AFAIK only two OEM make components employing variable gain to control analog volume are Ayre and Mola Mola by Hypex (both SS). The former employ such circuit in a $14k preamp and costly integrated amp, the latter in their $20k Makau DAC/preamp.
I have every reason to believe such circuit defines state of the art audio performance for volume control. Gain and noise/distortion are inseparable. The greater the gain the greater is noise and distortion.
If correct, why is such control not universally employed in reference quality components, both OEM and DIY? Is it extremely difficult to properly implement?
I have every reason to believe such circuit defines state of the art audio performance for volume control. Gain and noise/distortion are inseparable. The greater the gain the greater is noise and distortion.
If correct, why is such control not universally employed in reference quality components, both OEM and DIY? Is it extremely difficult to properly implement?
Could be rather dependent on what 'audio performance' means in practice? Sound or numbers? How have you uncovered that Ayre's using variable gain? From what I know of Charles Hansen's writings he uses an expensive Shallco motorized switch operating as a stepped attenuator in his top-end models. Lower end uses FETs as switches.
Ayre and Mola Mola couldn't be more different in audio philosophy - the former uses no GNFB and the latter, as much as humanly possible.
<later> Ah I see a description here - looks like Ayre's doing it open loop : Ayre Acoustics AX-5 integrated amplifier | Stereophile.com. The description's not clear enough to visualize the schematic but if it depends on using now-unobtainium Toshiba FETs that would be a good reason why its not been more widely adopted. CH bought up a huge number of J74s and K389s before they went EOL I understand.
Ayre and Mola Mola couldn't be more different in audio philosophy - the former uses no GNFB and the latter, as much as humanly possible.
<later> Ah I see a description here - looks like Ayre's doing it open loop : Ayre Acoustics AX-5 integrated amplifier | Stereophile.com. The description's not clear enough to visualize the schematic but if it depends on using now-unobtainium Toshiba FETs that would be a good reason why its not been more widely adopted. CH bought up a huge number of J74s and K389s before they went EOL I understand.
Last edited:
Review Ayre's website, specifically their "Reference" preamp circa $14k. They offered this feature for several years.
I refer only to variable gain stage volume controls in line level analog preamps. Does your FB note apply in this case?
If gain = noise and distortion, and it does, then how could any volume control outperform a variable gain circuit? That's my first question. The 2nd is why is it not standard high end feature?
I have no idea how different or similar are Charlie and Bruno's implementations. But they both seem to place it atop the volume control performance pyramid.
(Re. my audio experience: I recently borrowed the $10k Bricasti M1 SE DAC for a couple weeks, employed as follows:
M1 @ 0 dB/maximum output/maximum bit resolution > high resolution analog preamp/op-amp based ladder R volume, which thoroughly outperformed M1 digital volume > power amp.....I am almost afraid to seriously audition Bruno's $20k Mola Mola DAC/preamp w/variable gain volume control.)
I refer only to variable gain stage volume controls in line level analog preamps. Does your FB note apply in this case?
If gain = noise and distortion, and it does, then how could any volume control outperform a variable gain circuit? That's my first question. The 2nd is why is it not standard high end feature?
I have no idea how different or similar are Charlie and Bruno's implementations. But they both seem to place it atop the volume control performance pyramid.
(Re. my audio experience: I recently borrowed the $10k Bricasti M1 SE DAC for a couple weeks, employed as follows:
M1 @ 0 dB/maximum output/maximum bit resolution > high resolution analog preamp/op-amp based ladder R volume, which thoroughly outperformed M1 digital volume > power amp.....I am almost afraid to seriously audition Bruno's $20k Mola Mola DAC/preamp w/variable gain volume control.)
Last edited:
You'll need to say what 'outperform' means to you? I'm guessing you mean in THD+N but not all manufacturers are believers in traditional measurements as guarantors of audio quality. Amongst other audio designers, TVCs are considered to be the best volume controls and they generally have fairly poor THD especially at lower frequencies.
I've seen a preamp schematic of Bruno's for DIY use, I think (but can't be sure) he used opamps (LM4562) in a variable gain configuration with a potentiometer. This would necessarily involve oodles of NFB.
