[Steve Eddy]

While I've been aware of shunt attenuators for some time, I never gave them much thought because I use input transformers which demand a constant load impedance to perform their best. However a post on another forum got me thinking more about them and their claimed advantages and I came to realize that the advantages of the shunt attenuator are based on a myth and thought I'd write a little something about it here.

The claimed advantage of the shunt attenuator over the series/shunt attenuator is based on the notion that the "signal path" is a series path and that series circuit elements are of the most importance and shunt circuit elements are effectively "out of the signal path" because a shunt circuit simply goes to ground.

Based on this, the idea is that all you need to do is use just one very high quality resistor in the series position and then you can use lesser quality resistance for the shunt element without any loss of overall quality as the quality is defined by the series element.

For example, here's a quote from some product literature on Audio Note's website:

The volume control/attenuator is a hand assembled shunt attenuator, where the signal only traverses a single extremely high quality resistor before reaching the input to the line stage. All switch contacts and other components are in the shunt leg of the attenuator.

The notion that it's the series element that's the determinant of quality and what's in the shunt element doesn't matter so much because it's a shunt element couldn't be more wrong. The quality of the shunt element matters every bit as much as the series element.

That's because the signal seen at the output of the attenuator isn't the voltage across the series element, but instead is the voltage across the shunt element. So in a very real sense, you could say that the shunt element is the signal.

Let's look at it schematically:



We have V to represent our signal source, Rx is the series element and Ry is the shunt element.

The signal source sees an impedance equal to the sum of Rx and Ry and for a given signal voltage, a given current will flow through Rx and Ry. As per Ohm's Law, there will be a voltage drop (Vx and Vy) across each resistor as a function of the current divided by the resistance.

As you can see, the voltage drop Vx across Rx, the "single extremely high quality resistor," isn't even seen by the input that the attenuator is driving. Instead, the input sees only the voltage drop Vy across Ry, the shunt element, the one which the proponents of shunt attenuators claim doesn't matter so much in terms of quality.

However as can be seen, the quality of the shunt element matters every bit as much as the series element. And even though the voltage drop across the series element Rx isn't seen by the input, the current flowing through it must also pass through the shunt element so any deviation from the ideal of the series element will also manifest itself across the shunt element by way of Ohm's Law.

The bottom line is that the shunt attenuator has absolutely no technical advantage over a series/shunt type attenuator. And in fact it has a distinct disadvantage in the form of output impedance compared to a series/shunt attenuator.

The worst case output impedance of a series/shunt attenuator is 1/4 the end to end resistance of the attenuator. So if you compare a shunt attenuator with a 10k series resistor to a series/shunt attenuator with an end to end resistance of 10k, the worst case output impedance of the shunt attenuator will be 10k ohms, and for the series/shunt attenuator, only 2.5k ohms.

se

[Folsom]

I am not sure about that exact example but shunting works well for creating a log type response from attenuator/pot.

Also I think it might be possible to wire it so the signal never goes through a resistor. What you do is control the negative with an attenuator/pot. I threw it away cause it died but I had an old amplifier I am pretty sure was wired that way.

[DSK]

Steve, the technicalities are beyond my expertise but perhaps this is why the late, great Bob Crump chose the TKD PC65CS series attenuators (with Vishay S102 bypasses) over all other attenuators for the cost-no-object CTC BlowTorch. He had tried just about every attenuator under the sun (series, ladders, shunts) but said that the TKD "just sounded better". Someone told me that it was also used in the Lamm Ref2 and another high end pre-amp that I can't recall.

[Steve Eddy]

I am not sure about that exact example but shunting works well for creating a log type response from attenuator/pot.

Yeah, that's a bit different. In that case what you're doing is using a regular series/shunt attenuator, with a "law faking" resistor from the wiper to the bottom of the pot.

Also I think it might be possible to wire it so the signal never goes through a resistor. What you do is control the negative with an attenuator/pot. I threw it away cause it died but I had an old amplifier I am pretty sure was wired that way.

Control the negative? You mean negative feedback?

se

[Steve Eddy]

Steve, the technicalities are beyond my expertise but perhaps this is why the late, great Bob Crump chose the TKD PC65CS series attenuators (with Vishay S102 bypasses) over all other attenuators for the cost-no-object CTC BlowTorch. He had tried just about every attenuator under the sun (series, ladders, shunts) but said that the TKD "just sounded better". Someone told me that it was also used in the Lamm Ref2 and another high end pre-amp that I can't recall.

Could be. The TKD is a gorgeous attenuator. Wish they didn't cost so bloody much.

By the way, here's a photo of the innards of the Blowtorch:



se

[JohnR]

The worst case output impedance of a series/shunt attenuator is 1/4 the end to end resistance of the attenuator. So if you compare a shunt attenuator with a 10k series resistor to a series/shunt attenuator with an end to end resistance of 10k, the worst case output impedance of the shunt attenuator will be 10k ohms, and for the series/shunt attenuator, only 2.5k ohms.

