circuit that sumulates a high voltage zener with low temperature coefficient

Try:

It's raw..Guessing something like this...

| R | +--------+--------+------>1mA-->

| | | | R1 | | op amp | +---------|+\\ D | | |---G =>600V Mosfet R2 Vref--|-/ S | | | +--------+--------+ |

Might be a good starting point.

Needs more design work.

D from BC British Columbia Canada.

You didn't say how stable. "Back in the old days" people would run an NPN planar silicon transistor with negative on the collector (forward biasing the BC junction) till the BE junction zenered. The temperature coefficient of the resulting zener+PN could be quite low. So you _may_ be able to do well enough using something like a high- voltage transistor with a 100:1 (nom) resistive divider, collector:base:ground, with about a 5.6V zener emitter to ground. You need to use a high beta transistor so your resistive divider can be moderately "stiff" relative to the required base current. And you can overcome that by using a darlington connection with two 5.6V zeners in series, emitter to ground. Since high voltage transistors tend to not have very high beta, you could use a lower-voltage transistor whose collector is connected to a lower supply voltage. To keep transistor dissipation lower (1mA at 600V is still 600mW) you could put a resistor in the collector of the HV transistor; voltage divider still must come from the regulation point, of course.

A problem with the above is that current through the zener may vary too much to maintain the stability you want. The thermal may be OK, but not the load regulation.

Personally, I'd look at adding a transistor and a divider to a shunt regulator IC. I remember doing this in a design a long time ago, and it worked well (though you do have to be careful about compensation; I recall discovering that they didn't tell you everything you needed to know in the data sheet for the regulator, since they hadn't anticipated anyone would want to use it the way I was trying to... a different shunt regulator part number solved the problem nicely though.) You may even be able to find an ap note about doing this, or an example circuit in a data sheet. I'm thinking you might try TI, National, Linear Technology, Analog Devices... A TL431 looks like it would do just fine if you simply add an NPN high voltage "cascode" transistor between it's "cathode" lead (transistor emitter) and the regulation point (collector) with the base running at 5V or so. Be sure to use a high voltage resistor (or at least a string of several resistors in series) for the feedback from the regulation point. It would be courting disaster to use an 0805 SMT part with 600V across it...

Cheers, Tom

Indeed, with all that excess loop gain, first the opamp, then the mosfet, such a configuration will need lots of additional design work. Mook needs a circuit using 5.6 to 6.2-volt zener diodes, since that's where they have good properties. The classic stable HV "zener" circuit is made from a LV zener and one or two HV transistors.

Vz | +-----, | | | R1a c | b---+ e | | R1b | | c | Vz = (1 + R1/R2) (Z1 + Vbe) + Ib*R1 b---+ e | where R1 = R1a + R1b | | \\_|_ R2 /_\\ | | Z1 | ---+-----+----

The transistors may be Darlington form to reduce Ib in the Vz equation, Vz = (1 + R1/R2) (Z1 + Vbe) + Ib*R1.

le

uld

If it's for a reference, consider a current sink reference and pullup resistor to sense the HV, and an op amp and MOSFET connected in cascode to sink the current.

(warning: mediocre ASCII art follows:)

HV source + | | +-----------------------+ | | | | +--+--+ | | | +-| =3D > ||--[+5V] | >6Mohm +-| | > | | | |\\ >

| | >---+ +--------------|- / | |/ | /| +--| / +|---[GND] ||----< | +--| \\ -|--+ | \\| | +----------------+ | < < Rsense < | [-Vref]

The bypass capacitor across the HV sense resistor is to keep the op amp input capacitance from rolling off the frequency response. 'Rsense' will be a convenient place to install a trimmer if needs be.