Current boosted three terminal regulators and noise

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

ruffrecords

Well-known member
Joined
Nov 10, 2006
Messages
16,152
Location
Norfolk - UK
There are lots of circuits showing how to boost the current of a three terminal regulator using and external pass transistor. However, as far as I can see, they all work by sensing the current going into the regulator and, above a preset current, they start diverting it through the external pass transistor. But it seems to me that all of the ripple reducing properties of the regulator are not shared by the current flowing in the pass resistor because it is not included in the feedback loop. So does this mean that all such current boost schemes are useless for audio work?

Cheers

Ian
 
ruffrecords said:
There are lots of circuits showing how to boost the current of a three terminal regulator using and external pass transistor. However, as far as I can see, they all work by sensing the current going into the regulator and, above a preset current, they start diverting it through the external pass transistor. But it seems to me that all of the ripple reducing properties of the regulator are not shared by the current flowing in the pass resistor because it is not included in the feedback loop. So does this mean that all such current boost schemes are useless for audio work?

Cheers

Ian
The ones I've seen are in the feedback loop.... (I think).

JR
 
JohnRoberts said:
Yes...  It is still regulating the output voltage...

JR

In which case, as far as the pass transistor is concerned this feedback must be achieved by a change in regulator input current? So, going back to my original question, how is the output ripple/noise affected?

I admit I am finding it hard to get my head around this. There are lots of sites showing how to add a pass transistor but nowhere have I seen an analysis of how and how well it works. Even data sheets just show a 'current boost' application with no explanation or comment.

Cheers

Ian
 
Inside three pin reg is a pass transistor, usually emitter follower, and an error amp. Just imagine external transistor as an additional pass transistor in a configuration identical to output stage of an audio power amplifier in CFP topology.
 
ruffrecords said:
In which case, as far as the pass transistor is concerned this feedback must be achieved by a change in regulator input current?
Its all the same circuit.... When the regulator tries to deliver more current to the output that is reflected in increased current drawn at the input.

I fist saw this trick used to rig up high voltage/high current op amps from standard ones, by using the power supply draw to turn on external power devices (generally in common emitter configuration).  I built a modest audio amp for my little sister(RIP), into an old gutted hifi back in the 70s, using that approach. 
So, going back to my original question, how is the output ripple/noise affected?
I wouldn't expect it to be better, but could be similar.
I admit I am finding it hard to get my head around this. There are lots of sites showing how to add a pass transistor but nowhere have I seen an analysis of how and how well it works. Even data sheets just show a 'current boost' application with no explanation or comment.

Cheers

Ian
IIRC variants on these were published in ideas for design printed in electronic trade magazines back in 70s.

here is one such old op amp design 
Fairchild_Linear_Databook_uA777_Power_Amp.jpg


https://www.proaudiodesignforum.com/forum/php/viewtopic.php?f=12&t=699&hilit=current+booster&start=30

Here is a thread on Wayne's forum with sundry examples mostly of extended voltage range, but current boost works too.

JR
 
I remember amps like that from the 70s. On this side of the pond the HH range was the first I came across that used this technique.

I am going to sit down tonight in the quiet of my workshop and see if I can work out how current boosting actually works. Moamps' point about considering the series pass transistor is a great help. I am going to mentally disassemble the regulator into a reference, error amp and series pass transistor then add the current boost and see if it makes more sense.

Thanks for all the input.

Cheers

Ian
 
ruffrecords said:
I remember amps like that from the 70s. On this side of the pond the HH range was the first I came across that used this technique.

I am going to sit down tonight in the quiet of my workshop and see if I can work out how current boosting actually works. Moamps' point about considering the series pass transistor is a great help. I am going to mentally disassemble the regulator into a reference, error amp and series pass transistor then add the current boost and see if it makes more sense.

Thanks for all the input.

Cheers

Ian
Perhaps visualize what the waveform on a simple resistance in series with the regulator supply input would look like in response to load changes on regulator output. Using a fraction of that supply lead signal to provide more current into the regulator output supports the existing NF loop.

Of course the gain of this extra current assist could make the loop unstable if too strong.

Have fun...

JR 
 
Spent some time fighting LTspice tonight but got there in the end. Managed to simulate a TL783 regulating 300V down to 250V with a series pass PNP excited by a 10 ohm series resistor so it starts to kick in around 60mA. Bottom line is it is surprisingly good. For a load current change from 25mA (no current through external bypass) to 250mA (most of current through bypass) the ripple rose only by 50%.

I then disconnected the PNP by reducing the series resistor to nearly zero and repeated the experiment so this time the TL783 was handling all the current both time and in this case the ripple rose by 60%.

In both cases the ripple was about 60dB below the input ripple (10V pp at 100Hz). Adding a cap  from the adjust pin to ground significantly improves these results (by another 60dB).

Not sure how much of this is due to the simulation but it certainly augers well for the real world.

Cheers

Ian
 
ruffrecords said:
This article shows a typical current booster. Does that look like it is in the feedback loop to you?
You mean this? Indeed, it is in the NFB loop. the only issue there is that the current protection schemme still works for the LM317, but not for the pass transistor.
 

Attachments

  • lm317 boost.jpg
    lm317 boost.jpg
    28.4 KB · Views: 13
ruffrecords said:
There are lots of sites showing how to add a pass transistor but nowhere have I seen an analysis of how and how well it works.

