BPPBP - Bruno Putzey's Purist Balanced Preamp (well a balanced volume control really)

I can think of two distortion mechanisms that would be left uncompensated for when a series resistor is connected to the linear pot in the feedback network, these being the thermal effects due to the current density in the pot resistance tracks, and the voltage dependence of the track resistance.
Just how strong the effects of these nonlinearities on the distortion would be in the current case is not easy to answer without calculations and/or measurements.
Inadvertently, Bruno provided the distortion figures with a logarithmic pot in the feedback network in his original article, see his comment on the link below:
The G word: Demo project - A balanced volume controller | EDN
Since the pot used was 'a cheap 9mm "car stereo" pot', it may also have been a carbon track model, which is probably the worst case volume control device for this preamp.
I think that a speculation should be allowed at this point that a good quality conductive plastics or a cermet linear pot with a sensibly chosen series resistor for the purpose of improving the volume control law at low settings would in terms of distortion be much better than the pot with which - inadvertently - the measurements reported in the original article were made. And it were exactly these figures that got me thinking about building this fine preamp in the first place.

Regards,
Braca
 
I can think of two distortion mechanisms that would be left uncompensated for when a series resistor is connected to the linear pot in the feedback network, these being the thermal effects due to the current density in the pot resistance tracks, and the voltage dependence of the track resistance.
Just how strong the effects of these nonlinearities on the distortion would be in the current case is not easy to answer without calculations and/or measurements.
That makes me wonder if the distortions B.Putzeys is talking about are due to tempco and due to voltage dependance?
Inadvertently, Bruno provided the distortion figures with a logarithmic pot in the feedback network in his original article, see his comment on the link below:
I remeber the comment, pity the new data for the lin pot was not published. Or has it?
The G word: Demo project - A balanced volume controller | EDN
Since the pot used was 'a cheap 9mm "car stereo" pot', it may also have been a carbon track model, which is probably the worst case volume control device for this preamp.
I think that a speculation should be allowed at this point that a good quality conductive plastics or a cermet linear pot with a sensibly chosen series resistor for the purpose of improving the volume control law at low settings would in terms of distortion be much better than the pot with which - inadvertently - the measurements reported in the original article were made. And it were exactly these figures that got me thinking about building this fine preamp in the first place.

Regards,
Braca
Are you saying that it is a probable mismatch of tempco and voltage dependance of a separate resistor to the vol pot track, that risks an added distortion of the signal as it passes through the 3rd stage?
Resistor distortion effects are sometimes quoted down around -170dB. Way below what this pre-amp does.
But if there is a mismatch of parameters, maybe the "cancellation" of distortion you refer to, would no longer be cancelled.
 
Thanks for taking the time to explain that.

But I don't understand what your added diodes are doing.

A resistor is by definition a linear element, meaning twice as much voltage giving twice as much current.
Non linear means a non linear relation between voltage and current for whatever reason (like heat, voltage dependant track material, etc).
By adding diodes in my example I have just achieved that nonlinearity.
And although my example is very extreme, even then does the feedback topology give a completely distortion free gain.
That is what Bruno was trying to explain.

Hans
 
Under the assumption that the third section would amplify most of the time, I would agree with your observation.
However, and at least in my situation, this is not the case. A preamp with this modification has been a permanent part of my stereo system for the last twelve monthts, attached to a DAC and a RIAA preamp. The preamp drives a power amp with 26dB gain.

Hi Braca

it is quite obvious that your situation screams for a ladder network, either linear or logarithmic.
A pot is not meant to be used in the first 3% of its range, as in your even modified situation.
I do not know the reasons why you left the ladder network solution.
It gives endless flexibility without having to modify a single component on the PCB leaving all goodies intact.

Hans
 
Changing the pot from 10k to 5k, would reduce generated noise from the last stage by -2.5dB and going to a 1K pot would even reduce this noise further with -6.7dB.
However, the two stages before the last stage are also producing noise adding to the noise of the last stage.
For a 5K pot in the 0dB and in the -20dB position, overall noise will go down by resp. -1.4dB and -0.7dB, not a big deal.
A 1K pot in the 0dB and -20dB position will produce resp. -3.1dB and -1.4dB less overall noise compared to the 10K pot.
Be aware that the last stage also has the task of driving the external load, but 1K is still well within the specs of the LM4562 Op-Amp and worth considering.

