13.8V 20A power supply-Looking to add adjustable current limit?

[Xnke]

So normally in this situation, a pair of 723's would be used to get adjustable voltage and current limiting. I don't have any, AND I have PNP transistors for pass elements. I had a bit of time to sit down and play with this tonight and came up with the following:



Still need to figure the bleeder resistor on the filter capacitor, but it's not terribly important at the moment.

Input voltage is nominally 16V at 20A load, and output voltage is adjustable around 12 volts.

But, in this situation how should I add adjustable current limiting?

Yes, I know, use a pair of 723's. And ditch the PNP pass elements. Sure, sure. But, I have everything in the existing circuit to hand, and it's nice to be able to get by with what's in the parts bin.

So, how would YOU add adjustable current limiting to this? I'd like to be able to dial the current down to less than an amp, or up as high as 20A, with overcurrent situations behaving gracefully instead of crowbarring the output to ground and blowing a fuse, which is what I see commonly.

[Mazo]

You should add frequency compensation or this thing will oscillate.On the current sense problem you should decide first on the type of current sensor(a shunt or smth else?)If a shunt is okay then you should choose low or high side placement.If it were to me I would regulate the "negative" to still use the PNPs that you have on hand but getting the stability of emitter follower topology.If you insist on that schematic you should derive a current signal somehow and then probably just pull down the vref node to gnd based on the Isense signal and the setting of the current(a two op-amp solution is possible probably) that you want (another potentiometer)

[Xnke]

Frequency compensation will be needed-that is for sure. The 2N6051's have a minimum H_fe of 750, so some significant gain is available in the error amp loop for oscillations to occur.

I like the notion of pulling Vref down to ground based on Isense, a transistor in parallel with the Vref diode, (I will probably use a green LED or something, I've got a few zeners in appropriate values but the LED works fine.) and use the output of the Isense error amp to turn on the transistor, shunting current away from the diode and once 3-4mA is shunted off of it, it will shut off and the transistor's collector would become the voltage reference. This would tend to limit the lowest output current available to whatever the output voltage would be when Vref is a V_ce when the transistor is at saturation. This may be high enough still to cause a problem-I don't know yet.

I have thought about using one of the .33 ohm emitter ballast resistors on the pass elements as my current sense element-as the current through the resistor rises, the voltage drop across the resistor rises. So measuring the voltage drop across the ballast resistor and comparing it to the potentiometer setting would generate a current "error" that could be applied to the Vref node.

Using the Vref node in this manner I probably should drop the green LED and use a 6v or 9v zener diode, as this would give the current regulator more headroom to pull the current limit around. The again, the more room I give it to move in, the less likely it is to behave the way I expect it to...

I think by implementing the current limiter in this manner, any oscillation that developed would be current limited too, so if some outside influence or a deteriorating component in the frequency compensation network were to cause an oscillation, the current limiter should be able to cut the output current down enough to prevent damage to the pass elements. Maybe.

Going back to frequency compensation, I thought I had a good start on it, with the 10uF cap across the feedback loop to ground. There may need to be a miller compensation cap from the collector of the pass transistors to the positive-input of the op amp, but I will need some help figuring out how to figure that value.

[iMo]

This may work. You have to play with compensation, if required.
Simulation only.
The input voltage is an example, you need at least 18V thus it works.
I would add a transil (15-16V) to protect the output against peaks.
Also set the R_SHUNT as you need (now it limits at about 28A).

Adjustable current limit: to create a precise setting is not easy, you would need an additional circuitry. That would involve a propagation delay, you have to deal with then.
Or use a 1k trimmer (instead of the R12=100ohm) in the base of the current limiting transistor. Calibrate.
See another threads on the efforts in that regard. You may copy THIS as the current limiting transistor in LM723 is wired the same way as the Q5 in my schematics below.

No warranties of any kind are provided, use at your own risk

[Kleinstein]

The circuits with the output from the collector side usually  need a rather large capacitor at the output so more like >1000 µF than 10 µF. Compensation can still be a little tricky if fast response is wanted.
The problem is not the current gain (Hfe), but the voltage gain added by the output stage. The output stage is more like setting the current.

The circuit from imo might want a different OP, that gives a higher output voltage to get away with less input voltage.

[edavid]

It is better to use the PNP pass transistors in emitter follower mode, which means putting them on the low side of the supply.

Even with the high beta transistors, you still need a base driver to get 20A out.

You probably want a few volts extra for the control circuit supply, so you don't have to use an excessively high unregulated input voltage.  An extra transformer is cheaper than a giant heat sink.

[Xnke]

Putting them on the low side of the supply is fine-I can rework this as an emitter follower, but is the only reason to do so because "emitter followers are more stable"?

I'll get paid Tuesday or so, I could just wait and order some NPN pass transistors. But making 68$/week that is a bit of a superfluous expense.

I could add a second transformer for control voltage, that's no big deal. The parts bin is *very* heavy, and is about 10 meters x 20 meters. Same with big heatsinks...I sort big heatsinks by the pound, not by dimensions.

Now the last question about the PNP low side regulation-Will this (seeing as I need a negative ground) cause issues with other connected equipment by virtue of having a "moving" ground?