Wouldn't that be 5k worst case for the series/shunt? Which is the least common situation as well.

If we took the assumption that both the series and shunt elements affect the sound, then the advantage of shunt is that you can buy just one high-quality resistor and improve an affordable series element. Assuming you don't mind variable input impedance and 6db or so minimum attenuation.

[amplifierguru]

Hi JohnR,

I think the point Steve is making that you seem to have missed is that the ongoing signal IS the signal across the shunt component to ground. It is simply IRshunt the current determined by the (quality of at all audible frequencies) shunt resistance.

On the issue of 'seen' impedance going forward, an assumption that the series component contains ALL the source impedance so it's connected to a zero impedance source. In that case, if represented by a conventional pot, the wiper impedance is the input to wiper series R in parallel with the shunt R, which for a 10K pot is 5K//5K = 2.5K = 1/4 x 10K.

Loved the pic...

Cheers,
Greg

[JohnR]

Yeah, and the voltage across the shunt element is also determined by the series element. What you seem to have missed is that it's a voltage divider.

[amplifierguru]

Hi JohnR,

Covered in the I component which is Vin/( Rseries + Rshunt).

 

Greg

[Steve Eddy]

Wouldn't that be 5k worst case for the series/shunt? Which is the least common situation as well.

Worst case would be with no attenuation, i.e. volume turned all the way up, in which case the shunt resistance is infinite which means that your output impedance will be that of the series element, or 10k. Of course that assumes the ideal condition that the source impedance of the component driving the attenuator is zero. The source's output impedance will simply add to that 10k.

And yeah, it's not likely that you'll have the volume turned all the way up much of the time. But then I did say this was worst case, not average case.

Anyway, the output impedance issue was just a little side note. The main point was to dispel the myth that the quality of the shunt element is of less importance than the series element.

If we took the assumption that both the series and shunt elements affect the sound, then the advantage of shunt is that you can buy just one high-quality resistor and improve an affordable series element. Assuming you don't mind variable input impedance and 6db or so minimum attenuation.

Not sure I'm following what you're saying here.

se

[Steve Eddy]

Yeah, and the voltage across the shunt element is also determined by the series element. What you seem to have missed is that it's a voltage divider.

Yes, it's a voltage divider, but the salient point is that the only voltage seen at the output is the voltage across the shunt resistor as a consequence of the current flowing through it. Meaning that the quality of the shunt resistor is no less important than that of the series resistor.

se

[randytsuch]

Steve
I just wanted to make sure I understood what a "series/shunt Attenuator" is. 
I am thinking a pot, one end input, other end ground, with output at the wiper.

BTW, good argument about the shunt attenuator.  I never thought of it that way, but it is pretty obvious with your picture that the shunt element is creating the output voltage.

Randy

[amplifierguru]

Meaning that the quality of the shunt resistor is no less important than that of the series resistor.

and no more 

Greg

[Steve Eddy]

I just wanted to make sure I understood what a "series/shunt Attenuator" is. 
I am thinking a pot, one end input, other end ground, with output at the wiper.

Yes. Either a pot or a switched attenuator that works the same way, i.e. both the series and shunt resistances change. Some have called these "series attenuators" but if a "shunt attenuator" has a fixed series and variable shunt, a series attenuator might be mistaken as having a fixed shunt and variable series resistor.

BTW, good argument about the shunt attenuator.  I never thought of it that way, but it is pretty obvious with your picture that the shunt element is creating the output voltage.

Thanks.

Yeah, I hadn't really given it much thought before myself. Soon as I realized they had a variable input impedance, I pretty much forgot about them right there and then.

se

[WEEZ]

How about a lesson for dummies

How does a series attenuator or a ladder attenuator differ...pros? cons?

thanks

[Steve Eddy]

How about a lesson for dummies

How does a series attenuator or a ladder attenuator differ...pros? cons?

Here's a good piece for you to read over on Goldpoint's website:

How Stepped Attenuators Work

se

[Occam]

and a golden oldie from the mythic Holler'n Harry -
http://www.diyaudio.com/forums/showthread.php?s=8c459a6ce370ad40d5e2daf4ff3ac528&threadid=2608&highlight=

[WEEZ]

thanks gentlemen

[randytsuch]

and a golden oldie from the mythic Holler'n Harry -
http://www.diyaudio.com/forums/showthread.php?s=8c459a6ce370ad40d5e2daf4ff3ac528&threadid=2608&highlight=

FYI, Harry now posts at diyhifi.org as Fred D.

A while ago, I built one of these.  Sounds very nice in the right system.  Eventually replaced by a TVC, but those are a lot more money.  Passives want to have short IC's.

Randy

[markC]

Why not go with a ladder step and not worry about it?
I know it's more expensive, but if your looking for the purest signal path, then 2 high quality resistors at any given level has got to be best.
I've been running dual mono Goldpoints for a couple of years now, and except for the slightly irritating occasional pops, they work great.
Huge difference over the carbon wiper crap that they replaced.