Need analog chips to regulate +/-70vDC to +/-60vDC

[bdunham7]

Yup, or including a relatively high capacitance load -- can happen when a lot of bypassing/filtering is used -- regulator goes into current limit and the input charges up to ~nominal voltage while the output lags behind.  In which case the zener has to be rated to handle at least that surge energy due to charging the load, while perhaps not also the energy of discharging the main (input) cap.  (Or sinking mains fault current once the cap's been discharged; not to leave that out, either.)  And for short durations (some µs), yeah, the regulator won't respond instantly either, but that should be fairly small potatoes for a linear supply sort of thing, and it won't take much zener to assist that; response time doesn't mean anything though if we're talking about current limiting under startup/short conditions.

True for larger power supplies, but I hadn't thought of the non-thermal current limiting because I was only referring to the circuit posted above from the TI datasheet that used a 1W zener.  You'll notice that the fault current there is limited to below 0.9A under any conditions.  Beyond about double that I think it would be better to use a different design, probably one that doesn't depend on a fuse to keep it from going up in flames.

[bdunham7]

Why go to all the trouble using these lower voltage regulators and having to protect them

Because it is cheap and simple and works really well for low-power applications. 

[mawyatt]

Why go to all the trouble using these lower voltage regulators and having to protect them

Because it is cheap and simple and works really well for low-power applications.

And how is this simpler, cheaper than a simple shunt zener with series pass transistor? Granted maybe a little lower overall power consumption.

Best,

[T3sl4co1l]

True for larger power supplies, but I hadn't thought of the non-thermal current limiting because I was only referring to the circuit posted above from the TI datasheet that used a 1W zener.  You'll notice that the fault current there is limited to below 0.9A under any conditions.  Beyond about double that I think it would be better to use a different design, probably one that doesn't depend on a fuse to keep it from going up in flames.

Oh right, there's a series resistor too!  Which makes the fuse kind of dubious, but it says 1/8A and 100V 100Ω is 1A short circuit, it'll open eventually.

Anyway, the cascode one is the way to go, though it's a bit tight for OP's low overhead application.  A CCS pullup instead of the resistor might be worthwhile.

Tim

[TimFox]

Why go to all the trouble using these lower voltage regulators and having to protect them

Because it is cheap and simple and works really well for low-power applications.

And how is this simpler, cheaper than a simple shunt zener with series pass transistor? Granted maybe a little lower overall power consumption.

Best,

Note that higher voltage Zeners, say 62 V, have a relatively-large temperature co-efficient and will drift due to self-heating.
For regulating screen voltage on pentodes, I try to use a series string of lower-voltage diodes.

[mawyatt]

Why go to all the trouble using these lower voltage regulators and having to protect them

Because it is cheap and simple and works really well for low-power applications.

And how is this simpler, cheaper than a simple shunt zener with series pass transistor? Granted maybe a little lower overall power consumption.

Best,

Note that higher voltage Zeners, say 62 V, have a relatively-large temperature co-efficient and will drift due to self-heating.
For regulating screen voltage on pentodes, I try to use a series string of lower-voltage diodes.

Yes aware of the TC issue, we've even used a simple shunt type ~6V zener with additional shunt transistor (~ zero combo TC) that has worked well for HV applications. Certainly not in the highly stable regulator class but quite sufficient for our needs.

Best,

[TimFox]

The sweet spot for normal Zeners, where the tempco switches polarity, is somewhere around 5 to 6 V, IIRC.
A typical temperature-compensated "reference diode", where the tempcos of the Zener and forward junctions cancel at a given current, is a series combination of a 5.6 V Zener and a "normal" diode to give nominal 6.2 V, e.g. the 1N821 to 1N825 series.

