Hi All,
The first pic shows the Vishay 500mW Zener Diode datasheet. The second shows the same table with the tempco converted to V/degC.
There seems to be a generally accepted notion that Zener diodes in the 5-6v range (highlighted in red) are the most thermally stable. I have read this on countless reputable websites and assumed it as fact for a long time. Where does this come from? Unless I do not understand the attached datasheet properly (highly likely), it would seem these have a pretty reasonable positive tempco. The only thing I can think of is that this has been derived when using them in some kind of BJT voltage regulator, where they will in fact neatly cancel the -2mV/degC Vbe tempco of the BJT.
However, it seems to me that the 3.9V and 4.3V Zeners (1N5228 & 1N5229 - highlighted in yellow) have a near equal but opposite tempco. If one wanted a very stable voltage reference, the best thing to do might be to run an alternating string of these in series. This of course assumes each device is kept at the same temperature, an issue that applies to any multicomponent voltage reference.
Any thoughs?
Cheers,
Greg
The first pic shows the Vishay 500mW Zener Diode datasheet. The second shows the same table with the tempco converted to V/degC.
There seems to be a generally accepted notion that Zener diodes in the 5-6v range (highlighted in red) are the most thermally stable. I have read this on countless reputable websites and assumed it as fact for a long time. Where does this come from? Unless I do not understand the attached datasheet properly (highly likely), it would seem these have a pretty reasonable positive tempco. The only thing I can think of is that this has been derived when using them in some kind of BJT voltage regulator, where they will in fact neatly cancel the -2mV/degC Vbe tempco of the BJT.
However, it seems to me that the 3.9V and 4.3V Zeners (1N5228 & 1N5229 - highlighted in yellow) have a near equal but opposite tempco. If one wanted a very stable voltage reference, the best thing to do might be to run an alternating string of these in series. This of course assumes each device is kept at the same temperature, an issue that applies to any multicomponent voltage reference.
Any thoughs?
Cheers,
Greg
Last edited:
One possible source is NXP's datasheet for their BZX84 range. On this, the typical tempco changes sign between 5V1 and 5V6.
Vishay's data around 4.3V looks rather anomalous - the tempco jumps about a bit. Whereas on NXP the increase is smooth, above 3.9V. The red box highlights the min, typical and max tempco data in mV/K.
I found an open-circuit diode in a power supply that I was repairing.
When I looked up the part number, it turned out that it was a voltage reference comprising a 6v ish Zener diode in series with a normal diode.
it measured open-circuit because there were two back-to-back diodes in series, one being a Zener.
The idea behind this arrangement is that the tempco of a Zener of this voltage can be exactly compensated by the tempco of a normal diode, the two being equal and opposite.
When I looked up the part number, it turned out that it was a voltage reference comprising a 6v ish Zener diode in series with a normal diode.
it measured open-circuit because there were two back-to-back diodes in series, one being a Zener.
The idea behind this arrangement is that the tempco of a Zener of this voltage can be exactly compensated by the tempco of a normal diode, the two being equal and opposite.
I think the notion was that the ~6.8V zeners had the lowest dynamic impedance.
The data sheet table seems to support that.
I've never heard that they would have the lowest noise.
Jan
The data sheet table seems to support that.
I've never heard that they would have the lowest noise.
Jan
Ok thanks all it does indeed look like the Philips / NXP data is quite different.
It definitely pays to look at the datasheet, the difference between the two manufacturers is quite dramatic
I think this might be the answer, thank you...
It definitely pays to look at the datasheet, the difference between the two manufacturers is quite dramatic
TL431 is even better I'd wager. Any voltage you like at next to zero TC at the cost of two resistors.
Edit: Oh, same thing.
Edit: Oh, same thing.
TL431 is excellent in terms of stability but generally poorer in terms of noise. However, at least its noise level is predictable, zeners are rarely specified for noise and vary rather a lot between output voltages and different vendors.
Definitely true at 1mA on NXP BZX84. Though the 6.2V has the lowest maximum dynamic impedance at 5mA.