Re: The not-all-that-low distortion sine wave oscillator in a faster simulating version

On 20/12/2024 4:22 am, Edward Rawde wrote:

"john larkin" <JL@gct.com> wrote in message news:uoj8mj9t5vc84kl4mdr01n3spqtnra2u6v@4ax.com...

On Thu, 19 Dec 2024 09:53:03 -0500, "Edward Rawde"
<invalid@invalid.invalid> wrote:
>

"Bill Sloman" <bill.sloman@ieee.org> wrote in message news:vk0ehh$2o9dc$1@dont-email.me...

On 19/12/2024 2:10 pm, Edward Rawde wrote:

"Bill Sloman" <bill.sloman@ieee.org> wrote in message news:vk00um$2i900$1@dont-email.me...

On 19/12/2024 6:00 am, Edward Rawde wrote:

"Bill Sloman" <bill.sloman@ieee.org> wrote in message news:vjtgnp$24ubg$1@dont-email.me...

I've been playing with the circuit, and have got rid of one op amp, which made the simulation run much faster, but didn't
help
the
distortion performance.
>
Swapping the LT1115 for the LME49710 speeded up the simulation a bit more, but didn't make any difference to the distortion
either. A few of the ferrite beads have gone too.
>

>
I got a simulation speed of about 30us/s so I didn't wait the nearly 4 days it would take to complete.
I did an FFT on the first few cycles and it does look 100dB down up to 1.5MHz.

>
It rans at 68msec/sec for me and takes a couple of minutes to run the full ten seconds.

>
I used the circuit from your third post. One op amp had to be moved down a bit into position and then I hit simulate.
To be sure we're talking about the same circuit I've reposted it below.
I'm using LTSpice 24.0.12 with no new model updates available as of this post.

>
And I'm using LTSpice XVII(x64)(17.0.37.0) up-dated recently.
>
I finally got your version to work. As you say, U1 had to be moved into place, but I also had to change C10 on the output of U4.
I'd specified the capacitance as 3.3u. but the "u" symbol had vanished. When I specified the capacitance as 3300n everything
worked fine.

>
C10 is definitely 3.3u here. I tried changing it to 3300n but still less than 30us/s when I start the simulation.
>

>

If building this for real then ten turn trimmers would be used for:
R14 2.2k
R3 68 ohm
R16 100k
And I'd also want R19 or part of it variable.

>
Why a ten turn trimmer?

>
Can be 100 turn if you want. The point is only that fine adjustment would be a good idea.

>
Lots of turns don't always equate to fine setability.

>
Yes I agree. You might easily have the kinds of problems which were solved with anti backlash couplings in the days of drive cords
and tuning capacitors.

>
The long multi-turn trimpots are hard to adjust and expensive, and are
no better than single-turns for settability.

 Yes particularly if the resistive element is the same length, it may as well be single-turn.

Not always true.

Single-turn also has the advantage that you can see where it's set before you adjust it.

 >

Multi-turn is usually enclosed so hard to know where it's set before adjustment.

There was a line of multi-turn trimming potentiometers with a clear plastic top cover which did let you see where the brush was. Cambridge Instruments bought quite a lot of them. If you thought about what you were doing it wasn't all that much of an advantage.

 https://www.google.com/search?&q=multi+turn+trimpots&udm=2

DSMCZ seems to be an example of a multi-turn trimmer with a translucent case.

So I'd probably go for good quality open single-turn if I ever build the 120dB circuit.

Bad choice.

Single-turns have much lower HF parasitics too.

The resistive element in a modern multi-turn trimmer  is just a long thin film resistor on an alumina substrate - there's not a lot of RF parastic stray capacitance and inductance in that.
Wound element potentiometers were a lot worse.
--
Bill Sloman, Sydney