Sin/Cosine curves...

[SSLtech]

I'm trying to get as close as I can to a sin/cosine curve for a 'crossfade' circuit... I'm currently experimenting with a "panpot" type circuit (constant load impedance is not a big issue) and the circuit is thus:

Two feeds in via two equal "input" resistors (currently 15kΩ) these meet the two ends of a 10K linear pot, with the wiper grounded. The signal then leaves from the two ends of the pot via two "output" resistors (currently 15kΩ) to a virtual-ground summing circuit, which is gain-set to correct for the insertion loss of the circuit.

I've written an excel spreadsheet which plots the attenuation for each circuit and also compares them to a true sin/cosine plot (sin/cos gives equal power for non-correlated signals, and a -3dB at center for each signal...)

So far these numbers generate close enough to -3dB at center, but I'm wondering if there are any other circuits which will get closer to true sin/cos curves, without breaking the bank...

Thanks in advance,

Keith A.

 

[NewYorkDave]

Keef,

Drop me a line. electronic_dave at y*hoo. I can help.

 

[SSLtech]

Cool...For reference, here are the curves I am getting at the moment...

Keith

 

[chrissugar]

Hey Dave, I'm also interested in a solution for the sin/cos panpot. Can you help me too?

chrissugar

 

[SSLtech]

Okay... I get it now. Audio & Design recording (those fine folks who made the Compex, Vocal Stressor, Gemini, ScampRack etc...) actually had custom sin/cos pots made for their ambisonic decoder.

I've managed to plot a graph of RMS signal level against angle of rotation in my own spreadsheet (this is actually very cool... I'm learning a lot about excel!) -The RMS sum of the two signals actually varies by less than 3% peak, and the average error is about 1.3%... component tolerances are likely to have a greater effect on a built result, I suppose. :grin:

Learnin' a lot here!

Keith

 

[PRR]

> I'm trying to get as close as I can to a sin/cosine curve for a 'crossfade' circuit...

The math is elegant, but is this really what you want? Constant power seems nice, but DJ mixers find other curves more artistic.

In general: if there was a good way to fake a SIN with a LIN pot, there would not be $600 SIN pots (or used to be, when analog still ruled the world).

I suspect that your loaded-LIN pot plan is close-enough to constant-power that the ear can't say otherwise, nor would it really care. After all, you change the song, you inevitably change average level-- this isn't like the case of cross-fading from one 120.0V generator to another, the two sources ARE different.

 

[SSLtech]

PRR, This is part of a bigger circuit which does in fact call for some accuracy... with your permission, I'll PM you some links if you'd care to scan an eye over them for me?

Keith

 

[PRR]

4% peak 1.3% average error is probably plenty good for true constant-power cross-fade. The 0.33dB peak 0.11dB typical error is much better accuracy than typical audio precision. You will need some sharp test gear to find any two mikes, amps, tape-channels, etc matched that well, even if they are sold as a "matched set".

I think you should fly with it.