[soldar]

Now the last question about the PNP low side regulation-Will this (seeing as I need a negative ground) cause issues with other connected equipment by virtue of having a "moving" ground?

The ground is not "moving"; the ground is wherever you decide it is and everything else "moves" around that.

[Zero999]

Frequency compensation will be needed-that is for sure. The 2N6051's have a minimum H_fe of 750, so some significant gain is available in the error amp loop for oscillations to occur.

The output transistor's Hfe makes little difference to the voltage gain of the circuit and whether it needs frequency compensation or not. A high Hfe just means it needs less base drive for the same collector current, as a lower Hfe device.

The problem with the common emitter output stage is its transconductance depends on the load current, which makes if very difficult to stabilise across different loading conditions. A classic problem with this low dropout topology is it tends to oscillate when the output capacitor has too lower ESR.

[Mazo]

We say that emitter followers are more "stable" because they have voltage gain of a little less than 1.While your collector output topology has a voltage gain that also varies with load current and the stability is heavily dependant on the output cap.The extra voltage gain is a problem because the op-amp you are gonna be using is most probably unity-gain stable and will have something on the order of 45 to <90degrees of phase margin at it's unity gain(open loop that is) point.By introducing additional voltage gain you will degrade the phase margin by moving the unity gain point to a higher frequency which leads to (ringing and lots of it) or you are gonna make it 0(oscillation).It's not like the collector output is mission impossible it's just that it isn't the "slap a big enough integrator cap and be done with it" that the emitter follower offers.If you like the circuit you devised but want help stabilising it PM me or write here we will gladly help.

[floobydust]

NPN emitter-followers as pass transistors have a voltage gain of ~1 and thus the highest speed.
PNP Darlingtons are the slowest possible pass transistor configuration. You are using them with voltage gain, looking at the big (voltage) feedback loop including the op-amp. A few mV change at the base can cause many V swing at the PSU output.
I have stability issues with them because hFE can vary from 750 to 4,000 with 18,000 given as max. hFE sags under load, changes with temperature and the loop gain is all over the place. Stability can be difficult.

LM324 output stage is Darlington, you lose several volts for VOH. I don't think it can actually go high enough to shut off the 2N6051's.
You might need a start-up resistor, I'm not sure what the LM324 will do with both inputs at ground on start-up.

[MarkF]

I saved this from somewhere.  Maybe it will help.

   

[Xnke]

Worked out a trade with a local to obtain a few LM723's and 2SD111 transistors, that will pretty well push this design off the table. Started the metal work on the cabinet tonight, cardboard aided design is wonderful for sheetmetal cabinets.

Now the only issue involves how to quiet down this power transformer. It has a bit of hum when drawing 40A DC off of it, however it does not get hot...I can keep my hand on it indefinitely at that power level. I think it's up to the task. It's about the same size as a decent old "1000 Watt" microwave oven transformer (not one of these puny new ones...) that weighs 20lbs.

[floobydust]

I mount noisy power transformers on rubber isolation mounts. The vibration doesn't get coupled to the chassis. This is for 1000VA up. Very quiet afterwards.

[not1xor1]

Putting them on the low side of the supply is fine-I can rework this as an emitter follower, but is the only reason to do so because "emitter followers are more stable"?

I'll get paid Tuesday or so, I could just wait and order some NPN pass transistors. But making 68$/week that is a bit of a superfluous expense.

I could add a second transformer for control voltage, that's no big deal. The parts bin is *very* heavy, and is about 10 meters x 20 meters. Same with big heatsinks...I sort big heatsinks by the pound, not by dimensions.

Now the last question about the PNP low side regulation-Will this (seeing as I need a negative ground) cause issues with other connected equipment by virtue of having a "moving" ground?

There is another reason why it is better to use the positive rail as common (ground, not earth): single supply opamp like LM358/LM324 inputs and outputs can get very close to the negative rail, but not to the positive one.
In any case you can use simple charge pump (2 diodes + 2 caps per rail) to get two other rails and so overcome all opamp input/output limits.
The messy circuit below needs to be revised (it was a quick copy/paste from pieces of other circuit simulations), but shows:
- 1 more negative and 1 positive rail via charge pump
- adjustable voltage and current and fast current limit
- leds displaying CC/CV mode
- slow turn-on

[floobydust]

OP, so you've changed over from PNP Darlington to NPN power transistors and op-amps to LM723?

not1xor1, that looks reasonable and I use an aux. winding on the power transformer or a small transformer to make the control op-amp's power. There needs to be an undervoltage lockout to prevent the main output spiking as the aux. rails come up or power down at a different rate than the main raw DC.

[not1xor1]

OP, so you've changed over from PNP Darlington to NPN power transistors and op-amps to LM723?

not1xor1, that looks reasonable and I use an aux. winding on the power transformer or a small transformer to make the control op-amp's power. There needs to be an undervoltage lockout to prevent the main output spiking as the aux. rails come up or power down at a different rate than the main raw DC.