Hans

Thanks for the reply Hans! The reason I ask is that I'm looking for good quality pots with the best feel (for me that normally means they use proper race bearings in their mechanics) durability and tracking specs. Widens the options if there's variability in value.
 
Referring to Bruno's Mola-Mola blog (Volume control sores paragraph), he considered a high-precision P&G pot for his commercial preamp, but since there was no motorized version of it, he found a linear relay-switched resistor ladder solution for the feedback network. The latter is almost certainly a proprietary one - at least I'm not aware of such a device being offered, both commercially and in the "open hardware" spaces.

Regards,
Braca

Two things to take into account regarding his decisions which don't necessarily apply to the DIYer:

1) Precise, motorised Penny and Giles ARE available but in fader form - that might not be commercially acceptable, using a fader for volume, but it is certainly available for any individual to use. It doesn't even need to be put on the front, can be hidden behind a front panel with a flap..

Here's a brochure (I've no idea if this is their best spec motorised fader..) : http://www.cda-proaudio.com/Audio_Video_Controllers/PDF/PGFM8000_brochure_sept11(EN).pdf

2) Due to the construction of a Penny and Giles pot, they can be taken apart. It's not inconceivable for the DIYer to replace the central shaft with a longer one to introduce motor control on the back. The whole thing apears to be held together using a circlip. That's not the kind of thing people consider to be commercially viable though, requiring fiddly manual modification..

So, there are different options for the diyer, still using a precision motorised part.
 
Hi Braca

it is quite obvious that your situation screams for a ladder network, either linear or logarithmic.
A pot is not meant to be used in the first 3% of its range, as in your even modified situation.
I do not know the reasons why you left the ladder network solution.
It gives endless flexibility without having to modify a single component on the PCB leaving all goodies intact.

Hans


Hans,

Any chance you would be interested in offering a group buy or sharing your ladder solution pcb design that you published as a side to Bruno's article.
 
It's a stepped attenuator though which is not good for this design.

But the Cello the mechnical part of the attenuator looks like this (it's practically home-made anyway!) :

An externally hosted image should be here but it was not working when we last tested it.



and there's no reason why the same scheme couldn't be engineered to fit a good quality pot with a little engineering of the right pot... Would create that stepped attentuator feel on a pot.
 

I just went back through the thread to quote others as to why...

... only to see that the first discussion about it was with you! Post #83 or there abouts. Have you forgotten?

And then what's the point? Stepped attenuators are used due to sound quality improvements of resistors v a wiper track with the signal passing directly through it (across it, whatever correct language should be used). That has no meaning here where what is required is a linear divider ratio across the control in a feedback circuit. Why would you bother going to the effort of matching the individual resistors (or paying through the nose for someone else to do it to the right spec) when a linear pot does everything you need electrically already with no extra efforts or cash? AND you will also have to deal with the downsides of a stepped pot - that it only has steps and is not continuously variable! There is no real upside that I can see.

And then most commercial offerings seem to be shunt types - no linearity there.

If your only reason for wanting a stepped attenuator is for the stepped "feel" then again that has nothing to do with the electrical stepped nature but is only about notches in a piece of metal - that can be applied in other ways.

If it's a hang-up from an audiophile mindset then please take a cold shower and come back to this project afresh! The audiophile reasons for stepped attenuators do not apply to this circuit.

Hans' stepped ladder is all about implementing remote control.
 
Last edited:
Hi Braca

it is quite obvious that your situation screams for a ladder network, either linear or logarithmic.
A pot is not meant to be used in the first 3% of its range, as in your even modified situation.
I do not know the reasons why you left the ladder network solution.
It gives endless flexibility without having to modify a single component on the PCB leaving all goodies intact.

