Passive I/V: I/V resistor as attenuator

[Steve Eddy]

Has anyone using a passive I/V DAC tried using the I/V resistor as the attenuator? In other words, using a switched attenuator or pot configured as a rheostat for the I/V resistor.

Since a current-output DAC outputs a given current for a given signal level, reducing the value of the I/V resistor will reduce the voltage dropped across it and hence the signal level seen by whatever the DAC is driving.

One interesting aspect of this would be that the source impedance of the DAC would behave like a transformer volume control with the source impedance progressively dropping as the signal is attenuated more and more.

Of course unlike a TVC, the DAC would see a lower and lower load impedance as the signal was attenuated but then the output of the DAC is a current source (or sink depending on the DAC chip) which ideally wants to see a short to ground.

se

[Gridstop]

Wouldn't you need incredibly small resistors for the smallest settings?

If we limit the 0db setting to 100 ohms (and some people recommend even smaller values, like 10-20) then wouldn't say a -60db setting need something like... well I forget the math offhand but some tiny fraction of an ohm?

[jwb]

I would guess you can get down to the sub-Ohm level by switching in more parallel resistors. Eventually you'd approach the resistance of the switch/relay or the pcb traces.

Anyway the original idea seems weird, since you'd get the maximum output at the highest resistance and hence the furthest from the ideal virtual ground.

[Steve Eddy]

Originally posted by Gridstop
Wouldn't you need incredibly small resistors for the smallest settings?

If we limit the 0db setting to 100 ohms (and some people recommend even smaller values, like 10-20) then wouldn't say a -60db setting need something like... well I forget the math offhand but some tiny fraction of an ohm?

Well, 60dB is a ratio of 1,000:1 so with a 100 ohm resistor you're looking at what, 0.1 ohm? That's doable.

se

[Steve Eddy]

Originally posted by jwb
Anyway the original idea seems weird, since you'd get the maximum output at the highest resistance and hence the furthest from the ideal virtual ground.

Yeah, but if you're going to use passive I/V conversion at all you've already thrown that "ideal" virtual ground out the window, yes?

At least in this case, everything except 0dB of attenuation will have you closer to that ideal. Otherwise, the I/V resistor remains fixed at all times.

se

[Spartacus]

This is something I've wondered about too. It seems more wierd to me to go to great lengths to get distortion free voltage only to throw much of it away at a pot.

[Petter]

My understanding is as follows:

1. The process used for DAC chips is not typically good for analog circuitry such as amps - hence current output amps typically offer the best performance
2. The current output from such a DAC is very sensitive to impedance and needs to be buffered so it sees a "ligh load", typically held at a particular voltage. Hence a resistor as an IV without a buffer in front is not considered optimal in traditional datasheet terms.

Petter

[Steve Eddy]

Originally posted by Spartacus
This is something I've wondered about too. It seems more wierd to me to go to great lengths to get distortion free voltage only to throw much of it away at a pot.

True. Though it ain't exactly distortion-free voltage. Even if the resistor were perfectly linear, the current output circuitry of the DAC will fare worse in terms of distortion as you increase the load impedance that it's driving.

Increasing the load impedance seen by the DAC's output is rather like decreasing the load impedance seen by a voltage source.

But there is a certain elegance to replacing an active I/V stage with a simple passive resistor.

se

[Steve Eddy]

Originally posted by Petter
My understanding is as follows:

1. The process used for DAC chips is not typically good for analog circuitry such as amps - hence current output amps typically offer the best performance
2. The current output from such a DAC is very sensitive to impedance and needs to be buffered so it sees a "ligh load", typically held at a particular voltage. Hence a resistor as an IV without a buffer in front is not considered optimal in traditional datasheet terms.

Well, yeah, typically the I/V stage is an active one with the output of the DAC driving the virtual ground at the inverting input of an opamp. But many prefer the sound of passive I/V conversion using nothing more than a fixed resistor.

I'm just thinking that if one is going to take that approach, instead of using a singular resistor of a fixed value for the I/V conversion, you use a variable resistor or switched resistances? This would seem to optimize the performance of the passive I/V approach.

se

[fdegrove]

Hi,

See it already....a variable I/V followed by a variable mu triode...

Viva la distortion...Oops...Wrong planet....

Cheers,