Arny Krueger wrote:
"Pooh Bear" wrote
Audio Precision test oscillators offer THD residuals in
the 0.0006% region ( -104dB )

I looped-back a little XP PC with an Audiophile 24192 in it the other day,
running the freeware Audio Rightmark Program. *All* spurious
responses were 112 dB or better down (most in the -120 dB range),
and THD+N was something like -106 dB.

Not a toob in sight !

AFAIK toobed audio generators and analyzers never got within 2-3
orders of magnitude of -106 dB residuals. Something like 0.05%
midband, and 0.1% at 20 and 20 KHz was about as far as toobs got.

The biggest problem with low residuals back then was not, in my
experience, wether they used tubes or solid state, rather on how well
the unit was stabilized. There was a Krohn-Hite tubed oscillator, as I
recall, that was easily capable of well below 0.003% , you just had to
let it sit there and stabilize. AT the same time, some of the GR solid
state oscillators, like the 1309, could be tuned to meet those kinds of
levels, and MIGHT have been capable of far better, but the amplitude
stabilization network just just too noisy: you'd watch the residual
bouncing around and every once in a while you'd see it drop a good
20 dB below its average for a brief period (about a second). Bang the
case, upset the filament in the bulb they used for stabilization, and
you'd see the residual go all over the place.

The real secret to low-residual oscillators came with much better ,
lower noise and faster responding loop stabilization. The original
ST1700
had an oscillator circuit not substantially different than whatever
else was
out there but had superior stabilization and that was the secret to
their
significantly lower residual.