Op Amp sine wave oscillator at 50KHz is not easy to do?

[treez]

Hello,

I just read.............

https://www.ti.com/lit/an/sloa060/sloa060.pdf

And Op Amp sine generators, just to produce a 1V pkpk sine are very hard to do.....you need AGC etc etc.

Isn't there an easy way to get a sine?

 

[FvM]

Everything depends on the intended sine quality (distortion, amplitude accuracy). Generally I don't agrre about "very hard to do".

 

[treez]

i dont need good quality..may i confirm that the circuits relevant are "wien bridge" and "quadrature oscillator"

The wien bridge seems to need a lamp to be inserted in the circuit at some point?

 

[Syncopator]

The wien bridge seems to need a lamp to be inserted in the circuit at some point?

Wein, not wien.

The Wein bridge oscillator needs a.l.c. (automatic level control). It may be done with a small lamp, a thermistor, or a f.e.t.,

The commonly used thermistor, RA53, is hard to find nowadays so I would recommend the f.e.t. method.

 

[FvM]

Wein, not wien.

How do you know? It seems like the Wien oscillator invented by Max Wien is sometimes spelled Wein in English literature...

https://en.wikipedia.org/wiki/Wien_bridge_oscillator

P.S.: Regarding the technical question, the oscillator can work without amplitude stabilization, but the distortions are possibly unwanted high. A 3RC low-pass passive phase shift oscillator gives acceptable distortions when the output signal is accessed after the low-pass.

 

[LvW]

Ask the inventor of this oscillator type Mr. W.R. Hewlett. He used the WIEN bridge.
(Added later: This is an answer to Syncopator`s post#4)

 

[treez]

Page 63 of "practical oscillator circuits" says enough for me to think that the Wien bridge is pretty well unusable in its simple form............as its output cannot be stabilized , and it goes off to the rail.

Page 64 of the same book provides a Wien bridge schem with a jfet stabilizer, but its for a single supply use, and i need dual supply, as i want the sine to go pos and neg, so i am now seeking a Wien schem with dual supply op amps, and a fet stabilizer.

Is the quadrature oscillator any better (simpler) at producing sines............?

 

[LvW]

i dont need good quality..may i confirm that the circuits relevant are "wien bridge" and "quadrature oscillator"
The wien bridge seems to need a lamp to be inserted in the circuit at some point?

Treez, if you don`t need "good quality" - and if you can afford three opamps - I recommend the quadrature type consisting of two inverting integrator stages and one inverter stage (unity gain).
Here are the advantages:
* Normally, you don`t need an amplitude control (lamp, FET) because there will be only a slight clipping of amplitudes (the proof of this property cannot be found in textbooks).
* It is not a problem to design it for the required amplitude, see explanation below (this requirement is not easy to fulfill with the amplitude control in case of a WIEN oscillator)
* The frequency can be adjusted to the desired value without touching the oscillation condition.

Thus - for a non-experienced beginner the design process for this oscillator type is not complicated.

Some explanations:

a) Equal time constants for both integrating stages: T1=T2=T; wo=SQRT(1/T)
The oscillation amplitude at both outputs A1 and A2 will be approx. equal to the supply voltage Vcc=A1=A2 (in practice somewhat below due to opamp properties)

b) To get the desired amplitude, make both time constants unequal.
In this case: wo=SQRT(1/T1T2) Example T2>T1: Amplitude A1=Vcc and A2=Vcc*SQRT(T1/T2).
Thus, using different time constants you can realize the desired oscillation frequency as well as the desired amplitude.
This property of the quadrature oscillator is not mentioned in many textbooks.

Good luck.
LvW

 

[treez]

Heres an quadrature oscillator for 20KHz.................not enough for me...i need 50KHz and say at least 1.5V peak.

https://www.ti.com/lit/ds/symlink/4423.pdf

................for 50KHz, would the MCP601 be ok in a quadrature oscillator.

https://ww1.microchip.com/downloads/en/DeviceDoc/21314g.pdf

...the 2.5V/us slew rate seems a bit slow for 50KHz sine?

 

[keith1200rs]

You can stabilise a Wien oscillator with a couple of diodes and extra resistor in the feedback. It won't be perfect, but can produce a fairly decent sine wave.

For a fixed frequency, a filtered square wave works quite well.

Keith.

 

[LvW]

Treez,

here comes another advantage of alternative b) I forgot to mention in my post 8:
In case of unequal time constants only the output with the larger amplitude (approx. Vcc) is limited (clipped).
The other output is a lowpass filtered version of the other one and has a much more better quality.

 

[treez]

Thanks LvW,

https://www.ti.com/sc/docs/apps/msp/journal/aug2000/aug_07.pdf

i just tried to put the quadrature oscillator of this article (above) into ltspice, using the LT1006 opsamp, but couldnt get it to work...............

I was using a split supply, so didnt have the 2.5V level that they had......but it didnt work anyway..............i am wondering if the quadrature oscillator is a finicky circuit, needing lots of fiddly adjustments to get it to work.?

 

[LvW]

Hi treez,

I am wondering if the quadrature oscillator is a finicky circuit, needing lots of fiddly adjustments to get it to work.?

No, just the opposite is true.

Two comments:
1.) The chosen opamp is not "good" enough: Very low slew rate (0.5V/µs) and a GBW<1E6. For Fo=50 kHz you need at least a GBW> 3...5 MHz.
2.) I would not recommend both quadrature types as given in the referenced paper (Fig. 8 and 9).
The circuit in Fig. 8,indeed, needs carefully matched components and the circuit in Fig. 9 needs 4 opamps.
3.) My recommendation: Use a suitable opamp and the circuit as proposed in my post#8 (two inverting integrators and one inverter in a closed-loop). I promise: It works!