I've seen a preamp schematic of Bruno's for DIY use, I think (but can't be sure) he used opamps (LM4562) in a variable gain configuration with a potentiometer. This would necessarily involve oodles of NFB.
Last edited:
I presume TVC has limited step quantity and step magnitude that does not apply to variable gain volume control. To your point: I appreciate any reference to a reliable AB test, TVC v. variable gain volume. I know of no such AB test, because the latter is so rare. TVC is only compared to volume controls with performance below a variable gain control.
If you find out how Charlie's circuit differs from Bruno's, I'd love to hear about it.
Re. "performance:" The Bricasti M1 @ 0 dB > analog preamp @ -4 dB v. Bricasti @ -12 dB digital > direct to amp. In two AB tests, each with different program, four persons took seconds to prefer the former over the latter (one performed the test, the other three blind).
I wonder how the specs compare: former had maximum digital resolution > 8 dB of preamp gain w/12 dB of attenuation (preferred performance), latter has less digital resolution but preamp and IC removed from the circuit.
Knowing how digital attenuation degrades audio performance, Bricasti strongly recommends matching M1 internal gain settings to system requirements. A variable gain volume control matches gain at every single volume setting, not just within a certain range.
If you find out how Charlie's circuit differs from Bruno's, I'd love to hear about it.
Re. "performance:" The Bricasti M1 @ 0 dB > analog preamp @ -4 dB v. Bricasti @ -12 dB digital > direct to amp. In two AB tests, each with different program, four persons took seconds to prefer the former over the latter (one performed the test, the other three blind).
I wonder how the specs compare: former had maximum digital resolution > 8 dB of preamp gain w/12 dB of attenuation (preferred performance), latter has less digital resolution but preamp and IC removed from the circuit.
Knowing how digital attenuation degrades audio performance, Bricasti strongly recommends matching M1 internal gain settings to system requirements. A variable gain volume control matches gain at every single volume setting, not just within a certain range.
Last edited:
Yes TVCs typically have fairly coarse volume steps, definitely not continuously variable like a pot. Ayre's volume control is also quantized though doubtless in fairly small steps.
I would guess (as no doubt Mola Mola's internal schematics aren't public property) that Bruno's doing something similar to Doug Self (perhaps based on the work of Baxandall). You can read about Self's approaches in 'Small Signal Audio Design' Chapter 9, section 'Active Volume Controls'.
OTOH Charles Hansen looks to be using a FET OTA. So there's a voltage->current stage and the value of resistance chosen as load of the OTA determines the gain. A FET buffer would then follow to maintain constant output impedance.
I would guess (as no doubt Mola Mola's internal schematics aren't public property) that Bruno's doing something similar to Doug Self (perhaps based on the work of Baxandall). You can read about Self's approaches in 'Small Signal Audio Design' Chapter 9, section 'Active Volume Controls'.
OTOH Charles Hansen looks to be using a FET OTA. So there's a voltage->current stage and the value of resistance chosen as load of the OTA determines the gain. A FET buffer would then follow to maintain constant output impedance.
Re. TVC's larger step size v. other volume controls: suppose the TVC causes an SPL other than ideal. The mathematical difference between the ideal SPL and the TVC's nearest SPL (higher or lower) = X dB, which by definition equals a form of easily quantified distortion.
I bet this distortion is ignored by TVC proponents. And again, I know of no TVC proponent to participate in AB test with a variable gain volume control preamp.
I bet this distortion is ignored by TVC proponents. And again, I know of no TVC proponent to participate in AB test with a variable gain volume control preamp.
Last edited:
Fellas, I never heard TVC volume and I would avoid expensive parts if I can do well without. A few thoughts on the subject of volume control / variable gain:
1)Digital attenuation can be sonically neutral, it can be an excellent choice. Even leading mastering engineers agree on this (Bob Katz is one). I would expect 64 bits processing and a good dithering algorithm to guarantee neutrality. I have bookmarked this page but didn´t go through it yet, there might be some useful info: TPDF Dither | Airwindows
2)A combination of digital attenuation and analog gain control might be ideal. You would use some digital attenuation when your analog gain is at the minimum or, for example, when you want to avoid the non-linear range of a DAC (for instance TDA1545 is slightly non linear between -10dBFS and 0dBFS acording to a paper published by the designers of the dac).