Hans

Hello Hans,
As you know, I wasn't quite satisfied with 16 steps of 2db each, and since a solution with more steps wasn't available, I decided to stay with the modified preamp and pot for the time being. My time schedule in that period was also a factor.
But was also the sentence "Drive an amplifier circuit with a few kilo-ohms at your peril" of Bruno's in the original article that took me a bit aback. He referred to a log pot between the 2nd and 3rd stages, but it also applies to a stepped attenuator.
Now I'm building a preamp with a stepped volume control for a friend of mine, and he's chosen the 64-step RelaiXed Passive of Jos vanEindhoven. I've already finished and tested the volume controller (a few modifications were necessary), and the performance is excellent.
However, my ideal solution, i.e. a linear ladder network in the feedback path, is still not on the horizon. I tried to design one of my own along the R-2R concept, but after the first unsuccessful attempts on the paper I gave it a rest.

Regards,
Braca
 
I just went back through the thread to quote others as to why...

... only to see that the first discussion about it was with you! Post #83 or there abouts. Have you forgotten?

Post 83 is about wiper lift off, nothing to do with stepped attenuators
And then what's the point? Stepped attenuators are used due to sound quality improvements of resistors v a wiper track with the signal passing directly through it (across it, whatever correct language should be used).
Therein lies your mistake. In 99.0% of applications there is no measurable or audible difference with a stepped attenuator vs a pot.
That has no meaning here where what is required is a linear divider ratio across the control in a feedback circuit. Why would you bother going to the effort of matching the individual resistors (or paying through the nose for someone else to do it to the right spec) when a linear pot does everything you need electrically already with no extra efforts or cash?
In my case I need defined steps, preferable 2dB. I have to be able to adjust 7 channels by the same amount. An edge case, but one where a stepped attenuator or relay switched network is measurably superior.

AND you will also have to deal with the downsides of a stepped pot - that it only has steps and is not continuously variable! There is no real upside that I can see.
Your use case not mine.
And then most commercial offerings seem to be shunt types - no linearity there.
You've clearly not looked very hard.
If your only reason for wanting a stepped attenuator is for the stepped "feel" then again that has nothing to do with the electrical stepped nature but is only about notches in a piece of metal - that can be applied in other ways.
I wasn't the one asking about feel. I just noted that the best audio control from a tactile perspective was the Cello one.
If it's a hang-up from an audiophile mindset then please take a cold shower and come back to this project afresh! The audiophile reasons for stepped attenuators do not apply to this circuit.
I never said they did, but you said
It's a stepped attenuator though which is not good for this design.
and have not explained anything BAD about them other than control resolution, which can be fixed with more steps.
Hans' stepped ladder is all about implementing remote control.
How do you figure that out? You can remote that no more than you can a pot. The encoder solders to the board. You have read http://linearaudio.net/sites/linearaudio.net/files/hp bruno potmeter 3.pdf
 
Hello Hans,
As you know, I wasn't quite satisfied with 16 steps of 2db each, and since a solution with more steps wasn't available, I decided to stay with the modified preamp and pot for the time being. My time schedule in that period was also a factor.However, my ideal solution, i.e. a linear ladder network in the feedback path, is still not on the horizon. I tried to design one of my own along the R-2R concept, but after the first unsuccessful attempts on the paper I gave it a rest.

Regards,
Braca

Hi Braca,
Thanks for your reply.

As mentioned before, driving a R-2R is more than twice as complex as a logarithmic ladder network and has no further benefits at all.
So in this case "less is more".
I have no problem in designing a PCB for a ladder network with more steps of whatever stepsize ( I still advise 2dB steps) and making the Gerbers available if a substantial amount of people are interested in this.
But I do not want to be involved in supply, group buys or what else.

However, for 16 steps a relatively simple BCD encoder could be used.
In case more steps are needed, I see 2 options:

1) Leave the 16 steps encoder as it was, but add a separate toggle switch to operate a relay giving -16dB extra attenuation.
This can be achieved by reducing the value of the feedback resistor value around the output Op-Amp by a factor 0.16, all to be done on the same small PCB.
With 2dB steps, you will now have two overlapping ranges of 30 dB giving in total 46dB control range, which should be enough for any application.

