"Pooh Bear" wrote
in message

Arny Krueger wrote:

In fact audio generators with rediculously low
residiuals have been made using nothing more exotic than
NE 5532s.

Various construction projects like this have been published in the past,
including one by me. There was also a far more elaborate project by Cordell
that was published in Audio Magazine.

If one works with building audio generators of the traditional analog kind,
it turns out that the nonlinearity of the means used to stabilize the levels
is the major source of distortion, not the amplifier portion of the
oscillator. This is true whether a light bulb, a CdS opto-isolator, a
thermistor, a FET or a VCA is used. Been there and done that for all of
them.

I may shortly be diving inside our AP test set (
backlight needs replacing ) . Dunno what they use
actually. Maybe some exotic PMI or AD parts in selected
places ?

Maybe even discrete op amps, depending on the age.

One relevant parameter is the maximum amplitude that is provided. One
classic benchmark maximum output in the 10 vrms or +22 range. To provide
this at the generator's output terminals @600 ohms, you have to have a few
dB more at the op amp's output terminals.

You can't really do this with +/- 15 or +/- 18 supplies. One can stretch
NE5532s to +/- 22 but they tend to degrade over years.

The only high-voltage op amp chip that I know (one that shows signs of
hanging in with +/- 22) of is the OPA 604/2604. According to Doug Self,
they vastly underperform NE 5532s for nonlinear distortion.

Modern DAC chips are so good, and digital computation and function
management is so cheap and pervasive, that a modern sound card in a PC is
the most practical way to generate well-controlled sine waves these days.

It takes a lot of work to outperform a M-Audio Audiophile 24192 driven by
simple freeware software like Audacity and/or Audio Rightmark.