1n4571 vs 1n4574

[cnt]

I decided it would be fun to see how stable I could make a zener voltage reference.  I have no actual needs though so I'm not going to spend a ton of money on it (famous last words for every project I've ever done).  I figure I'll keep it at work as a really boring desk-toy and occasionally measure it when I have the nice meter out.

I think a 1n4571 fits the bill of only 5$ and fairly good but I see the 1n4574 the same "typical" characteristics but worse, worst case.  It's also 20$.  Are these the same just binned around the specified current?  Can I tweak the diode current up and down while adjusting the temperature to find the TC=0 point and just set it there?  Also they come in a -1 option with an "internal metallurgical bond".  Is that good? Do I want that?

Are there other diode options that are better drift and Tc per dollar? 

[Kleinstein]

The difference between the diodes is bining and maybe also more accurate testing. Chances are that the good ones end up as higher grades and the chance to find a good one (zero TC at 1 mA) is relatively small.  In most cases one will be able to get zero TC at a different current (e.g. 0.2 mA or 2 mA). A lower current would naturally lead to more noise and a higher current may be more effected by the series resistance and thus a slightly more critical resistor to set the current. With the low current to start with this would be less of a problem.
With the different related types also for higher / lower current binning may be even more complex.


The classical series of temperature compensated zener diodes is the 1N821 ... 1N829. These are for higher current (7.5 mA) and thus usually lower noise. A down side is that going to higher current with a 1N821 can become tricky. So there there is may very well happen that zero TC is at too high a current to be useful. As far as I have heart from 1N825 on it is usually possible to get zero TC at a reasonable current.

With these diodes there is anyway some 2nd order TC and for a really stable voltage over a larger temperature range (e.g. +-5 K or more) a stabilization of the temperature would be the logical way to go. With a good temperature control the TC is no longer that critical and some 50 ppm/K (e.g. 1N4571 or 1N823) could be usually good enough.