Accurate sine wave in terms of amplitude

[Terry Bites]

That DDS chip has an output stabilty quoted as "200  ppm/°C" not great, but then DDS is aimed at RF where no one cares much about precise amplitudes.

Brew up your own DDS, you don't need all that parameter register nonsense or a micro doodad to control it.

Back in the day, before low cost microcontroller platforms had been dreamt of, I made a sine source for an LVDT sensor using a lookup table on a M27C1024 eprom (still to be found on ebay and in my 90's junk box). I used a TCXO and some counters to create the address lines, 74LS590s but dont quote me on that. Driving an AD669 bolted to a REF01 10v reference. A fat buffer amp on the end and I had a 16 bit rock solid 10v pp waveform. It was overkill for my LVDT but fun to make. There's none of that control loop rubbish to worry about here.  A proud day in Terryland, no pay riase though.  Of course, you can easily tune it using the counter's clocks. Add your own precsion attenuators and gain. A very simple antialias filter might be needed to polish off remaining clock residue.

Calculate a Sine LUT here www.daycounter.com/Calculators/Sine-Generator-Calculator.phtml

[TimFox]

But I do know today that I am iterating options how to generate sine wave - as clarified in the initial message including an example circuit.

Quadrature oscillator might be "the fresh idea I was looking for". Just that none of the first examples I tried to simulate on LTspice didn't work. Like this one:
Probably a simulation related issue - I need to analyze it further...

Whilst that circuit has two waveform at 90', it also has 3 RC elements, so it is less suited to a knob controlled oscillator.
When simulating such oscillators, you need to adjust the idealized gain, to account for real opamp effects.
You then find a trade off between oscillator startup time and clipping of one node when stable.

Lab instruments (such as the Krohn-Hites) using quadrature oscillators require precision resistors on decade switches and precision capacitors on the range switch.
The OP's requirement is for a fixed, non-critical frequency, so matching resistors and capacitors using reasonable meters can give an acceptable solution for non-tunable frequency.

[sw_guy]

Thank you everyone who gave me ideas to start putting everything together. Based on preliminary simulations quadrature oscillator does exactly what I need. I also registered a few new circuits I can use in the future. One of them is precision rectifier.

Answers:

IanB>Do you develop software by any chance?
Yes.

PCB.Wiz>Do you need to adjust this generator across 1-10kHz and 200mV-5V ?
No. Fixed frequency and fixed amplitude.

xvr>Than use square waves. They can be considered as sinus with a lot of distortion.
I assume this was not a joke because square wave works, but in the final design there are (component level) limitations to use square waves.

Quadrature circuits started to oscillate on LTspice after enabling the "Skip initial operation point solution".

sw guy

[tggzzz]

Quadrature circuits started to oscillate on LTspice after enabling the "Skip initial operation point solution".

Realise that a frequency stable oscillator will have a high Q and take many cycles to stabilise. If your system requires frequency stability, that can be a simulation problem.

For system simulation, just use a spice voltage source set to output a sine wave. Only after the system works as desired with ideal signal, is it worth simulating with non-ideal signals.

[soldar]

I have never understood replies like tggzzz. Nothing valuable to say, full of negative load. Waiting for missing parameters. Oh well...

Suggested reply for the next time:
"It depends on the frequency response, output voltage and drive capability. Please give us more detailed specifications in order to give exact opamp examples. However, in general, I can say that typical opamps used for low-frequency applications, which are the common use cases for wien oscillators, are for example LT1632, LT1013, LT1001".

or.... just ignore the question.

Not referring to this specific OP but in general I am with tggzzz. I hate it when someone asks a question and is not even trying to be helpful or show interest. You have to extract information little by little, with much asking and effort. Yes, I can give it a pass but it is still rude on their part..