Path: ...!eternal-september.org!feeder3.eternal-september.org!news.eternal-september.org!eternal-september.org!.POSTED!not-for-mail
From: Bill Sloman <bill.sloman@ieee.org>
Newsgroups: sci.electronics.design
Subject: Re: Sinewave oscillator without gain control.
Date: Fri, 24 Jan 2025 15:52:14 +1100
Organization: A noiseless patient Spider
Lines: 59
Message-ID: <vmv6ac$22ufe$1@dont-email.me>
References: <vmum11$n7g$1@nnrp.usenet.blueworldhosting.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8; format=flowed
Content-Transfer-Encoding: 7bit
Injection-Date: Fri, 24 Jan 2025 05:52:30 +0100 (CET)
Injection-Info: dont-email.me; posting-host="667e4c1ceb183facb9d3e72e42ada600";
	logging-data="2193902"; mail-complaints-to="abuse@eternal-september.org";	posting-account="U2FsdGVkX1/OUveMZ/oP0fqHMwcd+imb9Jkr44xYXrI="
User-Agent: Mozilla Thunderbird
Cancel-Lock: sha1:w3u4/MhBP2UpkSPNkIX1Wxk5cbg=
X-Antivirus: Norton (VPS 250123-20, 24/1/2025), Outbound message
In-Reply-To: <vmum11$n7g$1@nnrp.usenet.blueworldhosting.com>
Content-Language: en-US
X-Antivirus-Status: Clean
Bytes: 4300

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?

> The circuit below was simulated in LTSpice 24.1.0 with all updates.
> There are reasons why I'm not sure I'd trust version 17.x of LTSpice to give the most accurate results but I haven't tested this
> circuit in any other version.

The accuracy of the results depends - first and foremost - on the 
accuracy of the Spice models. The only reason you could have to trust 
the LTSpice results would be that the real circuit gave the same results 
under the same conditions. John May has gone that far - you and I don't 
seem to have.

> After starting the simulation, wait until it has simulated about 7 seconds and then stop the simulation. Close the annoying log
> window which 24.1.0 will produce.
> You can now select a sample of about 100ms when it passes through 0dB (just under 2V) and run an FFT on it with Blackman-Harris
> window.

But why bother?

> This will show that this circuit is approaching 120dB down on harmonics.

For the circuit - as simulated. A real circuit will behave differently - 
it will have stray inductances and capacitances which are not include in 
the schematic and the device models merely approximate the actual 
performance of the device being modelled - mostly accurately enough for 
all practical purposes, but this map really isn't the territory.

LTSpice simulations can be a reliable guide to what won't work. They are 
less reliable guides to ultimate performance.

> So if you want to do better than that then you'll need an oscillator circuit which has better performance to begin with. You can
> then add gain control for constant level.

You do have to add gain control to get a constant output level, and you 
do have to do it without introducing too much extra distortion.

> I've not so far found an oscillator circuit in LTSpice with better harmonic distortion performance than this one.

Who cares. The distortion levels we are quibbling about are in the same 
range as the rounding error in the Spice numerical integration.

> You can, of course, try to use filtering to reduce the unwanted harmonics. Removing the first four harmonics produced by this
> circuit will get it approaching 130dB but don't ask me how that would be measured in reality or what it would cost to do so even if
> it can be measured.

You can get 20-bit A/D converters that sample quite fast enough to 
monitor a 1kHz oscillator, and you can do a Fourier transform on the 
stored output of the A/D. It wouldn't cost me any money because my 
current computer has an audio input that works that way, and some 
software from an old friend in Scotland that lets me use it as signal 
monitor. It would cost me time because I installed the software perhaps 
ten years ago and have only used it once.

-- 
Bill Sloman, Sydney