[sdiy] [Potential SPAM] Re: Mind blown -- TL074 saturated outputs different

[Neil Johnson]

Hi,

Matthias wrote:
> Sorry to be late on this ball, but I remember seeing (and saving) a schematic example of this, a single zener surrounded by 2 or 4 diodes in some bridge configuration - but now I cannot find it. The main point of it as I remember was to get a very symmetric output by avoiding having more than one zeners (since they are rarely as well matched as regular diodes), and instead use the same zener as the level reference for both the positive and the negative side. I guess a TL431 could give an even better result.
>
> Can anybody give a pointer to where this has been described?

One example from the literature is Figure 10.6 in Sergio Franco's
"Design With Operational Amplifiers And Analog Integrated Circuits"...
see page 234 of the PDF in the web archive:
https://archive.org/details/SFrancoDesignWithOperationalAmplifiersAndAnalogIntegratedCircuits1Pdf

David G Dixon wrote:
> The idea is that the zener bridge gives more symmetrical results than two back-to-back zeners.  However, it relies upon the assumption that the two pairs of regular diodes are identical.  This is probably more true than assuming that any two zeners of the same nominal voltage are identical, but it is still an assumption.

Well, almost, except without the ASSumption.  We don't ASSume the
bridge diodes are identical, it works on the basis that any two
randomly-selected silicon diodes' forward voltage drops are
significantly closer to each other than the reverse avalanche/zener
voltages of two randomly-selected Zener diodes.  For example, if we
consider a Vishay BZX85C5V1 zener diode its VZ is 5V1 +/- 300mV (the
spread gets wider for higher voltage diodes).  Compare this to, say,
BAT85 schottky diode, or 1N4148 silicon switching diode, and the VF  -
while still having some spread, is much tighter (in the region of a
few mV).  So no, we don't ASSume the diodes are identical, but for the
cost of four extremely cheap silicon diodes we get a much better match
of the positive and negative clamp limits without the cost of either
hand-selecting zener diodes or paying for tighter VZ binning, and as
others have suggested you can also use more esoteric devices like
TL431 and so on.

Neil
-- 
http://www.njohnson.co.uk