Diode-Stabilized Wien Bridge Oscillator

In this video, we present a sine wave generator called a diode-stabilized Wien bridge oscillator and show how to use LTSpice to analyze it. We then build the circuit on a breadboard, run it, and evaluate it using an oscilloscope. Can it produce a high-quality sine wave?
ADALM2000 (M2K) is not used in this video.
The author is not affiliated with Analog Devices, the maker of the ADALM2000 but rather is a user of it who thinks that it should be of interest to students and electronics hobbyists.
See also @Circuitsfromthelab ( / @circuitsfromthelab ) for videos on installation and basic usage.
Main page from Analog Devices:
www.analog.com/en/design-cent...
Suggested activities from ADALM2000 community:
wiki.analog.com/university/co...
Information on ADALM2000 hardware:
wiki.analog.com/university/to...

[@LutzSchafer]

Nice, I see you put a lot of effort into that. I remember in university, the Wien bridge oscillator was indeed a subject of discussions. In this example you need to make sure to "ride" on the forward knee of when the diodes just become conducting. This function is as you say nonlinear (approximately a parabola). Your second example works better because R5 linearizes this piece of the curve. Since I am no longer working in electronics development I haven't looked into newer versions. I am sure this control problem can be solved with arbitrary precision within a specific error correction time when the feedback is made with an integration component (which a lamp provides). The problem is similar to AGC, only requires enough precision to keep the oscillation criterion.

[@fer_fdi]

Superb!

[@aduedc]

Thanks,
It is great topology two diodes and resistor in feedback loop.
Also, you explain things really well.
I guess, we should be able to use this configuration (with some modification) in other types of oscillators to stabilize them.
By the way these two diodes with a parallel resistor are used in PLL filters for fast convergence. When the voltage of phase detector is too high, the diodes turn on and as you approach
low phase difference then the diodes turn off and resistor takes over.
Also, you should look at phase noise.

[@kensmith5694]

A few improvements for people to consider:
1) if the feedback part of the circuit is also tuned to have the gain of about 3.000 only at the running frequency, the high harmonics from the diode action will be suppressed.
2) Consider using "low leakage" diodes as your diodes. BAS116 is an example. The resistor in series is over 100K so you are working the 4148s at a very low current. The BAS116 has a higher "small signal resistance" at the zero bias point.
3) As well as #2 a resistor to ground from the end of the diodes can help. This lets you get the diode current and the extra feedback current to be different values so you can adjust them for best performance.

[@cocusar]

Is this documented somewhere? I'd like to give it a read, and/or a test!

[@kensmith5694]

@@cocusar Do you do LTSpice?
I could put together a model for you without too much effort. each point in my comment is fairly easy to model.

[@cocusar]

@@kensmith5694 sadly I don't, but I can easily learn. just wanted to know how that works for pure sinusoidal oscillators, because I might need one of those

[@kensmith5694]

@@cocusar No sine wave is ever pure. Diode based amplitude control doesn't do as well as the lightbulb. Thermistors are an alternative to the light bulb. Get your self a copy of LTSpice and play with it. To get you started, I will suggest you watch some videos about using it and then enter the design for an oscillator. Stick a 10mV pulse in series with some capacitor to kick start it
Voff=0
Von=0.01
delay=1uS
t-rise=1nS
t-fall=1nS
t-on=1E9

[@electronics.tinker]

Commenting on my own video.... My scope has 8-bit vertical resolution so it is hard to believe that it can reliably measure harmonics more than 48 dB down from the level of the fundamental. Two of the measurements in this video show no harmonics above this level. That suggests a clean sine wave relative to what can be expected with diode stabilization-- but don't be convinced that the total harmonic distortion is really less than 1%. The scope may not be showing it. This is a complicated topic...
the M2K has a 12-bit DAC. Could it be better than the scope? I doubt it. I suspect its other limitations would come into play. I'll make a video on this if I can figure out how to say something reliable given my test equipment limitations.

[@kensmith5694]

You can also make a "home made subtracter" or notch filter. Both methods allow you to reduce the amplitude of the fundamental by about 20dB without changing the harmonics. Since the full circuit is an oscillator and not an amplifier, chances are the notch filter method will work best. Buffer the signal and then use expensive capacitors and pots to make a twin-T

[@electronics.tinker]

Thanks for these interesting comments. Filtering the fundamental is an interesting idea. An effective filter design for the low frequency I am generating might be tough, though. My scope (Siglent 1104x-e) unhelpfully gives ">= 35 dB" as its SFDR spec. I think you have better equipment!

[@kensmith5694]

@@electronics.tinker A filter at lowish frequencies is very simple to make. Look up twin-T filters. You can get large value film capacitors up to 15uF on a budget you mere mortals can afford. If you need one that is easy to tune, the fairly common 3 op-amp filter design can be made to have a linear frequency with the setting of a 10 turn pot with a knob on it. There are two tricks to doing this:
1) Put the knob on so it goes 1.00 to 11.00
2) Use the resistance from wiper to each end as two gain determining resistors so that the knob's setting effectively gets squared.