2) Use a rotary encoder instead and drive a counter going from 0 to 24.
In case 1dB steps are preferred, the counter could go from 0 to 48.
Control cicuitry will become a little bit more complex, but a rotary encoder is very affordable compared to a BCD rotary encoder and gives the "feeling" of a pot.
Optionally even a display could be added to tell the position of the vol control.
The other advantage is that when using digital circuitry, a make before break circuit can be added, to fully eliminate glitches when changing the volume.


Regards,
Hans
 
Post 83 is about wiper lift off, nothing to do with stepped attenuators
Therein lies your mistake. In 99.0% of applications there is no measurable or audible difference with a stepped attenuator vs a pot.
In my case I need defined steps, preferable 2dB. I have to be able to adjust 7 channels by the same amount. An edge case, but one where a stepped attenuator or relay switched network is measurably superior.

Your use case not mine.
You've clearly not looked very hard.
I wasn't the one asking about feel. I just noted that the best audio control from a tactile perspective was the Cello one.
I never said they did, but you said and have not explained anything BAD about them other than control resolution, which can be fixed with more steps.
How do you figure that out? You can remote that no more than you can a pot. The encoder solders to the board. You have read http://linearaudio.net/sites/linearaudio.net/files/hp bruno potmeter 3.pdf

Oh my god! Calm down mate! I could tell you were just aiming for some kind of argument from your abrupt "Why not?" post.

Therein lies your mistake.

For most of your replies you just need to learn to read properly. And your view of the world is no more true than anyone elses regarding why people choose stepped attenuators. Plenty of people on DIYaudio going though subjective listening tests and choosing stepped attenuators too.

I never said they did, but you said

Hahaha! God, that's so childish... if you can understand the finer points of English you will notice I've used the word "If" in a number of places because I have no idea what your thinking is..

Billshruv, what is the point in this exchange? What exactly is it that you want to prove? You really don't have to wait to pounce on someone's post - you can just plainly post of your own volition any thoughts which come to mind.
 
Thought I'd post this up - nice to link this project to Bruno's thought processes on eliminating ground back in 2011 :

Like the UcD module, the NC has two feedback traces coming directly off the connector so any resistance between the negative terminal and circuit ground is canceled by the feedback loop. This is why the negative speaker terminal is labeled "LS-", not "ground". In our schematics, the differential feedback nets are not labeled "feedback" and "ground" but "feedback-pos" and "feedback-neg" and they're treated absolutely equally. In fact, the whole modulator and feedback circuit is differential. That doesn't mean the signals are symmetrical with respect to ground, but that ground potential is no part of the equation. Only the difference between signal pairs counts, whether or not any of them is held at ground potential at some point. I'll probably write an article explaining this style of circuit design for Linear Audio. I think many people would be helped by an explanation of how to eliminate ground from your signal path altogether even whilst using it to return current.
 
Would a linear version of the linked stepped attenuator work? (10K version??)

Kikitronic -- DIY Audio Kits: Attenuator for volume control

Thanks, Rene
Hi Rene,

Of course will a stepped attenuator work perfectly.
But do not opt for a linear version, keep it logarithmic.
The reason for a linear pot in Bruno's design was purely to cancel the non-linear properties of the track of a potmeter.

The dual use of the word linear is confusing in this context, one meaning that twice as much voltage does not result in twice as much current, the other one meaning that the slider in the mid position divides the impedance values in two equal parts (no matter how non linear these parts may be).

But in many real life situations a linear pot will mean that you will only use the first few percent's of the linear pot, making fine adjustment clumsy and not getting a good balance in volume between the two channels, forcing you to modify the PCB as Braca did to reduce overall gain.
A switched logarithmic pot with 24 positions and equipped with high grade resistor's, does not suffer from non linear track pot issues so the reason to go linear is no longer there.

In case of 2dB per step, you will have a control rage of 48dB, and with a listening level around -20dB, the stepped attenuator will be somewhere in it's mid position with excellent balance between channels.
A very good solution, though probably not very cheap.

Hans