3)I realize many of you guys use an analog source. For either analog or digital I am attaching two interesting and useful circuit ideas. One well known here, the Pass D1 analog stage, the other less known, MCI JH600 mike preamp. Mind the variable filters that are formed with the gain R.
4)Flat Phono Preamp Variable Gain
5)Finally, try an LDR for that shunt R? (The one that adjusts the gain.)
(My own I/V stage resembles the Pass D1 and sometimes I use that volume control scheme. I can switch the pot in and out of the circuit and I tend to use digital attenuation as well. I have the impression the TDA1543 dacs sound best when peaks don´t exceed -10dBFS just like the 1545. The gain implemented with LDR might be very nice, I haven´t tried.)
1)Digital attenuation can be sonically neutral, it can be an excellent choice. Even leading mastering engineers agree on this (Bob Katz is one). I would expect 64 bits processing and a good dithering algorithm to guarantee neutrality. I have bookmarked this page but didn´t go through it yet, there might be some useful info: TPDF Dither | Airwindows
2)A combination of digital attenuation and analog gain control might be ideal. You would use some digital attenuation when your analog gain is at the minimum or, for example, when you want to avoid the non-linear range of a DAC (for instance TDA1545 is slightly non linear between -10dBFS and 0dBFS acording to a paper published by the designers of the dac).
3)I realize many of you guys use an analog source. For either analog or digital I am attaching two interesting and useful circuit ideas. One well known here, the Pass D1 analog stage, the other less known, MCI JH600 mike preamp. Mind the variable filters that are formed with the gain R.
4)Flat Phono Preamp Variable Gain
5)Finally, try an LDR for that shunt R? (The one that adjusts the gain.)
(My own I/V stage resembles the Pass D1 and sometimes I use that volume control scheme. I can switch the pot in and out of the circuit and I tend to use digital attenuation as well. I have the impression the TDA1543 dacs sound best when peaks don´t exceed -10dBFS just like the 1545. The gain implemented with LDR might be very nice, I haven´t tried.)
Attachments
Either that or my dead bug style I/V is slightly happier with the reduced current swing, or both.
So far nothing in this thread indicates anything to contradict that variable gain defines state of the art volume control.
Again, in the hope that someone knows: Why do only two known OEMs employ such volume control? Are the two circuits mentioned above exceptionally costly and/or difficult?
I won't say the designer's name because the info might have been confidential. Designer X makes an exceptionally high value low-mid priced DAC with proprietary features. X works on a variable gain volume control, and said it is a difficult circuit to properly implement.
> only two OEM make components employing variable gain
Nearly all "mixers", stage and recording, use variable gain in many stages. Alexandre's second pic is (in related forms) universal in mike-amps, from $100K consoles down to the "$10 mic amp" project.
The level coming off a harpsichord through a dynamic, or a bass drum through a '414, can be VERY different. Gain "must" be controlled. Before clean gain-knob amplifiers, engineers kept pads handy for knocking-down excess signal.
At "this" end of the chain, we "know" what signals we will get. (One thing we know is we will never get more than 2V analog out of a standard digital source.)
And in (good) practice, our in-use hiss level is the hiss in those studio mike-amps (and the mikes, and the studio ambient).
IMHO a simple passive pot can give excellent performance. The "best" will tend to a low-value pot (which may force a buffer before) and a low-hiss amp after that.
Also the simpler vari-gain amplifiers will not cut gain to *zero*, for silent play(?); and most have some minimum hiss level higher than a full-down pot, and perhaps higher than a sharp pot recovery amplifier.
Nearly all "mixers", stage and recording, use variable gain in many stages. Alexandre's second pic is (in related forms) universal in mike-amps, from $100K consoles down to the "$10 mic amp" project.
