Super Regulator

Pooge, he stated that +/-22 is the limit for opamps. Since one opamp is used per rail, this tells you that for each rail you can go nearly as high as the maximum opamp supply voltage. You don't want to go higher, not to damage the opamp.

So if each opamp can handle a 44V rail to rail voltage, this means that each rail should comfortably give you around 40V. Two rails of 40V equals +/-40V. :)
 
AX tech editor
Joined 2002
Paid Member
Jan,

PR2201 can do 80V. JFET input on top.

PR2201 / PR2202 80V OP-Amp, High Voltage Op Amp

But I think you might not like the noise figures. :)


Patrick

Yes, also the BW and slew rate isn't that hot, and Iout is pretty low.

If you need more than say 40V from these regs, a good way is to hang a 3-pin reg off the output and supply the opamp from that.
Then again, also 3-pin regs have limited voltage range in absolute terms.
Or maybe just a zener in series with the opamp supply pin.
Haven't tried that though.

Jan
 
AX tech editor
Joined 2002
Paid Member
Stupid Question, : What are the maximal output current of the regulator ?

Not a stupid question, rather to the point.
I would say about 1 A. If you really want to go to such high currents, you probably want to select a high-Hfe pass transistor.
Also keep an eye on the dissipation - with 5V across the series transistor at 1A that's 5W. Check the degrees/watt figure of the heasink you use to avoid too much temp rise.
For instance, with a heatsink with 5 degree/W that 5 watt causes a 25 degrees temp rise. That's probably as high as you want to go, so a taller (lower degrees/W) heatsink would be better.

jan
 
AX tech editor
Joined 2002
Paid Member
Pooge, he stated that +/-22 is the limit for opamps. Since one opamp is used per rail, this tells you that for each rail you can go nearly as high as the maximum opamp supply voltage. You don't want to go higher, not to damage the opamp.

So if each opamp can handle a 44V rail to rail voltage, this means that each rail should comfortably give you around 40V. Two rails of 40V equals +/-40V. :)

Yes, thanks.
 
> If you need more than say 40V from these regs, a good way is to hang a 3-pin reg off the output and supply the opamp from that.

Powering the opamp is one thing, and you can probably use a TL431-based shunt regulator plus a CRD for that.
The opamp output DC, on the other hand, has to be output voltage + 0.65V - startup-Zener voltage.
Are you prepared to use a high-voltage Zener there then ?

> What are the maximal output current of the regulator ?

Equal to the max. current of the opamp x hfe of the pass device ?
Assuming of course you have sufficient heat sinking and a powerful enough pass device.
You can of course use a (depletion) FET. They don't load the opamp so much.


Patrick
 
Last edited:
AX tech editor
Joined 2002
Paid Member
> If you need more than say 40V from these regs, a good way is to hang a 3-pin reg off the output and supply the opamp from that.

Powering the opamp is one thing, and you can probably use a TL431-based shunt regulator plus a CRD for that.
The opamp output DC, on the other hand, has to be output voltage + 0.65V - startup-Zener voltage.
Are you prepared to use a high-voltage Zener there then ? Patrick

Ohh yes; the voltage across the zener doesn't matter so its noise is largely irrelevant, and the current is controlled by the opamp anyway.

>

> What are the maximal output current of the regulator ?

Equal to the max. current of the opamp x hfe of the pass device ?
Assuming of course you have sufficient heat sinking and a powerful enough pass device.
You can of course use a (depletion) FET. They don't load the opamp so much.


Patrick

Actually, you must first make sure that the current source has enough current to fully open the transistor at say 1A. Then you must make sure that the opamp can absorb that current for the no-load case.
 
i'll have to look at the test data but the NE5534 wasn't in the same class with the AD825 or AD797. I would suggest that you try the OPA604 which will do +/-24V, has low noise and a reasonably high slew rate. As we know from experience, the AD797 tests wonderfully, (most of the time), but is prone to instability.
 
