Adapting this NSC app note for my +150V regulator

At the high output voltage you want this is usually done via Opamp and discretes, using FETs. See Art of Electronics Chapter 6.2. In my copy it's on page 369, figure 6.47.

Thanks, I'll take a look at that when I get home. (I don't have a copy of the book here at my job).

Another thing that comes to mind is an LM723 floating regulator with an external pass transistor, something like the +100V regulator shown in Fig. 22 of this datasheet:

I want to stick to common, inexpensive parts if possible.

By the way, I am aware that there was a big thread here about +150V regulator options a while ago--but alas, i can't get much out of it because the copious ASCII schematics posted there all come out garbled on my end.

--Dave

Not really, since the base drive would be a bit complicated for a +150V output.

The uA723 would be an excellent contender, just not like in Fig.22. Take a look at AoE when you get home, that's how it's done.

Might have to switch to fixed-font, then they work ok.

BTW, bottom posting would be nice, especially when a thread becomes longer.

The TIP50 is not good enough. Taking the approach of short-circuit limiting at 2x the maximum regulated current rating, as in LB-47, exceeds the SOA of that particular transistor. Even strict limiting at

Good point. I know there are some transistors intended for horizontal output service in TV sets that are pretty beefy, with high C-E breakdown voltage ratings. I'll look into those.

Also, take a look at big FETs. They have become rather inexpensive.

Power MOSFETs avoid the 2nd-breakdown problem, which can be a big deal in HV regulators. However, high-voltage power MOSFETs love to RF oscillate at from 10 to 30MHz when used in the linear mode at even modest currents, like 5mA, and with more than say 25V across their drain-source. The easiest way to prevent this is to use two ferrite beads and slip one over both the gate and source leads of a TO-220 part. Sometimes a gate resistor alone can spoil the oscillation (they're always recommended anyway, and especially if you use a gate-source zener, isolate the gate with a resistor), but I have better results and am more comfortable with two ferrite beads.

I have had good luck with just putting a SMT resistor right at the gate lead. 51 Ohms usually works. Don't put a long trace between the resistor and the gate.

The most direct adaptation of LB-47 using a HV MOSFET would be like so, that particular FET is available from Digikey for not much more than a TIP50: View in a fixed-width font such as Courier.

. . . . . V+ . | . .-------+-------+ . | | | . | | | . [Rb] [Rg] | . | | | . | | |< IRFU9310PBF . | +-----| P-CH . | | |-. . | | | \\ . +-------|-------|--------|

And if you use beads spend a drop of compliant (soft) glue. Else they rattle in transport, maybe shatter, lodge their pieces in not so good places and ... click ... phssst ... WHADDABANG!

This sims well and requires only low voltage components for compensation, be sure to have good bypass at the input and output: View in a fixed-width font such as Courier.

. . . . . V+ . | . .------+-------+------+----+ . |2W | | | | . [100k] [330] [3.3k] | | . | | | -/ | . | | |1N4742^ | . | | 0.47u |12V | |< IRFU9310PBF . | '--||---+-[47]-+--| P-CH . | | |-. . | | | \\ . +--------------|-----------|--------|

Strictly speaking, this would be better: View in a fixed-width font such as Courier.

. . . . . V+ . | . .------+-------+--+--------+ . |2W | | | | . [100k] [330] [3.3k]| 1N4742| . | | | -/ 12V | . | | | ^ | . | | 0.47u | | |< IRFU9310PBF . | '--||---+--+-[47]-| P-CH . | | |-. . | | | \\ . +--------------|-----------|--------|

Excuse me, I use sensible small beads that have no chance to short the leads -- recommended. In fact, among the beads in my inventory, it's hard to find many large enough to cause trouble.

A p-channel part here, awkward, not quite my cup of tea, I'm afraind.

Probably a good idea, although the beads I use are so small and light, it's hard to imagine any trouble.

Kind of reminds of the bad batch of MMBT2222's we have sitting in our bins that are good for nothing but saturated switching circuits. If you use this batch in a linear circuit, there is a point on the curve that it actually oscillates. Go above or below this point and we're back in business again. This was found to happen on 2 different test circuits to insure it was not bad board and component design. We found this out after debugging a circuit where these were being used to control small servo's. Putting some padding in the circuit removed the higher frequency component and thus still effected the end results by causing low freq distortion in the curve at that same point.

A new batch was sent to us and all was well once again.

Looks like a faithful reproduction of Maida's original intent which is a common base voltage isolation level shift relative to a reference bootstrapped off the output. The LM317 is not too bad for this app, the only disadvantages are its fixed internal current limit and relatively large operating current overhead.

It doesn't have much to do with weight. It's the permanent rattling during transport that does them in. I've seen bottom panels of machines with a whole enchilada of parts, shards, ferrite dust and so on. At the end of the day ferrite is ferrite, and that's always very brittle.

Hmm, I use the MMBT2222 a lot in linear apps. Never had a problem.