Low voltage shunt regulator

On a sunny day (Tue, 19 Aug 2014 07:13:05 -0700) it happened John Larkin wrote in :

See...

HighVoltageOperationOfTL431.pdf

on the S.E.D/Schematics page of my website.

Replace TL431 with LM285-ADJ and "Toss in" a conventional MOSFET (I see no advantage using a depletion-mode FET) and perhaps a few protection diodes for safety during start-up.

Looks like a reasonable expectation would be about 1 Ohm per stage, so the higher the MOSFET voltage the better. ...Jim Thompson

Not a whole lot in that toolbox there, Mr. Goodwrench.

Oooops!

Some issues... at least according to the LM185-ADJ Spice model I have, the knee ain't so marvy... needs about 20uA cathode current to be "stiff".

Also, the reference location is between CATHODE and ADJ, rather than between ADJ and ANODE as in the TL431, so the "sex" in reversed.

...Jim Thompson

• Final voltage to be 400.0 +/- 0.1; total noise/ripple 100mV; regulation 100mV or better 20uA-500uA; response time < 1mSec; anything "faster" in self-noise,etc should be by-passable with 0.1uF or less.

• MUST be isolated, eg: act like a theoretical zener.

Well, technically it is not oscillation, the ranDUMB noise is caused by sporadic micro avalanche breakdown. One can always find some (small) current range where that does not happen,BUT all zeners i have looked at (4.7V to 180V, numerous brands) exhibit the avalanche noise somewhere (or in most of) in the 20-50uA region. That configuration is not useful; it is not a 2-terminal shunt regulator.

That would be utterly silly for a 400V app.

Looking at schematic, it seems no FET is needed. But i do not calculate the 45V. Mister Ohm, NOT related to Sherlock, tells me a total voltage of

As i said, zeners are not useful: too damn noisy; the good-old avalanche spike crap which gives a negative resistance and possible oscillations.

Gotcha! Vref for a TL431 is 2.5V ;-)

Vref for the LM185-ADJ is ~1.245V ...Jim Thompson

I think we have a little thread drift here. Somebody was talking about HV stuff for old color TVs.

It is wore than that...burn 10uA for a divider and that leaves a huge

Looking again in my shunt regulator archive, once upon a time (2002), STM made the TS431 which had the exact control you mentioned. Minimum cathode current 40uA (typ). Your Spice model for the TL431 seems congruent with the STM part. I assume that your LM185-ADJ Spice model is appropriately "flipped". I found some Nat Semi LM185 models, so now i have a nice disparate set where no doubt none of them follow reality in that low current region. But at least one of them will show what to expect.

Thanks.

I've seen a bad batch of MMBT2222 have what looks like a tunnel diode effect in a particular bias region and thus appeared on the scope as a spur. Like i said, bad batch. I still have a hand full with a note on them indicating to be used as switches only.

As for zeners oscillating, I've never seen one that cause an oscillation that was bad enough to actually cause an issue. Only when they come under very low current maybe, but I think that falls true with many pn junctions?

Are you sure you didn't experience this using some glass envelope types that could be acting photovoltalic? If sure would make since when being used in very low current circuits..

Jamie

Looks like you're up a creek. Got a paddle?

Cheers, James Arthur

What's wrong with...

In fact, with max 17uA total supply current between the two, you've got an extra 3uA to spare on the supply; that could be burned in a 5V zener. Obviously, you'll need a ~20M series dropping resistor for that, and the feedback divider must be over 133M.

With minimum gain > 10^5, the opamp output will rail over a range of less than 10uV at the input. Or for a 1/160 feedback divider, no more than

The initial accuracy of the reference is not quite 0.025%, but there are tighter references out there, or you can trim it, as its stability is good.

Don't forget to use extreme precision film/foil resistors; have fun with that >= 133M part.

As you did not provide a cost or size constraint, I'm going to stamp this design exercise as "complete". Ship it! ;-)

Tim

Check; pawn to Q3.

• OK; bad terminology. IV curve has negative resistance and lots of "fuzz" in that region; the "fuzz" being ranDUMB avalanche spikes which looks a bit like oscillation. All zener diodes without any regard to package do this s*it somewhere in the 2uA to 10mA region. ONLY when one pushes the recommended operating current (umpeen mils) do they always behave.

NOPE! Look carefully at the TI datasheet: TYP min operating current Vout=Vref is 7uA for the LM385, and 8uA for the LM185/285.