[mawyatt]

Around ~6V a typical "zener" diode has ~+2mv/C TC. In the concept shown above this +2mv/C TC compensates the transistor -2mv/C TC, yielding a net overall control TC of ~0, and since the series pass element (transistor, darlington or NMOS) is "inside" the feedback loop it's TC gets divided by the Open Loop Gain. If the feedback resistors R1 and R2 have a similar TC, then their TC won't matter much either. If more precision is required then the transistor Vbe should be in included with the zener voltage in the equation (Vz) shown, so would be:

Vo ~ (Vz +Vbe)(1 +R1/R2)

Best,

[Zero999]

Why not use the TL431, rather than a plain zener?

To work at higher voltages than 36V, it can be cascoded or a zener connected in series with it. See the thread I made about it a few years ago.

https://www.eevblog.com/forum/projects/tl431-common-base-amplifier-stability/msg599580/#msg599580

Add another pass transistor, with the TL431+cascode in the feedback loop and you have a high voltage linear regulator.

[bdunham7]

And how is this simpler, cheaper than a simple shunt zener with series pass transistor? Granted maybe a little lower overall power consumption.

I'm not claiming it is necessarily simpler, just on the same order.  But they are probably a bit simpler to design with parts that are laying around--so easy a cave man could do it.  And the performance of the simplest implementation is probably better as well.

[magic]

Add another pass transistor, with the TL431+cascode in the feedback loop and you have a high voltage positive linear regulator.

FTFY

[mawyatt]

Why not use the TL431, rather than a plain zener?

Yep, see post #24.

"Of course one can employ a TL431 and a couple more transistors & resistors for a better performing regulator, but may not be necessary."

Best

[TimFox]

In a similar circuit, I have used a TL431 with a highish-voltage Zener in series with the + terminal, but a Shottky diode between the + terminal of the TL431 and the control pin, and a resistor across the TL431, are needed to avoid smoke (preventing the control from becoming more positive than the + terminal, and ensuring current through the Zener to limit the voltage across the TL431).

[nctnico]

Why go to all the trouble using these lower voltage regulators and having to protect them

Unless the OP has some tight regulation requirements a simple series pass transistor with shunt zener is hard to beat for simplicity. A couple zeners, a NPN and PNP pass devices, and a couple resistors (shunt caps for better noise/ripple) and you have a +-60V regulator

Of course one can employ a TL431 and a couple more transistors & resistors for a better performing regulator, but may not be necessary.

Simplicity yes, but I have a thing for current limiting and thermal protection. It is nice when circuits can survive a short at the output.

[bdunham7]

Simplicity yes, but I have a thing for current limiting and thermal protection. It is nice when circuits can survive a short at the output.

Yes, there's no reason not to include all that in such a basic design, someone can thrift it out later if they just don't want it.

Here's a very straightforward 'cave-man' solution for the OP to look at.  There might need to be a second 11K3 resistor across the negative output if it is operated with no/low load currents and you can add capacitors for filtering depending on the inputs, but I think this is a minimal solution that will provide 'pretty good' results and is pretty well protected.  Shouldn't even need heatsinks and the BOM is $9 in single quantities at Mouser.  In stock too!

Edit:  So much for cave men.  I would might have to lower the hold current of the PTC or raise the voltage of the zener--or both--to fully protect it.  But those parts are in stock too.

[enut11]

Thanks @bdunham7.
The 70v raw inputs for the regulators will be filtered with 680uF caps.
The load is an audio amplifier kit and the rail caps are 100uF.
enut11

[bdunham7]

Thanks @bdunham7.
The 70v raw inputs for the regulators will be filtered with 680uF caps.
The load is an audio amplifier kit and the rail caps are 100uF.

Are those raw as in direct from a transformer and rectifier?  If so, do those inputs feed anything else or just this circuit?  If 1) yes and 2) no, then are those capacitor specs set in stone?

[enut11]

@bdunham7
The raw 70v is direct from rectifier/filter and the capacitance can be varied.

The audio amp input caps are more difficult to change due to space on the PCB

The regulators will only feed the audio power amp.
enut11

[magic]

The zeners could be 5W or TVS as they need to trip the fuse/PTC. Note that when the load draws constant 0.18A, the regulators may overheat and go open circuit and it will fall on the zeners to conduct that current.

It may be a good idea to have diodes clamping the rails to ground so that neither can reverse polarity if something shorts them together.