Not a stupid question, rather to the point.
I would say about 1 A. If you really want to go to such high currents, you probably want to select a high-Hfe pass transistor.
Also keep an eye on the dissipation - with 5V across the series transistor at 1A that's 5W. Check the degrees/watt figure of the heasink you use to avoid too much temp rise.
For instance, with a heatsink with 5 degree/W that 5 watt causes a 25 degrees temp rise. That's probably as high as you want to go, so a taller (lower degrees/W) heatsink would be better.

jan

Thank you for fast and meaningful answer.
 
> Can you expand on that?

The pass device is a BJT, which means the base voltage is higher than the emitter (= output) voltage by about 0.65V (Vbe). Else no current will pass.

The opamp is powered by the output. So even if the opamp is Rail to Rail, it cannot possibly supply the required voltage to the base of the pass device directly. So you need a start-up Zener to make up some voltage. In Jan's schematics, that zener is 6.8V.

If the output is higher than 44V, you would either have to use a HV opamp, or you need to reduce that output voltage to something the opamp can take, say 30V. So let's say output is +50V, and we supply the opamp only with 30V. That supply can be made up by a TL431 configured to 30V, plus a CRD (say 5~10mA).

Now things get interesting. You can hang that 30V opamp supply and hence the opamp itself at the "top", i.e. at the output, or you can put it at Gnd level. I personally do not like hanging the opamp to a "floating" voltage, even though it should be regulated. So I want my opamp to see 0V / +30V for a positive regulator. Now even if the opamp is R2R output, its output can only be 30V max. But the pass device base needs to be at +50.65V. So you would need a ZD of say 24V, ideally ~30V, before the regulator will work at all.

Not for beginners.....

Clear enough ?


Patrick
 
400V 50mA J-D Pos Super Reg

To take it to the extreme, and for the sake of stimulating discussions,
this is a concept (i.e. not-proven, build-at-own-risk) schematics for a 400V 50mA Jung Didden Positive Super Regulator.

At least in Spice simulations, performance is excellent.


Patrick

.
 

Attachments

  • 131213 400V Jung Didden Pos Concept Schematics Public.pdf
    106.3 KB · Views: 1,147
AX tech editor
Joined 2002
Paid Member
Nice Patrick, that should work well!

Two things to watch out for: start up may give issues with the opamp locking up at zero output, and on the other hand the supply to the opamp may have switch on transients above the save value.
These things are not always modeled well in Spice.

The lock up may be fixed with a large resistor across the pass device, while the opamp can be protected with a zener across it's supply pins.

Since the MOSFET pass device will have lower transconductance than a BJT performance may be slightly less, but will be of no consequence.

If you use depletion mode MOSFETs like the 2540, be aware that it has a peak Id limit not much higher than its avarage Id limit. That means that if you have large capacitors at the output, the switch-on (high current loading of the output cap) may destroy the pass device.
That's the reason that in my HV supply I went away from a depletion mode device; it did increase the circuit complexity somewhat but made it much more robust.


Jan
 
Member
Joined 2011
Paid Member
Like the TL071, the OPA140 uses JFETs on its inputs. (Pchannel for the TL071, Nchannel for the OPA140).

Like the TL071, the OPA140 can tolerate a very large delta-V across its inputs; there are no emitter-base junctions to worry about, and no VEBO breakdown voltage to fear.

This is why the TL071 is the unanimous choice for generating triangle waves in PWM switchers: the (integrator + SchmittTrigger) triangle circuit applies more than 2/3rds of the supply across the opamp's input pins. That would be disastrous for a BJT input opamp, but no problem at all for a JFET input device like the TL071. Or the OPA140.

So why include diodes D6 and D7 in the schematic of post #34? They are across the input pins of a JFET opamp.
 

Attachments

  • diodes_opa140.png
    diodes_opa140.png
    53.5 KB · Views: 3,317