I found this article, you might find it interesting (it's about PSRR of LDO connected after SMPS but there are some interesting graphs)
https://www.analog.com/en/technical-articles/increasing-output-current-of-the-ultralow-noise-ultrahigh-psrr-lt3042-200ma-linear-regulator.html
 
ruffrecords said:
Spent some time fighting LTspice tonight but got there in the end. Managed to simulate a TL783 regulating 300V down to 250V with a series pass PNP excited by a 10 ohm series resistor so it starts to kick in around 60mA. Bottom line is it is surprisingly good. For a load current change from 25mA (no current through external bypass) to 250mA (most of current through bypass) the ripple rose only by 50%.

I then disconnected the PNP by reducing the series resistor to nearly zero and repeated the experiment so this time the TL783 was handling all the current both time and in this case the ripple rose by 60%.

In both cases the ripple was about 60dB below the input ripple (10V pp at 100Hz). Adding a cap  from the adjust pin to ground significantly improves these results (by another 60dB).

Not sure how much of this is due to the simulation but it certainly augers well for the real world.

Cheers

Ian
I would think that for a HT PSU, regulation and noise were not very critical, since you probably have additional RC filtering along the way.
The differences shown in the article suggested by moamps seem to be almost irrelevant in the context.
I understand the need for analysing and assessing, though.
Beware that, in practice, running the LM783 at 60mA with 50V across may pose thermal problems.
 
abbey road d enfer said:
I would think that for a HT PSU, regulation and noise were not very critical, since you probably have additional RC filtering along the way.
It's a double edged sword. The SRPP output stages for example have a PSRR of just 6dB. They are the major consumers of HT so it make sense to feed them from a stiff supply. Early stages do indeed have additional RC filtering
Beware that, in practice, running the LM783 at 60mA with 50V across may pose thermal problems.
Which is one reason for considering sharing it with an external pass transistor. I very large heat sink is already in place.

I have zero experience in selecting high voltage high current PNP transistors. I prefer one in a TO220 package. A readily available one seems to be the MJE5852G. Is this a good choice? Are there better ones?

Cheers

Ian
 
abbey road d enfer said:
You mean this? Indeed, it is in the NFB loop. the only issue there is that the current protection schemme still works for the LM317, but not for the pass transistor.
It seems like the current limiting will still work in the pass circuit, after a fashion. When the regulator stops increasing current output, its draw from the supply pin will be limited also, limiting the pass device current being commanded. The total current limit output threshold will be modulated by the gain/ratio contribution of the added pass circuitry.
===

Thermal protection shut down and direct sensing functions will be operating open loop but should be serviceable if boost circuit components are well matched.

JR
 
abbey road d enfer said:
You mean this? Indeed, it is in the NFB loop. the only issue there is that the current protection schemme still works for the LM317, but not for the pass transistor.
Maybe I am being a bit thick but I still don't see it. The only NFB  I see is from OUT to ADJ. OUT is independent of IN and the current in the bypass transistor resistor depends on the load.

Cheers

Ian
 
ruffrecords said:
Maybe I am being a bit thick but I still don't see it. The only NFB  I see is from OUT to ADJ. OUT is independent of IN and the current in the bypass transistor resistor depends on the load.

Cheers

Ian

the regulator has an active differential stage inside comparing output V to reference V.  The pass transistor is part of the active current output stage (indirectly via the supply pin).

Just like a simple active stage with NF you can make it unstable by putting too much gain in the external current boost stage.

Sorry if this is not clear...

JR
 
JohnRoberts said:
It seems like the current limiting will still work in the pass circuit, after a fashion. When the regulator stops increasing current output, its draw from the supply pin will be limited also, limiting the pass device current being commanded. .....

The current thru regulator is limited with Vbe/R so regulator will never reach limiting current except when the output pass transistor is faulty  so the pass transistor isn't protected of current overload or short circuit. That's the reason why in some designs  one more transistor and sensing resistor is added (https://www.st.com/resource/en/datasheet/cd00000444.pdf , page 24).
 
ruffrecords said:
I have zero experience in selecting high voltage high current PNP transistors. I prefer one in a TO220 package. A readily available one seems to be the MJE5852G. Is this a good choice? Are there better ones?

Usually one small PNP  and power NPN is used as shown in attached schematic. You can find easily high voltage high current NPN type. Or you can use power MOSFET.
 

Attachments

  • high power reg.png
    high power reg.png
    45 KB · Views: 11
Thanks moamps; I understand those additions necessary to make a practical circuit. Right now I am trying to get my head around the fundamental operation, so..

@JR

Here is my sketch of a typical adjustable regulator with current boost:

currentboostedregulator.jpg


As it says in the data sheets, the regulator works to maintain 1.25 volts across the resistor between the ADJ pin and the OUT pin. You can see the diff amp doing this.

For a light load, the BJT is off and the circuit works like the original three terminal regulator.

If we assume the current out of the ADJ pin is very small compared to the load (which it is) then the current leaving the regulator output pin equals the current going into its input pin. When this current reaches a value large enough to start turning on the BJT, the BJtT starts to supply current to the load. If the load increases, the increase in current from the regulator is sensed by the 10R reissotr, tuens the BJT on some more and provide additional current to the load.

So it seems to me there is a separate feedback mechanism operating where the the BJT senses the regulator output current independent of the feedback via the diff amp in the regulator.

To me, the more obvious topology would have been to use and NPN BJT as an emitter follower so the regulator only has to provide bas current like this:

efollower-Regboost.jpg


The regulator raises the NPN base to ensure it is turned on enough to maintain 1.25 across the resistor between the OUT and ADJ terminals and the NPN definitely IS in the feedback loop.

But of course this will not work because with a three terminal regulator we cannot connect the actual supply output to the diff amp input so you have to invent some other method (such as the one in the first pic). To do the emitter follower version you need a four terminal regulator??

Cheers

Ian
 
Back
Top