zener voltage reference, what is stable temp voltage?

[snafflekid]

zener voltage reference

I am making a zener based UVLO. I thought that the temperature stable zener diode voltage was around 5.6V but I am finding literature stating it is 6.3V The technology I am designing with has 6.3V zeners, so I am likely stuck with that. An explanation would be nice anyway. Here are links to what I am talking about.

**broken link removed**

https://focus.ti.com/lit/an/slyt183/slyt183.pdf

 

[erikl]

Re: zener voltage reference

I thought that the temperature stable zener diode voltage was around 5.6V but I am finding literature stating it is 6.3V

5.6V is the better fitting figure. But the exact value - i.e. the relative contributions of zener resp. avalanche mechanisms - depends on the steepness (doping concentration and depth) of the p+n(+)-junction.

The technology I am designing with has 6.3V zeners, ...
An explanation would be nice anyway.

So your zener will most probably own a small negative temp.coeff. (s. citation from Allen/Holberg book below). Your foundry should tell you its value. If it's -2mV/K , you could perhaps compensate it by the temp.coeff. of a forward biased pn-junction (≈ +2mV/K).

 

[dick_freebird]

To get a true flat tempco you probably have to comp
it with some fraction of a forward diode (all, or none,
are both unlikely to be ideal). This may mean two trim
networks (if you're looking for that kind of accuracy).

 

[snafflekid]

I have determined I need to build a UVLO using a Zener reference. I have breezed through CURRENT SOURCES & VOLTAGE REFERENCES by Linden T. Harrison and noticed that the flattest tempco of a Zener is around 5.5V. This jibes with what I have learned prior. Zener voltage references also add a pn junction and the Zener is set to have +2.1mV/C TC. For reasons I have not researched, this combination operated at a particular current has a near zero TC and results in a total voltage of 6.3V. However it seems to me that the Zener itself could be set to have a zero TC alone. My guess why this is not the common method is that this doping level is more sensitive to process variation than the level that has +2.1mV/C. I just don't know

Now, I am going to have to indirectly ask questions of the technology group to determine why the Zener we have is set to 6.3V. I have a suspicion that someone told someone that the zero TC voltage is 6.3V without understanding that this includes a Vbe! ARGH! Monday is going to be fun.

 

[erikl]

... My guess ... is that this doping level is more sensitive to process variation than the level that has +2.1mV/C.

Below pls. find a 9-year-old report on zener breakdown voltage dependency on ion implant dose. The process was 0.35µm std. CMOS. Foundry-specific info has been removed.

... determine why the Zener we have is set to 6.3V.

"Zener" breakdown at this voltage is much steeper than at lower breakdown voltages (s. the report below), because - with increasing breakdown voltage - the avalanche mechanism dominates more and more the zener breakdown mechanism. Thus the breakdown voltage at a certain current shows a much lower tolerance.

I have a suspicion that someone told someone that the zero TC voltage is 6.3V without understanding that this includes a Vbe!

I don't think you can get a zero TC voltage of 6.3V, even with a series circuit of a 5.6V "Zener" diode and the forward Vbe of a normal silicon diode (≈ +2mV/K), because the TC of a 5.6V "Zener" diode - which is far more an "avalanche diode" - tends to be positive, too, about +1mV/K, s. this curve. In order to compensate the ≈ +2mV/K TC of a forward Vbe voltage, a real Zener diode with a negative TC of about -2mV/K is required, and this necessitates a zener with a breakdown voltage of about 4.0 .. 4.5V , s. the a.m. curve (commercially available zener diodes of 4.3V show this TC). So a (near) zero TC voltage of such a series circuit would be about 5V.