DIY Attenuator Pad help

[stevedc]

Newbie issues here - hoping for some help.

I have an audio signal that's too "hot" for the "EXT IN" input on my Tascam digital recorder. (These are all unbalanced RCA/phono type connections.) I have an old inline attenuator cable, around 60dB, but it's way too strong, so I thought a fun project would be to build my own.

Output impedance on my source device is 47 ohm  (and voltage is 2.2v).
Input impedance on the digital recorder is 25K ohm.

My plan is to make a simple L pad:
1) Set the shunt resistor to 47 ohm (or as close as possible)
2) To achieve an 18dB attenuation, set the series resistor at about 326 ohm by using the formula Rseries = Rshunt * ( 10^(db/20) - 1 )

In researching all this, I found lots online about T pads and Pi pads and matching asymmetric impedances and it wasn't clear how much of all that I needed to worry about for a simple audio application.

So I ended up thinking the L pad was the way to go, and that I should have the output of that circuit closely match the output impedance on the source. I'm not sure if I should be setting resistance values in the circuit based on what the load (input of the digital recorder) will see, or what the source will see.  (By my calculations, I could also set the shunt resistor to 6 ohms and that would give about 47 ohms toward the source with the same 18dB attenuation, but I'm not sure what the implications are that vs the values I proposed above -- or if there is another approach that's better.)

Any insights into this would be appreciated.
Thanks!
 Steve

Happy to hear any thoughts on the impedance issues and whether I'm missing anything in my proposed design.

[Richard Crowley]

Since most modern audio gear has quite low output/source impedance (47 ohm in your case) and rather high load/destination input impedance (25K ohm in your case) the easiest (and most flexible) would to simply use a dual audio-taper pot.  That way you can "gain-stage" all the links in the chain for optimal signal-to-noise ratio.  I would use something like this....

http://www.allelectronics.com/item/apd-10k/10k-audio-taper-dual-pot-pc-mount/1.html

It is cheaper than the four resistors (for stereo, I presume) and it is continuously adjustable.
Can't hardly beat that.

If you really have your heart set on a fixed pad, this is my go-to resource for calculating audio pads:
http://www.uneeda-audio.com/pads/

[stevedc]

Thanks, Richard.

Would you expect there to be sound quality/distortion differences using a pot vs a fixed pad using good quality metal film resistors?

PS - yes, I've spent a lot of time at the Uneeda Audio site - that's what got me this far!

[Richard Crowley]

Unless you use a really grungy, dirty 2-cent pot, I would NOT expect "sound quality/distortion differences" from a pot vs. a fixed-resistor pad.  There is more distortion in the Tascam recorder (and likely in the unidentified source gear) than you would expect from a half-way decent audio pot.  Chances are that both your source gear and the Tascam recorder are using similar pots already.  Even that 25-cent pot will NOT be the weakest link in the audio chain.

[Alex Nikitin]

The main problem with dual pots, especially log ones, is that they will have a rather large imbalance (up to 2-3dB) between two halves at least at some points on the adjustment curve. It would be correct to build a fixed attenuator in this situation. I would go for a total resistance around 20K, it would not load the source too much (remember, it is an audio device, and 47 Ohm output impedance is just that and do not require any matching load, moreover, to use a load below 10K would be unwise). I would use 22K and 3.3K resistors do make an attenuator.

Cheers

Alex

[TimFox]

Check the specification on your source's output.  It may have a 47-ohm output impedance, but might not be happy driving such a low load resistance.  For example, a standard op-amp output with a 47-ohm series resistor would ordinarily expect a load of at least 2000 ohms.  The 47-ohm resistor is to avoid ringing when driving long cables.
Your L-pad design is reasonable, but I would use a first resistor of 2000 to 5000 ohms and a second resistor to achieve the required attenuation.

[stevedc]

So if i go the L pad route...

I'm looking for 18dB attenuation and from the comments above I need about 22k ohm total resistance.

So that means I'd have ~ 19k2 on the serial resistor and ~ 3k2 on the parallel.


Does that sound reasonable to everybody?

[Electro Detective]

Play with the dual audio-taper pot arrangement suggested above,

once you get the result that works for your system, measure the values

and knock it up with fixed resistors that won't ever get wiper fuzz

or jump carbon tracks at the sweet spot