Help with Impedance Matching

[mudihan]

My cd player's output impedance is 200ohm (output is about 3v, which is just enough) and my power amp has 22k input impedance. What value of pot should I use for a passive volumn control (input-pot-output)? Stepped pots are not considered due to high price. My cables are made of Belden 1695A, which is 75ohm and has low cap (16.2pf/ft). They are also less than 2 feet long. Thanks!

[sreten]

Hi,

A 10K pot is suitable for the job.

/sreten.

[mudihan]

Thanks!

[AndrewT]

Hi,
check that both the CD output and the power amp input have a DC blocking capacitor in the lines.

Pots do not like a permanent DC voltage across the wiper. I have heard it reduces their life and/or makes them noisy.

You may get away with a 5k log pot, but since your cables are so short it will offer little improvement.

An alternative is to use a linear pot with log faking resistor.
For this you choose a pot about 5 to 10times the impedance (50k) and place a resistor of about 10% to 15% (5k to 7k5) of pot value from wiper to common on each channel. This improves the tracking between channels when listening at very low volume. But it suffers from high source impedance at very loud volume, again that should not be a problem.

[sreten]

Originally posted by AndrewT
.
....But it suffers from high source impedance at very loud volume.......

Hi,

How so ? would be highest at mid position, 2K to 3K, and would not
vary much over most of the range, but be low near max and min.

It is a very preferable option over a log pot for the reasons you
mention, channel tracking at low volumes. The other option to
help with this is add a level switch (say -10dB), so that the
pot is generally never used near is minimum settings.

/sreten.

[richie00boy]

Indeed, at full volume the source impedance is simply the source impedance in parallel with the pot value and the faking resistor.

A law faked pot offers broadly comparable source impedance to a log pot, throughout its rotation. The only difference is the loading on the source is more with a faked pot, at higher volume settings.

Also, to implement a faked pot, one would sensibly utilise the input impedance of the amp as the faking resistor, provided it was of a reasonably low value (say 22k or less).

[AndrewT]

Hi,
a normal volume pot on the input has the usual variable source impedance as seen by the following input stage.
You correctly identified the highest impedance condition when volume is at -6db relative to max rotation volume. i.e. two half track resistances in parallel giving an effective source impedance =track resistance/4 (there is a small correction to make for the source impedance of the stage before the pot).

With a law faking resistor from wiper to common the impedance variation is not similar to the above.
eg. at half wiper travel the track resistances will be 25k and 25k and the faker is 5k in parallel with the lower half track.
The effective source impedance is now 25k//25k//5k=3k57 (again correction for source before pot).
This example shows an increase of 43% in source impedance of faked cf. log. If the usual 10times(100k) & 15%(15k) were applied then the source impedance jumps to 9k4 (+276%), very much worse.
I would call this high source impedance when set to high volumes.

The source impedances of log pot and law faked pot are not comparable at high volumes settings.
At the extremes of rotation both types have the same impedance characteristics.

If you feel inclined, you can make the source and input impedance corrections and then draw a graph of db attenuation vs effective source impedance for both types and you will find that my statement holds true.