The transistor degradation did not only happen at high currents, but also a rather low currents like 10 µA. The article also mentions a change in the breakdown voltage of up to 20 mV, which would be bad for a reference. So there seems to be no truely safe level, just slower degradation. Another interesting point mentioned in the article is that the gain degration is reversible with annealing. So there may be positive effect of operating at a higher temperature, as it is faster to espeblish an equilibrium between the generation of new defects and the annealing.
Another problem with using transistors is that there is quite some scattering on voltage were they actually break down. So chances are only a small percentage may work and testing may be kind of destructive, as the avalanching also tends to increase the transistor noise.
Dual BJTs are getting rare and hard to get. IN comparison the 2DW23x are relatively easy to get.
The termal coupling is one of the smaller problems - so just single transistors should be good enough. No real need to destroy expensive matched pairs for this.
A somewhat interresing point may be annealing part. Here the use of a transistor can be an advantage, as one can measure the annealing part via the gain and thus without generating new defects.
Home
Help
Search
Login
Register
