Life expectancy of a transistor wired backwards?

[MaxPower]

The old simple sawtooth oscillator circuit: npn transistor with the collector to ground, no connection to the base, and the emitter through a resistor and pot to 9v, and a cap from the emitter to ground.

The Elektor schematic shows a 5.6k resistor in series with the pot and v+ from 5v to 20v iirc. I'm using a 1k resistor as it allows a wider frequency sweep. The circuit is picky about which transistors will work, but I happen to have a transistor type (model?) that seems to work. I've tried six so far and all six have oscillated.

Anyway, would hate to sell/give a synth drone/sequencer/noisemaker only to have it fail shortly thereafter.  Thanks.

[duck_arse]

funny that, exactly the same circuit I was werkink on tonight. I'm aiming at a modded tri-negistor from moosapotamus, just sorting transistor types now.

looking at the waveforms on the cro, some transistors are VERY erratic, in the good, jumpy, predictable-now way, and others just don't start. some start with the voltage wound up a few.

as for your question - erm - don't know. RG tells us a transistor junction degrades each time it's reversed biased, doesn't he?

what type number reliable starters did you find, if I might enquire?

[MaxPower]

PM sent. I wonder if hfe might be a reliable indicator.  The transistors in question have an hfe about 100 (+/-10) according to my dmm.

[PRR]

There are less abusive ways to make a sawtooth.

[GibsonGM]

There are less abusive ways to make a sawtooth.

But are they that easy?   

[Rob Strand]

I can't comment on the reliability of the saw-tooth circuit but when you use reversed bias transistors as noise sources or voltage regulators I've seen no tendency for them to fail.

as for your question - erm - don't know. RG tells us a transistor junction degrades each time it's reversed biased, doesn't he?

I'm pretty sure the context of RG's comment is:  Once a transistor is reversed bias it can become noisy when used in normal forward mode for audio.  If you drive more current in the reverse mode it tends to do more damage.

If you have ever had a fault on a circuit using LM833 opamps, which don't have input protection diodes like the NE5532,  you will know the circuit can become very noisy.  Some faults can take out a lot of opamps this way.

[anotherjim]

I understood RG's comments were for using the "Zener'd" transistor for normal amplifier duty too. And one used for noise or Zener doesn't get much worse in continuous use (first time is a charm). For noise generation, it would be great if they did get worse!
I take it that in the oscillator, it is doing what a neon was used for in tube circuits -  reset a ramp generator when a certain voltage is reached.
The transistor Zener effect doesn't work reliably at 9v and the minimum it can work will be nearer 7v -  depending on the transistor - that's why it's picky about transistors at 9v. 12v should be much better.

[duck_arse]

the moosapotamus page izz here:
http://www.moosapotamus.net/trinegistor.html

this schem is there:


if it was only a simple sawtooth you wanted, this isn't the method. the empty base Q1 in particular shows a few modes of oscillation, sometimes going f, 2f, 3f, then off, f etc. in steps. sometimes sweeping/ramping at audio rates, sometimes at lfo rates. so huge fun for rhythmic hunka chunkers.

as for gains - I don't think it is a reliable guide. age, maybe - get yer hands on some ~1980~ BC548's or BC178/9's or some Fairchild PN3566's. oddly, I could not get a single [Natsemi] PN2222A to start in Q1.

[late edit: ] there needs to be a decouple cap added before the output level pot.

[Rob Strand]

I could not get a single [Natsemi] PN2222A to start in Q1.

I think part of RG's "warning" is by trying out transistors you actually stuff them up and if you put them back into your stock it's at you risk.

Your post made me remember the Negistor article in Popular Electronics (Dec 1975, p69).   
From that I got a few more links:

https://web.archive.org/web/20020220180450/http://www.keelynet.com/zpe/negistor.htm
https://jasonacox.com/NDR/
https://www.physicsforums.com/threads/quiz-why-how-does-this-circuit-oscillate.657701/

Shows enhanced ckt:
http://www.cappels.org/dproj/simplest_LED_flasher/Simplest_LED_Flasher_Circuit.html


I don't if it's true or not but I couldn't help thinking that the transistor's "reverse beta" is a factor?   Mr. Cappel's enhanced circuit kind of points in this direction as well.

Reverse beta is the gain of a transistor when you swap the C and E terminal ie. you treat C and E and E as C.  The gain is quite low in this configuration, perhaps 1 to 20 but more like 2 to 7 or so.

The motivation here is you can measure reverse beta without damaging the transistor.

The reverse beta can be measured with your multimeter's hFE tester, you don't get much resolution though since DMMs  only display units.   Another caveat is transistor gain depends on the test current.   There's probably a best test current to reflect the necessary reverse beta in the negistor.  I  don't know if that's going to be at low current or high currents.

