What am I missing?

[jeff mai]

Here is the scenario:

Passive I/V stage (2K7 to ground) --> 4.7uF coupling cap --> 12H log tapped autotransformer (100 ohms DCR) as volume control --> 22K input impedance at amplifier.

I figured on some easily measurable (and possibly audible) LF roll-off due to:

1) lower AC impedance to ground at LF for the I/V stage
2) voltage divider effect of the cap + autotransformer at LF

Tried it - couldn't hear any roll-off. Measured it (with CRO) - I get 1 or 2db down at 10Hz. No humps, bumps or dips from 20Hz to 20kHz.

It could be that I don't fully understand how the passive I to V works from the DAC, but I'd guess that the DAC's current output causes a voltage to develop across the resistor to ground.

I guess the other possibility is that the autotransformer is much more than 12H, but I don't think so. That figure came from the maker.

Any thoughts on what I'm missing?

 

[jeff mai]

More info I forgot to mention:

The I/V stage is attached to a TDA1543 DAC chip.

The position of the volume control during the test is such that the reflected load will be very, very high, making coil inductance + DCR the dominant portion of the autotransformer impedance at low frequency. Even at a 1:1 ratio (full volume), the reflected load of 22K will be 10 times the autotransformer impedance at 20Hz.

 

[SY]

Dumb question- I've never used the 1543, but I did a lot of playing around with passive I/V using a 1541A. For that DAC, the resistor used as the I/V had to be MUCH smaller than 2k7, otherwise the linearity and dynamic range went to hell. Something like 100 ohms was appropriate. The small resistor load meant more voltage amplification was needed, so that was a big trade-off. Does the 1543 have a spec for the maximum acceptable resistance load?

 

[jeff mai]

Does the 1543 have a spec for the maximum acceptable resistance load?

Don't know, but even were 2K7 too high, I don't think this would affect the FR. It certainly wouldn't flatten out by coincidence!

What I hear and measure make no sense. I must be overlooking something.