The level coming off a harpsichord through a dynamic, or a bass drum through a '414, can be VERY different. Gain "must" be controlled. Before clean gain-knob amplifiers, engineers kept pads handy for knocking-down excess signal.
At "this" end of the chain, we "know" what signals we will get. (One thing we know is we will never get more than 2V analog out of a standard digital source.)
And in (good) practice, our in-use hiss level is the hiss in those studio mike-amps (and the mikes, and the studio ambient).
IMHO a simple passive pot can give excellent performance. The "best" will tend to a low-value pot (which may force a buffer before) and a low-hiss amp after that.
Also the simpler vari-gain amplifiers will not cut gain to *zero*, for silent play(?); and most have some minimum hiss level higher than a full-down pot, and perhaps higher than a sharp pot recovery amplifier.
Last edited:
Hi PRR
I learned to love a good minimalist recording, it´s often just two mikes. Be it the orchestra or just a small ensemble or even one guitar. And I desire to extract all the enjoyment contained in it (which is immense and not present in most systems I heard. I seem to get more with every improvement I make to mine).
Best result I got so far is with non oversampling DACs going to a balanced variable gain I/V stage (open loop), to a balanced bridged power amp with low gain (this one is very high feedback, low gain). As for hiss I have none whatsoever, not with my ears on the tweeters.
I need to match most of the components because the circuits are balanced. Could be a PITA for production but for my personal system it´s worth it. The orchestra went from boring to stunning as the system evolved.
That MCI schematic I find interesting, they used a dual transistor similar to the MAT duals for input. The inverting input of 5534 is cascoding the collector. Loop feedback from the outputs to the emitters of the dual. And one resistor (make that an LDR) to control the gain. I think stellar results can be had from this topology if you don´t exaggerate the gain.
I learned to love a good minimalist recording, it´s often just two mikes. Be it the orchestra or just a small ensemble or even one guitar. And I desire to extract all the enjoyment contained in it (which is immense and not present in most systems I heard. I seem to get more with every improvement I make to mine).
Best result I got so far is with non oversampling DACs going to a balanced variable gain I/V stage (open loop), to a balanced bridged power amp with low gain (this one is very high feedback, low gain). As for hiss I have none whatsoever, not with my ears on the tweeters.
I need to match most of the components because the circuits are balanced. Could be a PITA for production but for my personal system it´s worth it. The orchestra went from boring to stunning as the system evolved.
That MCI schematic I find interesting, they used a dual transistor similar to the MAT duals for input. The inverting input of 5534 is cascoding the collector. Loop feedback from the outputs to the emitters of the dual. And one resistor (make that an LDR) to control the gain. I think stellar results can be had from this topology if you don´t exaggerate the gain.
Last edited:
psychoacoustic noise shaped dithered digital volume control?
flat TPDF is easiest but shaped dither is better as the last digital processing step before analog conversion
flat TPDF is easiest but shaped dither is better as the last digital processing step before analog conversion
Self and Putzeys, Baxandall inspired, active volume controls offer a bit more than very high performances :
First, they use linear pots so in stereo versions, their tracks are well matched and they are less expensive than non-linear pots.
To me, the gain changing with the knob variation gives a much nicer progression and better feeling than the so-called log pots.
The circuit is easy to implement, you can breadboard it, the cost is very low,
I have made my own version which has non-inverting, inverting and balanced outputs :
Active volume control with balanced output using 2 op-amps
Used op-amps have been NE5532s, LM4562s and even a TL084 just to test how noisy it can be (nothing audible my loudspealers).
The Bruno Putzeys's version
The G word: Demo project - A balanced volume controller | EDN
shows a most interesting property. The only load of the pot cursor is the inverse input of an op-amp (at the right of the circuit) which has a huge impedance. The input DC bias of an LM4562 being very low, the current in the contact between the pot track and the cursor is almost null, avoiding any non-linearity coming from it and preserving the circuit to become scratchy with time.
I think a very low load of a pot cursor makes it more state of the art than a very expensive component.
- Status
- Not open for further replies.