Anyway, something to think about if you don't want to kill your transistors.  If it works it would be a less mysterious way to choose transistors.   It might not be the only factor but it might be a necessary factor which can be used as a kinder way of filtering good candidates.

For example,

The transistor Zener effect doesn't work reliably at 9v

I've notice that as well, which made me give up trying to use BJTs as a noise source on 9V.
To start, best to play with these things at 12V to 15V;  negistor could be more finicky.

[MaxPower]

Thanks for the link and schematic. I tried one 2n3904, one bc549, and one 2n5088. None of them worked. I'll try 12v and 18v just to see if it helps. But I'm going to try another six of the transistors that worked and see how that goes.

Going by my ears they sound like a pretty consistent sawtooth throughout the frequency sweep. I have a cheap digital scope from banggood (dso-138 or some such thing). The scope shows a somewhat rounded ramp with a straight (vertical) drop as the transistor triggers and discharges the cap I guess?

[Rob Strand]

The scope shows a somewhat rounded ramp with a straight (vertical) drop as the transistor triggers and discharges the cap I guess?

If the ramp is rounding off because the circuit is an RC network.  The cap charges through the resistor,

https://image.slidesharecdn.com/transientanalysis-170603060752/95/transient-analysis-17-638.jpg?cb=1496470183

If the supply voltage is high compared to the zener voltage of the transistor the output waveform would 'aim' at a higher voltage and trip in the earlier part of the curve.   That would make the rising part more linear.

Another way to get a linear slope is to use a current source instead of a resistor.    An interesting question is if the resistor was replace by a current source would it work?  and would it help the chances of the circuit working?

FYI,  most current sources have a voltage drop so the supply needs to be a little higher than the zener voltage of the transistor.   Nonetheless, if you are aiming to get the linear part of the curve,  I'd expect a current source would end-up with a lower supply voltage than trying to get the same result with a resistor.

When you put a scope probe on the circuit the probe 1M (x1) /10M (x10) impedance will always cause some 'rounding' even if the circuit itself doesn't have any rounding.    Since the resistor much less than the probe's 1M/10M  you will be seeing the effect of the resistor not the probe.   With a current source you might see something with the probe.

[duck_arse]

the really mad transistors make all the difference. some types will work, but provide no interest in the tri-negistor circuit, whereas a whacko makes all the switches work. I have some very different cap values on the bb at the moment, and a 2N3566, 2N4401, 2N3566 or BC548.

oh - and the empty base Q1 is sussseptible to hum. if you touch the cro, or the ground line, or the transistor, or the V+. so you can add a piece of wire standing straight up, connected to base, and wave your hand at it for even more fun.

[MaxPower]

Great info Rob. I had wondered, is this setup a vco or current controlled oscillator? My guess is cco but what do I know? I came across a cco in some old electronics mag, I'll have to give it a study. It does sound like a pretty good sawtooth wave so maybe it's not as rounded as it looks. My scope settings may have squashed the waveform making it look rounder than it is.

Duck, I did notice the theremin potential when I setup a dual negistor. I didn't change my caps though (I was going for the detuned sawtooth sound in my circuit) so both oscillators were in the same frequency range. I didn't get any interesting results. Need to change cap sizes for a proper test. Anyway, tried another six transistors and they all oscillated (12 for 12). How about mixing in a cd40106 (6 oscillators)?

[Rob Strand]

Great info Rob. I had wondered, is this setup a vco or current controlled oscillator? My guess is cco but what do I know? I came across a cco in some old electronics mag, I'll have to give it a study. It does sound like a pretty good sawtooth wave so maybe it's not as rounded as it looks. My scope settings may have squashed the waveform making it look rounder than it is.

It's generally called a relaxation oscillation.     VCO and CCO  means the frequency is *controlled* by a voltage or current but might not indicate how the actual oscillator works.

http://hyperphysics.phy-astr.gsu.edu/hbase/Electronic/relaxo.html

FYI: The unijunction device (UJT) was built for the job of having a negative resistance.  It was mostly used as an  oscillator in with the circuit shown.   The output at the 'emitter' is a sawtooth, very similar to the negistor circuit and the output at base 1 is a narrow negative pulse.
http://hyperphysics.phy-astr.gsu.edu/hbase/Electronic/unijun.html#c1
http://www.electricalfun.com/WorkbenchFun/images/TThomas3_image3.png

I suspect the rounding filters some high frequencies.  However, it's not *exactly* like  low-pass filter as the sharp falling edge remains sharp so the filtering is less noticeable.  I'm not even sure if it's high or low pass filtering as we are using the RC ramp to get the linear ramp in the first place.  If it sounds good it is probably is good enough.