]

For just example, consider a 2V signal being attenuated 30 dB for listening, so the signal is reduced to 63 mV. An active preamp will have an SNR about 30 dB lower than its measurement at max volume. So if it measures 110 dB SNR at max, it will be about 80 dB at this volume setting.

Even worse, preamp SNR is generally specified referred to maximum output, which more often than not is 6 to 9 Vrms. With most power amps, you generally don't need more than about 1.5 Vrms for full rated output, at times not even that. So that's potentially 12-16 dB wasted right there, assuming that the noise floor at the settings required is still the same as at normal listening level (which is not necessarily true). This makes a 124-128 dB(A) spec a fair bit less impressive than it looks at first.

My rule of thumb for output noise floor in a fairly decent preamp (straight 16.5 dB gain, 5532 class opamp) is about 4 µV(A). About -111 dB(A) ref. 1.5 V. With power amp gain of 26, 29.5 or 33.5 dB following, that's 80, 119 or 189 µV(A) at the speaker output, respectively (or -91, -87.5 or -83.5 dB(A) ref. 2.83 V). The first two are almost certainly inaudible on normal sensitivity speakers (85-88 dB / 2.83 V / 1 m). Things will be noticeably different if your speakers are big and horny with the sensitivity to match.

The above is how even DACs of relatively modest dynamic range specifications by modern standards have managed to beat most ordinary traditional preamps - the unbalanced output may only be good for 117-120 dB(A), but that's referred to 2 V, equivalent to 2.8-2.0 µV(A) worth of output noise.

Now obviously, you can do much the same in an analog preamp... for example, a PGA2320 on +/-15 V rails is good for 8.5 Vrms of output and up to +31.5 dB of gain while its typical noise floor is a relatively unexciting 10.8 µV(A). Giving this an output stage with a gain of -14 dB (1/5) results in much more favorable values.

A 10k passive attenuator at -30 dB will have 2 resistors of 316 and 9,684 ohms. The output impdance is these in parallel, which is about 306 ohms. Thermal/Nquist noise at 306 ohms at room temp and 20 k bandwidth is 3.15e-7 Volts, with a signal of 63 mV, the SNR is 106 dB. The SNR is more than 20 dB better than the active preamp.

Now, mind you, it is not common for power amps to have quite the sort of input noise density of a microphone preamp (and much less so if a balanced input is involved), so they would not be able to take full advantage of this. I would estimate a good one might reach about a 2.5 nV/√(Hz) input voltage noise density, that's 0.35 µV in 20 kHz. Add your thermal noise, and we're at 0.47 µV, which still is a lot lower than 4 µV. (About -130 dB below 1.5 Vrms.) Even if we're talking a relatively mediocre 15 nV/√(Hz) input (e.g. LM3886 and the like if memory serves), that's still only 2.1 µV (and maybe 1.7 µV(A)).

 

[AuricGoldfinger]

There are different designs for passive attenuators, so I'll clarify that with a ladder attenuator (voltage divider) the effective impedance for computing noise is its output impedance. It has the same input impedance (5k, 10k etc.) at every setting but its output impedance varies with volume position. Worst case highest is 1/4 the input impedance so 1.25k for a 5k attenuator, and lower at other settings.

Indeed. I just logged in to fix and addendum that reply, but you’re quicker to the draw.