Re: Final final 1kHz oscillator

"Edward Rawde" <invalid@invalid.invalid> wrote in message news:vgrbl7$30bo$1@nnrp.usenet.blueworldhosting.com...

"Cursitor Doom" <cd@notformail.com> wrote in message news:ee92jjllg5kudr83e8bvvbsjba754aecjs@4ax.com...

On Sun, 10 Nov 2024 18:00:01 +0000, Cursitor Doom <cd@notformail.com>
wrote:
>

On Sun, 10 Nov 2024 17:45:00 +0000, Cursitor Doom <cd@notformail.com>
wrote:
>

On Sun, 10 Nov 2024 12:12:34 -0500, "Edward Rawde"
<invalid@invalid.invalid> wrote:
>

"Cursitor Doom" <cd@notformail.com> wrote in message news:bs81jjt7d064jc1ktmvihgn30qkhf67taj@4ax.com...

On Sun, 10 Nov 2024 00:49:36 -0500, "Edward Rawde"
<invalid@invalid.invalid> wrote:
>

"Bill Sloman" <bill.sloman@ieee.org> wrote in message news:vgnkgg$3p31a$1@dont-email.me...

On 9/11/2024 7:39 pm, Edward Rawde wrote:

"Bill Sloman" <bill.sloman@ieee.org> wrote in message news:vgn4gm$3lcor$1@dont-email.me...

On 9/11/2024 7:43 am, Edward Rawde wrote:

This is the simplest circuit I can come up with.
Harmonics are more than 80dB down in simulation.
FFT the last 30 seconds.
There is only one harmonic stopping me claiming 90dB but the exact FFT result can depend
on exactly how much output is selected for the FFT. You may see different results.
>

>
I have fixed my circuit but it raises more questions than it answers.
This circuit deliberately has no text and no u symbols so should be ready to simulate.
If you let it complete (about 20 minutes on a fast PC) the FFT should show at least 80dB purity and there aren't many
harmonics
standing in the way of 90dB.
Startup isn't fast and it takes at least 20 seconds of simulation to fully settle.

>
Spectral purity might be fine for all I can tell but the output
amplitude is all over the place,

>
Change R9 to 1.07Meg does it amplitude stabilize now?
Adjustment will be needed in any real circuit.

>
Indeed, but for a simulation we should be seeing identical results!
I've made the change as you suggested and it's increased the sim time
considerably. I'll post again when it's completed.

>
Update: looking *much* better at 25% done. Settled after 7.5s. I will
let it run its course anyway and post again at the end with the final
outcome.

>
Looks really stable now; 4.5V peak to peak and none of the
fluctuations of the earlier simulations. Weird how it took *so* much
longer to complete this time,  though.

>
An FFT on the last 10 seconds should show peformance approaching 90dB down.
>
I likely won't be working on this any further unless anyone else wants further discussion.
>
The goal was to get the purest possible sine wave at the lowest cost and in simulation I don't think I can improve on this
circuit.
I'm not able to build this circuit and even if I did I don't have equipment capable of measuring distortion 90dB down.
>
I'd still like to know why LT1057 doesn't work in position U2. It's not like rail to rail is needed.
LT1678 works exactly as expected. LT1057 probably would in reality but not sure I'd risk it.
>

To answer my own question a wild guess says that the simulation model is hard coded to keep the output 2V away from the rail.
Seems pretty poor to me that an op amp operating on 6V can't do more than 4V out.
I wonder what a real device would do.
The LT1678 works because it's rail to rail.