interesting negative resistance circuit

[ledtester]

I found this on circuit on the electronics stackexchange:

https://electronics.stackexchange.com/questions/528743/constant-current-limiting-circuit

It was shown to exhibit negative resistance for a range of input voltages and because of that it can oscillate a tank circuit.

Attached are LTspice files.

The original used a 4.7V Zener in the BZX84C series. The closest I found in LTspice had was the 6.8V version. Does the 4.7V model exist for LTspice or is there an way to synthesize/derive it?

[mawyatt]

Interesting circuit, thanks for showing!!

To answer your question, you could create your own Zener diode model for a ~4.7V breakdown, or use the LTspice 1N750 model.

Best,

[razvan784]

Also works in a simpler configuration (no Zener, no cap) and at lower current:

[magic]

This is what "DC sweep" simulation mode is for

[mawyatt]

Does work indeed in real life!!

Here's with DSO & AWG (with buffer amp), then with Tek Curve Tracer!!

Noise on DSO plot is due to crappy 12V SMPS supply used with AWG buffer amp to get to 20V peak unipolar ramp (AWG can only supply 10V peak unipolar ramp).

DSO & Tek Curve Tracer plots are 2V/div Horizontal and 1ma/div Vertical.



Anyway, fun circuit and thanks for posting.

Best,

[mawyatt]

This is what "DC sweep" simulation mode is for

Yep.

.dc V1 0 20 0.01

Best,

[T3sl4co1l]

For further reading, see: foldback current limiter circuit.

Tim

[Benta]

I hope everyone realize that this is actually unwanted, uncontrolled, parasitic oscillation?
I mentioned it in some other threads on emitter-follower type LED current limiters.
Oscillation frequency depends purely on parasitics in the two transistors and cannot be modified or controlled in a sane way.
Also, the transistors are partially forced inot saturation-, so DC biasing shifts unpredictably.
Not useful in practice, but a nuisance in real life.

[mawyatt]

Agree it's not a circuit one would normally use as an oscillator, but will oscillate at the ~LC parallel resonance over a range of L & C. This will begin to oscillate with a voltage across the network of ~1.8V out to beyond ~8.5V, then hysteresis entered and it will not oscillate until the voltage drops to ~6.4V.

What's happening is as the DC voltage is increased to 1.8V the circuit exhibits enough effective negative R (-R) to kick off oscillations, once the oscillations start the the waveform swings above and below the DC set bias point. As the bias voltage approaches and increases beyond the point where the -R subsides (~ 6.4V) the waveform "swings" into the + R region, but enough of the waveform swings into the -R regions so the waveform cycle average is still effective -R. As the bias voltage increases eventually reaching ~8.5V where the waveform average is no longer -R (more into the +R region than the -R) and the waveform collapses and oscillation subsides. Dropping the bias voltage back towards zero, and ~6.4V oscillation start up again.

We've confirmed this behavior by measurement.

Anyway, interesting circuit with some interesting behavior, but maybe not very useful, and reminds of the chaotic oscillator based upon the Chua Diode

Best,

[Benta]

Several IEEE papers have described this unwanted behaviour, which happens in emitter-followers fed from a negative impedance.
The most famous instance (I believe) was the Apple II, which had a discrete crystal oscillator at a few MHz for the main CPU/graphics clock.
What no one noticed was, that the clock generator additionally whistled along at over 50 MHz due to this parasitic oscillation.
Don't remember if there was a recall, but it created FCC trouble.

Thankfully, the fix is easy.

[TimFox]

If you really need a negative-resistance oscillator that depends neatly on an external L-C resonant circuit, try a transitron with a 6AS6 tube.
https://books.google.com/books?id=e_oZ69GAuxAC&pg=PA79&lpg=PA79&dq=transitron+oscillator&source=bl&ots=v-nK1uhV3n&sig=ACfU3U0HN6HCz8jiyvgkR2MbEeWDbD7lig&hl=en&sa=X&ved=2ahUKEwioksu7yZr7AhUBpokEHfQlBvQ4ChDoAXoECBAQAw#v=onepage&q=transitron%20oscillator&f=false

[mawyatt]

Several IEEE papers have described this unwanted behaviour, which happens in emitter-followers fed from a negative impedance.
The most famous instance (I believe) was the Apple II, which had a discrete crystal oscillator at a few MHz for the main CPU/graphics clock.
What no one noticed was, that the clock generator additionally whistled along at over 50 MHz due to this parasitic oscillation.
Don't remember if there was a recall, but it created FCC trouble.

Thankfully, the fix is easy.

Aware of the -R with Emitter Followers, serious issue with very fast SiGe bipolar transistors if one doesn't control the impedances "Seen" from the base as well as the emitter. In the SiGe case this is caused by the excess phase shift in the BE junction, we often included a series resistor with the base just for this purpose!!

Didn't know about the Apple II issue tho, that was our 1st real computer. We developed a Digital Scope and Spectrum Analyzer for this when it first came out. This was way back when Apple provided the schematics, BIOS and everything, then with the intro of the Mac took an opposite position and provided almost nothing!!

Best,

[Benta]

Well, the base resistor is the only fix that works reliably.
I wouldn't want to fiddle with SiGe on this, just a simple BC547 or 2N2222 makes trouble.   

[David Hess]

Well, the base resistor is the only fix that works reliably.

There are all kinds of ways to stop emitter follower oscillation.  For tough cases I have gotten the best results by placing a ferrite bead in series with the emitter where the current is high enough to take advantage of it, but a base series resistor is usually be the best option.

https://www.eevblog.com/forum/beginners/noise-on-emitter-follower/msg1297122/#msg1297122