 

[treez]

Thanks,

Ive just found a quadrature oscillator schem in the tl071 datasheet that works but it doesnt give the equation for how to get the sine frequency.

Presumably one gets the AB transfer function, and equates it to -1, then uses complex numbers and s = jw to get the frequency?.........then one needs a formula for the amplitude.....and to know if it will clip or not?

......oope sorry LvW i now see your explantion of amplitude, but to which schem does this apply?.is it to any of those that consist of two integrators and an inverter?

Ill look for the 3 opamp one you kindly speak of

Youre right , ill change the LT1006.

i wonder if 2.5V/us is ok for slew rate for doing 50KHz sine?




..I found the article here on the web (below) which states a schematic and how to find the frequency for it.....but it doesnt say how you work out the amplitude.............i wonder how to work it out?..i need to know otherwise it may clip?

Research article
"Quadrature oscillators using operational amplifiers"
by
Jiun-Wei Horng

 

[LvW]

i now see your explantion of amplitude, but to which schem does this apply?.is it to any of those that consist of two integrators and an inverter?
..I found the article here on the web (below) which states a schematic and how to find the frequency for it.....but it doesnt say how you work out the amplitude.............i wonder how to work it out?..i need to know otherwise it may clip?

When you read again post#8 you will see that I gave you the formula for the oscillation frequency as well as the recommendation (formula) how to set the desired amplitude.
The oscillation amplitude depends on the supply voltage and the time constant ratio that can be adapted to your needs.

Treez - just for your information: The oscillator as described in the paper mentioned by you is NOT the circuit I have proposed to you.
It is a third-order circuit and I propose to use only two simple integrators (and one inverter).

 

[treez]

Thanks LvW.....now all i need is the schematic.(2 integrators and a inverter)........this one below doesnt work....

https://i45.tinypic.com/33tjpro.jpg

 

[LvW]

" does not work" - what does this mean?

Did you provide a little "help" to start? (Initial condition or a short voltage pulse at one of the non-inv. inputs.)***
More than that select a sufficient resolution (e. g. <10us) for a 100ms simulation time.

*** Or don`t use continuous supply but switch one of the supplies on at t=0.

Sorry - but now I am off for today.

 

[treez]

This is the only sinusoidal oscillator i have got working..its based on the quadrature oscillator.......none of the schems i have found on the web can i make to work.

https://i48.tinypic.com/4l1rb.jpg

...is this ok?...how do i calculate the amplitude from it.?

.....the sine wave only occurs at the output of the follower.

It only gives one sine...but i only need one......................at least the frequency is calculable from f = 1/2piRC, but the amplitude has been hit and miss.


as LvW says, I see that noise plays a part in quadrature oscillator........and maybe this is why the simulator is not making them work.....but it made the one in the above picture work so it is confusing.



***************************************

OK here's the 50KHz sine oscillator..one sine clips (U2) but the other is ok.......................theyre in quadrature.........
https://i46.tinypic.com/20hllbs.jpg

.............i put the two inverting amps (unity gain) in so that i could change the amplitude without changing the frequency..................but it changes both amplitude and frequency if you adjust the gain of either of these........................do you know how i can change the gain without changing the frequency?...the frequency is nice at 50KHz now, and i want to keep that.

............the oscillation frequency is by 1 /(2*pi * sqrt (r3*r5*c2*c3))

 

[FvM]

All of the shown oscillator circuits work in real live (if you use suitable OPs) and can "work" in a simulation, e.g. the simple quadrature simulator in post #16.

The circuit shows a rather slowly rising amplitude, so you'll want to use an .IC statement or similar technique to make it start from a non-equilibrium condition. That's the simulation specifique point.

It's more interesting in my view to think about the conditions for low distortion. It's your decision which stage(s) is clipping - by selecting stage gaina. Clipping in front of the dual integrator gives least distortion. A quadrature oscillator which low excess phase will show slow amplitude rise and respectively low clipping degree.

Instead of collecting more oscillator circuits from the internet, you should try to understand the principle behind it and learn to calculate the circuits yourself.

P.S.:

how do i calculate the amplitude

Starting from the maximum undistorted amplitude, defined by the stage that clips first.

 

[treez]

Well....its the old story all over again...........if in doubt, refer to Babani books............."how to use op amps" by E.A.Parr, PAGE 56, 57 (this book cost me £3.99)

the below is the best quadrature oscillator circuit that i have found ALL DAY!!!!!!!!!!!!!!!!!

The frequency is simply the frequency of the low pass filter, and the amplitude is simply the size of the zener....and only two op amps.....................it even gives 2 sine waves in phase quadrature....................instead of one sine wave being clipped.

..............The circuit explanation.....(as always with Babani books), is beautifully simple..........and it goes like this............

the opamp integrator gives a phase delay of 270 degrees........................then the low pass filter opamp gives another phase delay of 270 degrees............thats equivalent to a phase delay of 180 degrees..........and so when ever a loop gain of unity occurs.we have oscillation at the filter corner frequency...............

.....and having the zener in there ensures that the loop gain is unity.

QUADRATURE OSCILLATOR SCHEMATIC
https://i46.tinypic.com/wnz1t.jpg

frequency = 1/(2*pi*sqrt [(R4+R3).R2.C1.C2])


....the oscillation frequency is not affected by changing R1 and C3, as long as the loop gain is > unity.



Any better and simpler and more robust quadrature oscillators gratefully accepted please.......................................must have frequency of oscillation calculable from the components and amplitude calculable from the components, as with the Babani "special.................please feast at your leisure.