The load is an audio amplifier kit and the rail caps are 100uF.

Any known PSRR requirements?

[Zero999]

@bdunham7
The raw 70v is direct from rectifier/filter and the capacitance can be varied.

The audio amp input caps are more difficult to change due to space on the PCB

The regulators will only feed the audio power amp.
enut11

Just one question: why are you using a regulated power supply, for an audio amplifier? It's not normally necessary, since the power supply rejection of a well designed amplifier makes voltage regulation unnecessary.

If the design you're using is for 60V, but you have 70V, on the filter capacitors, then it's better to modify the circuit for the higher voltage, than to add an unnecessary voltage regulator. It's probably as simple as increasing the voltage ratings of a few components and using larger heat-sinks.

EDIT:
I've just read the original post again and I'm confused.

I want to regulate a raw +/-70vDC supply to +/-60vDC at less than 100mA. It is a low noise metrology project so aiming to use analog chips.

I found the TL783 125v will do the positive side but have yet to find a suitable negative chip.

Any suggestions from the analog gurus?
enut11

Now you say it's an audio amplifier.

Please post a schematic or data sheet for what you're powering.

[bdunham7]

The zeners could be 5W or TVS as they need to trip the fuse/PTC. Note that when the load draws constant 0.18A, the regulators may overheat and go open circuit and it will fall on the zeners to conduct that current.

I intended to have the zener never to have to conduct significant current under any external fault.  I think that works as drawn as long as the 70V input is exact or very close.  Note that with the 68R resistor, if the current exceeds about 120mA the voltage drop reduces the input voltage and the regulator drops out and thus will dissipate minimal power.  Also my 'no heatsink' statement reflected a room temperature ambient.  If any of that changes, then as I said in the edit,  mutatis mutandis -- alterations might be needed.  The key to this 'floating high' arrangement is that the current and thermal overload functions have to be taken away from the LM3x7. 

[bdunham7]

The regulators will only feed the audio power amp.

What I meant was does the 70V unregulated power feed anything else besides the regulators?  And what are the characteristics of the transformer, etc?  Is that 70VDC no-load, estimated, exact within +/- x.0 volts, ??

[magic]

I intended to have the zener never to have to conduct significant current under any external fault.  I think that works as drawn as long as the 70V input is exact or very close.  Note that with the 68R resistor, if the current exceeds about 120mA the voltage drop reduces the input voltage and the regulator drops out and thus will dissipate minimal power.

Guess you're right about the dropout
And I suppose 68Ω is also supposed to limit short circuit current to 1A so that the regulator has no trouble passing it.
I'm starting to wonder if the zener is even of any use at all. It looks like internal current limiting is the only condition which endangers the regulator.

Rail clamping diodes may still be a good idea.

[bdunham7]

I'm starting to wonder if the zener is even of any use at all. It looks like internal current limiting is the only condition which endangers the regulator.

Rail clamping diodes may still be a good idea.

The zeners only conduct when the regulator reaction time is not fast enough to react to a sudden load increase.  Or if the regulator shuts down, but that would be a catastrophic event.  Protecting against that would require a pretty big device, although it certainly could be done that way in an alternative design.

There are many things you can add to improve the characteristics of a circuit like this, but I'm not sure whether they are needed or not without more exact information so we know what to worry about.  Diodes are cheap, I suppose.  My main concern would be getting the overcurrent protection right, which might require a higher power rating on the input resistor or a different PTC or both.  That and the zener should be a good margin above the max input voltage minus 60 volts.

[enut11]

The regulators will only feed the audio power amp.

What I meant was does the 70V unregulated power feed anything else besides the regulators?  And what are the characteristics of the transformer, etc?  Is that 70VDC no-load, estimated, exact within +/- x.0 volts, ??

The transformer is a 50VA toriodal and feeds only the amp via the regulators. 50vAC to 70vDC unloaded. Variability depends on the 240vAC mains fluctuations.
Transformer load is less than 50mA as it supplies voltage (not power) to a DMM.
enut11