From: email@example.com (Tom Bruhns)
Subject: Re: Output Impedence
Date: 30 Oct 2002 12:07:46 -0800
References: <firstname.lastname@example.org> <email@example.com>
NNTP-Posting-Date: 30 Oct 2002 20:07:46 GMT
John Woodgate wrote in message news:...
> I read in sci.electronics.design that Tom Bruhns wrote
> (in <firstname.lastname@example.org>) about 'Output
> Impedence', on Sun, 27 Oct 2002:
> >Well, it does matter some. What's the input impedance of your mixer?
> >Use a transformer matched to it; that's more important than matching
> >the output impedance of the soundcard, which likely is pretty low. If
> >you load the transformer with a load equal to the transformer's rated
> >secondary impedance, then the primary will present (nominally) its
> >rating to the sound card output, and that shouldn't be a problem.
> No, we don't do that. The mixer input impedance is likely to be 10
> kohms, and if you want a transformer that 'matches' that, it implies a
> primary inductance well above 10 kohms at 20 Hz, which is a LOT of
Yes, I'd agree that matching to 10kohms is not appropriate. Mixing
boards I'm familiar with do have 600 ohm (nominal) inputs, though.
But...if the mixer is high impedance input, then if you want really
flat response over a wide range, I'd say to use a good "600 ohm 1:1"
transformer, and put a load on its output which makes it flat (as flat
as possible or as you need).
I thought it would be interesting to remind myself of what a real
transformer's frequency response does as a function of output load, so
I dredged out an old 600:600 "bridging transformer". What I see is
that the low end isn't terribly affected by load (-0.1dB at 0.6Hz,
yeow!), but the high end is quite sensitive to load. The transformer
package has a built-in 10k load, so I didn't go above that, but with
that load, there's a resonant peak of +9.3dB at 186kHz and it's +3dB
at 97.5kHz. It's +0.1dB at 12kHz. With a 600 ohm net load, it falls
off to -3dB at only 58kHz, and -0.1dB at 11.7kHz. However, with a
load adjusted for widest +/-0.1dB flatness (about 2kohms), it goes to
109kHz (-3dB at 208kHz), almost ten times the +/-0.1dB bandwidth you
get with the 600 ohm load.
Wish that was easy to display here as a graph. But the summary is:
peaking at high freq with a 10k load. High freq falloff with a 600
ohm load. Both limit the freq response (+/-0.1dB) to about 12kHz.
But an optimized load extends the response to 109kHz. There are
similar benefits if you measure to +/-3dB bandwidth.