Stable bjt voltage drop

Hello,

I need a bjt transistor which provides a stable voltage drop on the pn junction. Voltage drop must be stable against temperature, vce and ic. Vce voltages will be lower than 10 volts and ic is less than 1mA.
Bjt voltage drop will be used as voltage reference for current limiting, so NPN or PNP options are available.

Thanks

You cannot have this, generally - temperature is a prime
factor in junction forward voltage. At best you may find,
for some specific device, one "magic point" where Vbe is
temperature-invariant (due to competing trends in, say,
junction potential and series resistance) but this then
needs another proper-tempco current bias....

The scheme you allude to has been common in IC op amps
and the like, only without any great accuracy expected -
you will see gross variation in current limit limits on the
datasheets.

It may be that a current limit which gets lower as temperature
gets higher, is in fact a good thing (as it's meant to ensure
product reliability under fault conditions, and most reliability
detractors are strongly temperature accelerated).

As is often true, your requirements want sanity checking.
Picking an arbitrary behavior which looks good on paper
may be both unnecessarily difficult (or at least out of
proportion to its actual value) and potentially not even
the best behavior for what the purpose is.

Thanks,
that I was asking the most stable.
Someone knows a bjt with a "stable " voltage drop? [Reference]

Any silicon semiconductor junction forward voltage has a temperature coefficient of about -2 mV/K. Asking for a more stable voltage drop is simply ignoring laws of physics.

It makes sense to ask for devices with less type variation of forward voltage, specific leakage or current gain characteristics. But forward voltage temperature coefficient has to be either accepted as fact or must be compensated in special circuits.

Suggest to read about temperature compensated circuits.

FvM said:

Any silicon semiconductor junction forward voltage has a temperature coefficient of about -2 mV/K. Asking for a more stable voltage drop is simply ignoring laws of physics.

It makes sense to ask for devices with less type variation of forward voltage, specific leakage or current gain characteristics. But forward voltage temperature coefficient has to be either accepted as fact or must be compensated in special circuits.

Suggest to read about temperature compensated circuits.

Click to expand...

Okey, let's ignorate temperature effects. Just NPN or PNP with less variation of forward voltage.

Thanks.

Now you must quantify how much "less" you need. In
my experience the Vbe of older bipolar technolgies (as
used in high quality piece-part op amps) varied not much
and mismatches on the order of a few mV was the norm
(low and stable enough that laser trim at probe could
pull everything to center).

"Less" tells nothing without some quantitative reference,
and "enough less" requires a quantitative goal. With those
in hand you could challenge a given foundry to provide
the process control statistics and make your selection
if you find more than one.

The thing is, most of the really high quality (in the sense
of caring about and controlling this param; many of the
commercial BiCMOS technologies are mostly about the
CMOS and BJTs are opportunistic and unoptimized, or
made "just good enough" for general purpose but not
leading-specs parts) foundries are quite uninterested in
adding analog ASICs / custom ICs to the foundry - or
at least until you show up with the right number of
zeroes on the check.

In any case you want to be considering post-fab trim
as a means of getting to "perfect" - counting on the
fab for month-in, month-out consistency will eventually
disappoint you

The transistor you're looking for would have to be immune to thermal runaway. It will be big news the day it gets invented.

Here is a biasing method which may be useful. Articles say it has a stabilizing effect despite temperature changes. Suppose collector current rises due to heating. This draws current away from the bias, causing collector current to drop back down.



You'll need to experiment to find which combination of resistors works best, whether resistance should be high ohm, or low ohm.