Re: Sinewave oscillator without gain control.

On 30/01/2025 10:25 am, Edward Rawde wrote:

"Bill Sloman" <bill.sloman@ieee.org> wrote in message news:vnck1i$295d5$1@dont-email.me...

On 24/01/2025 5:02 pm, Edward Rawde wrote:

"Bill Sloman" <bill.sloman@ieee.org> wrote in message news:vmv6ac$22ufe$1@dont-email.me...

On 24/01/2025 11:14 am, Edward Rawde wrote:

Before you can make a 1kHz sinewave oscillator with constant output level and better than 120dB harmonic distortion you'll need
an
oscillator which does better than that when it passes through the required output level.

>
This is obviously true. Why do you think you need to tell us about it?
>

>
You do make me laugh sometimes Bill.
Your circuit doesn't come anywhere close to 120dB but it does depend on which simulator you ask.

>
I never said it did. I was looking for comments on the way the current mirrors might screw up the waveform. I didn't get any.
>

Adding a FET gain stabilizer to the circuit I posted seems to have negligible effect on the distortion.

>
Then you haven't looked at the current waveform going through the the FET.
>

So I'd concentrate on the oscillator distortion level rather than the gain controller contribution if I wanted to attempt better
than 120dB.

>
And you'd probably better dump LTSpice 24.1 - it has been claimed that it has a serious bug. That might explain why your
simulations of my circuit run a thousand times slower for you than they do for me and a couple of other people.

 If U3 and U8 (LT1115) in your circuit are replaced with LT1678 then on my computer with LTSpice 24.1.1 and all component updates,
the simulation completes almost as fast as I can start it, at about 55 ms/s.
A sample near 10s is not quite 60dB down at 3kHz and 5kHz and 70dB down at 7kHz, approaching 80dB down everywhere else.
The parasitic 23MHz signal has gone.

The LT1678 looks like a nice part, but at $10 each it has to be. The application doesn't need a rail-to-rail op amp.
John May has pointed out to me that the LT1115 LTSpice model is just a variant of the LT1028 Spice model. I used the LT1028 in a rather demanding application some forty years ago and while it had its idiosyncracies, generating a 23MHz parasitic oscillation wasn't one of them.
The performance you are now getting sounds like what I've been seeing all along, and presumably reflects the defects of the current mirrors.
I've tried to improve them a bit but never got more than a dB or two of improvement. I need better insight. Talking to people often helps develop that but it hasn't worked here.
--
Bill Sloman, Sydney