Passive balanced volume control using a single attenuator? Possible?

[ewildgoose]

Someone was recently trying to talk me around the idea of a balanced passive volume control using only a single attenuator. I think the idea was basically to use the attenuator to wire up the hot and cold signals rather than sinking each to ground using two attenuators.

Can anyone comment on this circuit? A sketch of what would be required would be nice?

Also, I need this to work with the output of a computer soundcard which has a pseudo balanced output, and for example seems to respond badly to inadvertent tying of -ve and ground lines together... Is this design likely to work well in this case?

For those wondering why bother... Well, I have 8 balanced output channels, and I can't afford a decent 16 channel attenuator, whereas an 8 channel one seems more reasonable... Any other suggestions for 8 balanced channels of volume control gratefully received...

[Peter Daniel]

Here's an example:

[OliverD]

If I remember correctly, this principle was used in Pass Lab's "D1" D/A-Converter. A service manual is available for download at their website.

[Variac]

A comment I have seen regarding this is that most of the resistors required are quite low value- some less than an ohm, so are hard to find in a decent mix of values. A search might turn up some of these other threads. I'm not sure what to search on though..

[Pan]

Someone, a well known manufacturer, raised the issue of the input stage of the amp "looses" the damping factor when the resistors to GND is not left in the circuit.

I don´t think this is a problem, the balanced input see the shunt resistance between the + and - and in my mind a bleeding off of stored energy could as well take that path.

Another solution I have used is that first you use a series element and than you use two resistors to GND as usual. You scale these resistor for the max volume you need, I never need full ouput from source, especially running balanced.

This way you have a relatively low impedance to GND and then you use the stepped attenuator to shunt the signal between the + and -.

Try to use as low value of resistors as possible as long as your soundcard can handle it without loosing performance.

/Peter

[ewildgoose]

Pan, thanks. I think this is the design that was being discussed.

I think the idea was two resistors in series, followed by the attenuator which joins +ve and -ve to each other and then on down to ground. Then I seem to remember that the sketch had two more resistors in series after the attenuator, but I'm not quite sure why?

Like you I won't need a full range signal, typically I am running about -40db down from max output for example.

I was going to get one of those 10K DACT stepped things. Could someone suggest suitable starting values for the other resistors in this scheme?

Also, rough finger in the air estimate here, but is this design going to be worth doing at all rather than simply using an unbalanced output into the DACT? Rest of the system is an RME card (50 ohm output imp), then preferably 1-2 feet of cable (yeah, I know, will try to minimise this), then the DACT-in-a-box, then straight down to the ZapPulse amp input (perhaps 3-9 inches since two amps, input impedence 17K ohm)

Any thoughts?

[Pan]

Typically a passive sound best with as low resistance as posible, I don´t know what kind of load your card would like, but I´d guess the current capability isn´t all that great.

A series resistor for a shunt attenuator like this is bets kept at 1k or so, but that means an typicall output impedance of 1.1k or less including the shunt resistors. This might be to low, but again it might work. 10k is some kind of "standard" and will probably be ok, but lower is better.

One thing about the DACT I don´t like is that the resistors is soldered in a loooong serie (series atenuator). I prefer one only resistor to GND in the shunt position. You can always buy the elma switch and make your own balanced shunt attenuator, that I would recommend.

When using passives, it´s always wise to go balanced due to the noise issue in higher impedance circuits, especially since you have balanced out from source and I believe the latest Zapulse has balanced inputs as well, right?

/Peter

[Pan]

Peter Daniel,

checked the amp in the link...

The opamp at the ouput, is it a DC servo or an "error correction amp" working over the audio range?

/Peter

[ewildgoose]

Hi Peter, thanks for the advice. So I will start with a 1K resistor and see how I get on. What are the likely signs of lack of current capacity though? Distortion increase? Failing that what I think you are saying is move up to perhaps 2K, 5K or 10K?

There is a picture of the card itself here if that gives any clues to opamps used and potential current output capacity:
http://www.wildgooses.com/downloads/HDSP9632mod.jpg

(FWIW: the notes on the jpeg were a friends suggestions on improvements to the standard card)

I agree about your comments on the series resistors not being ideal, but it's beyond my time and abilities right now to construct a better alternative, especially assuming an 8 way switch. (Anyone want to quote me to build one for me?)

Thanks for your advice, will see how it turns out. I assume that in practice this design will be sonically pretty similar to buying an extra 8 way DACT thing and doing it the obvious way with two attenuators per channel?

Zappulse has balanced inputs, and the source is all balanced, so yes, this seems to be the preferred option (and the source is a PC with all the obvious associated noise).

Cheers

[Pan]

Guess you´ll have to try and see how it sounds, adjusting from there. Remember that if you use a 10k series attenuator as a shunt with a low, say 1-5k fixed series resistor, then it is possible the values are to big in the attenuator. In practice this means you´ll end up with very few volumesettings as you run out of attenuation range.

Not familiar with the components on your card but the advices you allready have been given seems good if your in tweaking mode .

Using the approach we discuss here will reduce the "bad" with a series attenuator, since the fixed three resistor voltage divider are good quality and soldered. Even if a component is in paralell with the signal it effects the signal in the end, but less so in this instance.

A variation of this for a two channel rig is to use the balanced three resistor voltage divider and then add a plastic pot, a good high performance passive for those on a budget.

/Peter