TL431 Oscillator

This one isn't too serious.

When shabaz did a blog on using a TL341 as a Low Battery Warning Indicator, I wondered what else you might use the device for. Since it includes an amplifier, it shouldn't be too hard to get it to oscillate, so that's what I thought I'd have a go at. See if I can get a sinewave out of it.

As I'm just doing this for fun, I'm going to slap together the circuit without too much detailed design and see what happens. I wouldn't recommend that you actually use this for anything, though, for a whole variety of reasons, not least of which will be the high current consumption.

Here's a picture of the breadboard I ended up with

and here's the circuit

This is using an LC circuit to determine the frequency. The tapped capacitor [C1-C2] for the feedback makes it a Colpitts oscillator.

I wasn't very sure at all of what the frequency response of the amplifier in the TL431 would be like, so it seemed safest to go for a fairly low frequency of oscillation, which meant having quite high values for the capacitors and the coil. I found a couple of old 2.2uF film capacitors [so, 1.1uF in series] and a 4.7mH [17 Ohms series resistance] inductor that looked like they'd suffice to get me going and would be easy to use on the breadboard. In theory, that makes for a resonant frequency of 2.213kHz, but the circuit will move that a bit because that's the frequency where the LC network itself looks real, whereas the oscillation criteria is that there be zero-degrees phase shift right round the whole loop, including the small delay through the regulator. [I think that's right: if you want a masterclass, best ask one of the experts here to do it rather than me; it's a long, long time since I last sat down in a lecture hall.] In practice, in my circuit, the frequency could be higher or lower than that because the coil and the capacitors will have a large tolerance on the values.

That LC network needs to be driven by the regulator amplifier, but we also need to have an eye for the dc biasing of the whole thing, so I'm going to do what you'd do with a JFET or a bipolar transistor and have the load up to the supply. That's a nice arrangement because current can run down through the coil to power the regulator.

Next I need to consider the regulator. I'm going to arrange it in a way that's analogous to what you would do with a common-gate or common-base circuit with a transistor. Firstly, I'm going to dc bias the output [the cathode] at twice the voltage of the reference pin by making the two resistors R1 and R2 the same. As the voltage between the anode and the reference will settle to 2.5V, that means the cathode will be at least 5V (in practice, more, because I'm going to have to lift the anode above ground in order to drive it as the input). The (ac) input voltage at the anode will then cause the difference between pins 2 and 3 to vary in the opposite sense to how it would if behaving as a regulator (the 100nF capacitor at pin 3 stops it moving around), so the loop, instead of countering the change through negative feedback, will reinforce it through positive feedback and, with a bit of luck, we should see the circuit oscillating.

Not sure I explained that very well, but the short version is that the dc bias is set with negative feedback [hopefully stable] and the ac operation is positive feedback [hopefully unstable, but in a nicely controlled sort of way].

Does it work? Er, yes, sort of! Here it is oscillating

Hardly the purest sinewave you'll ever see, but it's trying its best. I think I've got a way to go before I can claim to be an analogue designer.

Here's the FFT from my oscilloscope - lots of even harmonics there.

Does it work in the simulator? It did after I gave it a initial kick [by holding the LC circuit at zero volts at the start with the IC [Initial

Condition] symbol you can see on the circuit. I also selected the simulation parameter-set that is specifically for oscillators.

Here's the oscillation and the power supply current waveforms:

This is the measured average supply current for the breadboard circuit, which is reasonably consistent with the above.

Some scope for improvement there, if you felt inclined to play with it.

Finally, here's the open-loop response around the loop, from the simulator. This won't be very accurate, but it does nicely show the resonance peak due to the LC circuit and the fairly abrupt change in phase at that point [though you can see the effect resistance has in damping the peak].

If you found this interesting and would like to see more blogs I've written, a list can be found here:
https://www.element14.com/community/people/jc2048/blog/2018/10/24/jc2048-blog-index