Optimizzin Al-gorithym <[EMAIL PROTECTED]> wrote: > You can also use common guard structures to isolate the "HV" part of > the chip, without dicking with the Delicate Recipes (process) which > you Don't Want To Do And Probably Wouldn't Be Allowed To Anyway. > Also helps keep digital switching noise out of the source.
True, but if the customer wants low supply voltages, you're screwed. Consulting a process guide I happen to have handy (it's a pretty recent fast bipolar process), emitter-base breakdown is quoted at 2.6V nominal, so I guess from anything greater than 2.6V + Vcesat supplies (maybe 3V, certainly 3.3V, 5V), you're all set. Of course, if you're willing to sacrifice some current protection, put resistors on your wafer, and reduce PSRR, you could go as low as 2.7V for the cost of one 10 ohm resistor. > Actually, we're interested... I don't have any detailed data on it handy, but there are some indications in the same process guide that emitter-base breakdown carries with it some pretty serious hot-carrier problems. I believe that over time this effect drops off (after the worst of it has had its run), so the transistor won't continually degrade. Furthermore, most problems associated with hot-carrier effects won't concern us if the transistor is being run only in reverse breakdown---you'll lose some beta, BVeb might go down a little, etc. Thus, you'll almost certainly want to sacrifice the headroom and use a mirror to drive the emitter instead of a resistor (else your breakdown will increase over time, which will probably make the degradation more severe), but you should be OK. If I ever get the opportunity, I'll do some emitter-base burn-in testing and see how the entropy of the output is affected by hot-carrier-induced transistor degradation. > man, you don't want to have had too much coffee trying to land the > probes.. ..looking at analogue measurements with spectral analyzers > and sampled data with statistical tools) I've done some probing with way too much coffee. The worst was probing MEMS devices that way. Man, oh man. :-) -- Riad Wahby [EMAIL PROTECTED] MIT VI-2/A 2002
