Reversing collector and emitter on a PNP 2

[Comcokid]

I am working with a product problem. This industrial product was designed around 1970, and consist of a mixture of discrete transistors, and LM741/LM308 opamps. In other words, the engineers who designed it are long gone, and the circuit approach is not-at-all what would be done today. But I need to fix it with minimum changes.

Through shear accident, it was discovered that the problems with the design could be fixed by reversing the collector and emitter on a 2N3906 PNP. This transistor is part of a multivibrator circuit (NPN and PNP in a SCR-like arrangement) being fed from a 11 ma constant current source and capacitor through a double emitter transistor (to create a ramp). Reversing this transistor keeps other opamps off the rail and solves temperature drift problems of the circuit as a whole.

Here is the issue. I can find little data on using a PNP transistor with collector and emitter reversed, just a Wikipedia reference about how the gain is much, much lower. Since the transistor is driven by a current source it is protected. What other information can anyone tell be about using a PNP transistor with the collector and emitter reversed? What other characteristics do transistors have when used this way?

 

[itsmoked]

They are symmetrical with the following exceptions:

1) Leads sometimes change position in manner that prevents easy reversal via simple insertion changes.

2) The collector is the die's base substrate and is the big thermal connection to the package a reconnection/insertion may reduce the thermal capabilities.

3) The gain h_fe is reduced. Not all that much! I have never had an application that cared a twit about the difference.

4) Because of the die geometry the transistor's speeds will be a little different.

Generally short of some ragged edge application, it shouldn't matter much.

Keith Cress
Flamin Systems, Inc.-

http://www.flaminsystems.com

 

[nbucska]

You get beta<1 but very low Vsat on the enitter (acting as Collector}

Plesae read FAQ240-1032
My WEB: <

http://geocities.com/nbucska/>

 

[MacGyverS2000]

The gain hfe is reduced. Not all that much! I have never had an application that cared a twit about the difference.

Keith, you serious on this one? I would expect gain to drop like a rock, but I can't say I've ever tested the theory...


Dan - Owner

http://www.Hi-TecDesigns.com

 

[itsmoked]

I messed around with a Tektronix curve tracer and some 3906's, , 3904's, and 2222's and was surprised by how little difference I saw. I also have miss-stuffed boards with no change in function. Now some RF transistor or high speed amps would more than likely have problems so I wouldn't go out of my way to reverse transistors.

Keith Cress
Flamin Systems, Inc.-

http://www.flaminsystems.com

 

[IRstuff]

> NPNs can take about a 40:1 hit in beta in reverse mode. The discrepancy may be at the current that you measured the beta at.

> Beta will also probably take a bigger hit over temperature

> The VBcbo will be drastically reduced, as will the VBceo

> The base resistance will go up

> Miller capacitance will be MUCH worse

> Collector-base leakage will be worse

TTFN

 

[OperaHouse]

That reminds me of something that I have wanted to test again for more than 40 years. Back when I was only about 14 I read in article about making a reversible train power pack. Just a step down transformer, pot and power transistor forming a half wave rectifier. Must have been a PNP germanium from the age. Pot was connected from C to E and the wiper went to B. I remember it had a lot more gain in one direction in the other.

If what everyone else is saying is true, it should have conducted in both directions and not have been a rectifier you could reverse.

 

[jimkirk]

I remember many years ago hearing about a circuit, I think it was developed by Bell Labs, that was built with one transistor and two resonant circuits. It was an oscillator that would operate at one of the two resonant frequencies, depending on the polarity of the power supply.

 

[IRstuff]

The only time you can get close to equal forward and reverse gains is if the transistor is a lateral transistor with base width photolithographically defined. But then, the gain is crappy in both directions ;-)

Vertical BJTs are inherently asymmetrical, just by virture of the fact of that the emitter/collector area and doping profile differences.

TTFN

 

[Comcokid]

Thanks everyone. This got me the answers I needed. When you tell higher managers that the answer is to 'install the transistor backwards', you need technical info to backup the resulting political flak.

 

[analoguealan]

I remember production line testing linear adjustable bench power supplies around 1970. One of the problems was identifying germanium transistors in the control circuitry that had been wired in the wrong way around.
The problem was they were simple alloy junction transistors where the collector and emitter regions were the same size. The voltage and current capability was equal either way around. Hence no electrical test could identify a line wiring mistake. We just had to do a thorough visual exmination.

Unlike modern silicon planars were the junction widths are delibrately asymetric to get good voltage breakdown between base /collector.
The base emitter has very thin region to,
Increase transistor efficiency by easing the flow hole/electrons into the collector region.
Reduce capacitance and hence improve high frequency performance.
Thats why modern silicon planars only hav a reverse base emitter breakdown capability of of about 5 to 7V. The old germaium alloy junctions could take about 30V reverse base emitter, same as the Vcb.

Analogue Alan