For our 2025 Component Abuse Challenge there have been a set of entries which merely use a component for a purpose it wasn’t quite intended, and another which push misuse of a part into definite abuse territory, which damages or fundamentally changes it. [Ken Yap]’s use of a transistor base-emitter junction as a voltage reference certainly fits into the latter category.
If you forward bias a base-emitter junction, it will behave as a diode, which could be used as a roughly 0.7 volt reference. But this project is far more fun than that, because it runs the junctions in reverse biased breakdown mode. Using one of those cheap grab bags of transistor seconds, he finds that devices of the same type maintain the same voltage, which for the NPN devices he has works out at 9.5 volts and the PNP at 6.5. We’re told it damages their operation as transistors, but with a grab bag, that’s not quite the issue.
We’ve got a few days left before the end of the contest, and we’re sure you can think of something worth entering. Why not give it a go!
Erm…a diode you mean?
A very unstable, unrepeatable voltage reference that is worse than Zener diode. Not only the breakdown voltage increases with temperature, but is also dependent on current and characteristics of a particular transistor. There’s a reason no one uses them as references. Also, quantum noise, which is often used in RNGs and noise sources.
There’s a reason no one does anything the way described in the stories submitted to this challenge. That’s kind of the whole point.
Isn’t the point of challenge to get something USEFUL from the abused component?
This reference, frankly, is not useful.
I once designed a dual channel DDS generator, with the idea of overclocking the main chips to 100MHz because that would extend the available frequency range at a cost of some additional losses, heat and higher noise and harmonics beyond the range. I also overclocked a PIC microcontroller just to see, what would happen. It ran faster. That’s useful, too. I’ve also seen somewhere a stereo headphone amplifier made of two TL431s. I even built it p2p style, and it worked very well. That’s a legit component abuse – going beyond the part’s parameters or applications.
There’s a nice collector-base diode with a negative TC in series with that positive TC reverse biassed base-emitter diode. Choose the right device type, do some sorting, and the TCs cancel.
Apparently, the use of any BJT in base-emitter reverse-bias breakdown mode damages the device for use in normal forward-bias mode by decreasing current gain and increasing leakage. I have been unable to find a physical reason for this action that covers all BJTs.
For any BJT, allowing too much current to flow in reverse bias mode might result in localized or widespread heating, disrupting the crystal structure or allowing rediffusion of dopants. But if current is limited? …
For planar BJTs, the explanation proposed is that the “hot” electrons create recombination sites (traps) between the silicon and the silicon oxide. I assume that once again disrupting the crystal structure is to blame.
What about grown-junction and alloy-junction BJTs? (Were these only germanium?) The active regions are buried; they can be (I think) constructed without any place for traps to form in an oxide interface. If so, and degradation still occurs, what’s the mechanism? The same question applies to BJTs that are not passivated.
Does anybody have good knowledge of this?
So basically a Temperature Controlled Voltage Reference TCVR.
Choose the desired voltage by increasing or decreasing the heat