[GRe]

"patrick-turner" wrote in message
...

[lot of snips]

These pots seemed to wear out all too easily. Trying to get GOOD log pots
isn't easy. I tried a 20k dual Alps Black log pot but the distribution of F
along the dial was really bad, stretched out on one section and bunched up
in another.

-Like I said, not my experience, for the wearing out part that is.
But true for the "unusual" frequency distribution along the dial, but easily
solved by a frequency counter i.s.o. a dial.

I have a single F oscillator to make 1.0kHz, with tuning cap to get +/-
20Hz. I used a single tiny globe that is 60r cold and about 200r with
2.5Vac across it. I got 8.3Vrms out with NE5534 with +/- 15.6Vdc rails and
THD = 0.004%.

-I had a look at your single F osc. A THD of 0.004%/-88dB looks promising.
However, I'd like to keep F variable.

-While going trough my junk stock I found a 3-gang tuning cap. with about
10-575pF per section. Unfortunatly, but usual I guess, the cap's chassis is
common for all three sections, so, with common connected to the NE5534 +
input the cap's chassis must be mounted insulated from ground. I'll see what
teflon or other insulating standoff's can be found in junk stock, then mount
the cap "floating" in a cookie tin.

-Just measured 25pF parasitic capacity between the tuning cap floating in a
(carrots and peas) tin. The parasitic 25pF is in fact parallel to C2, so I
guess I need an identical cap/trimmer parallel to C1 for compensation.

-BTW, just curious, what does happen, if anything at all, if 2 cap sections
in parallel were used for C2 and at the same time R3||R4 is lowered to half
the value of R1||R2? After all, in that case the basic condition
Xc(C2.R3||R4)= Xc(C1.R1||R2) is still met.
I'm asking because the max. capacity of all 3 sections differ slightly. But
half the capacity of section 1 || 2 is almost equal to the capacity of
section 3. So, in that case the cap has better symmetry.

Pot was fuct, so I replaced is but I didn't like the result. There was no
room to install radio tuning gangs which give much less bouncy trace. But
vari caps are limited at LF and I want 1Hz.

-Don't you get stability and/or THD problems at that low frequency caused by
the limited thermal capacity of the lamp?

Thanks for the link to BWD141. The schematic there was what I had. I have
not removed the schmitt trigger sq.wave amp yet.

-When not in use, the sine-to-square converter is an asymmetrical load to
the sine wave generator output. My gut feeling is THD might improve a little
when the converter is disconnected from the sine wave generator (f.e. at
R12).

The 141 seemed to have attrocius DC offset on all outputs, quite
intolerable!, so rather than just fix this peice of junk, it better to
entirely re-make it. I should end up with a square wave at 1MHz which has
its highest F present of about 5MHz, OK for audio where ringing due to
square waves needs to be examined.

-By design the output has a little positive DC offset (a seldom case where
positive is negative but negative would'nt be positive) via RV3 and R54
caused by the bias across R5. Eversince it's such an old generator did you
check C17/C14?.

An old Elektor publication describes a wien bridge osc. with 2H at -85dB
and 3H at -80dB, but before I start building I'm curious what kind of
figures you obtain with your circuit(s)?

Good enough figures for me. If you want variable F AND low THD 0.001%,
then it gets much more difficult and complex.

One could spend years fiddling around with oscillators, and my time is
limited and I have many amp projects to commplete.

-Exactly the reason I have to be satisfied with building either the "Elektor
oscillator" or the NE5534 based one. So, at the moment for me no
subsequent-to-oscillator filter-gear to get another for tubetronics
unrelevant -15dB on top of a more than perfect -85dB.

Regards,
Gio Re