C17 allows slow rise of output voltage at turn-on
I fully simulated a variation of that circuit on turn-on/off via a voltage controlled switch and it looked like it behaved nicely.
I'm slowly building a first low voltage-low power (18V-1A) prototype and will soon check how it behaves in the real world.

[spec]

Hi Xnke,

I see you have changed tack and abandoned the original circuit- a shame because it would have been interesting to see the development of that architecture which is great for DC performance, with two reservations about your implementation, but more difficult to compensate, as has been discussed.

I started drawing up an outline circuit illustrating one method to fit a precision (apart from a small standing current) and variable current limit to your original circuit, which may also be of of interest, even for your 723 approach, so I thought I would post it anyway (frequency compensation is not included):

By the way, with a few other mods, you could fit NPN output Darlington transistors to the circuit shown and get a conventional PSU, with an emitter follower output, which has no voltage gain and is thus easier to compensate.

UPDATE #1 2019_01_30  The transformer voltage and current indicated on the schematic are a hang-over from a previous design and are not correct for this application.

[floobydust]

Bench lab power supplies frequently fail or misbehave. I've repaired so many and I'll play my harp here about the design issues that get missed. Spice sims need to include several scenarios.

Because your pass-transistors are normally full on, the op-amps needs to be up and running to back off the output.
When the power supply is switched on, there is a delay for the op-amp rails to come up, so the op-amps are late controlling the pass transistors. This causes a nasty overshoot.

When the power supply is switched off, there is a delay for the op-amp rails to falloff.
If, at the moment, the power suppy was running a heavy load, the main filter caps drop off fast and no drama.
But a light load can mean the op-amp rails falloff before the main filter caps are discharged and you get a nasty upsurge because the op-amps are powered down before the main filter caps are empty.

On the day OP needs to charge a car battery, the power supply dies:

If the power supply is off when a battery is connected, you get backfeed through the pass transistor's E-B junction. Same if the load has a large capacitance.
If the power supply is on when a battery is connected, but at a lower setpoint (say 5V with 12V battery connecting), you also can get backfeed through the pass transistor's E-B junction.

These scenarios can cause the pass transistors and/or the op-amps to fail, or your load to get damaged.

[Doctorandus_P]

Power suplies as drawn with PNP transistors in this configuration are inherently less stable than with (usuallly) NPN transistors in emitter follower configuration.

If the base voltage of an emitter follower is held at a constant voltage, the output will not change much regardless of the current.

You've drawn your PNP transistors as a current source and are using the controll circuit to maintain a constant voltage on the output. This can be done, and is used in "low drop" voltage regulators, but it's more difficult to tame.

You have 3 emitter resistors to stabilize the current through your power transistors.
The voltage over these resistors varies lineairly with the output current.
You can use 3 resistors(100 ohm to 1k or so) to average the voltages over these resistors.
Then you can use this voltage for your current limiting.
The easiest way  is to use an PNP transitor with the Emitter connected to V+ and the base to the common point of these 3 resistors. This transistor wil be opened when there is >600mV over your power resistors.

But your LM324 probably can not deliver enough base current for your power transistors. Usually the power transistors are used in a darlington or shizalsky configuration.

For stabilisation you have used a simple zener + resistor. Stabilisation of this ciruit is not so good. A very nice (and well known) circuit is to not provide the current through that resistor from the input, but from the output of an opamp that buffers the zener voltage. This opamp has a constant voltate output, which results in a constant current through the restor, which stabilizes the voltage over the zener diode. You've got plenty of opamps in a lm324, so with the addition of 2 resistors for a bit of amplification you're done.

[spec]

You have 3 emitter resistors to stabilize the current through your power transistors.
The voltage over these resistors varies lineairly with the output current.
You can use 3 resistors(100 ohm to 1k or so) to average the voltages over these resistors.
Then you can use this voltage for your current limiting.
The easiest way  is to use an PNP transitor with the Emitter connected to V+ and the base to the common point of these 3 resistors. This transistor wil be opened when there is >600mV over your power resistors.

That is a neat idea but the OP wants current limits of 1A to 20A. How would that be realized with your approach?

But your LM324 probably can not deliver enough base current for your power transistors. Usually the power transistors are used in a darlington or shizalsky configuration.

You didn't notice that the output transistors are Darlingtons.

[Doctorandus_P]

You have 3 emitter resistors to stabilize the current through your power transistors.
The voltage over these resistors varies lineairly with the output current.
You can use 3 resistors(100 ohm to 1k or so) to average the voltages over these resistors.
Then you can use this voltage for your current limiting.
The easiest way  is to use an PNP transitor with the Emitter connected to V+ and the base to the common point of these 3 resistors. This transistor wil be opened when there is >600mV over your power resistors.

That is a neat idea but the OP wants current limits of 1A to 20A. How would that be realized with your approach?

Can be done in a multitude of ways.
You can use one of those INA high-side current sensor chips to measure that voltage.
You can build a simple current mirror with 2 transistors.
You can use an opamp in which the inputs work (slightly) above the Power supply rails. LM324 is not fit for this, but another old jelly bean part such as the TL072 can do this easily.
It would also need a potentiometer to compare it to.

A fun (and unusual) circuit to implement the output of the current limiting section is to put the transistor of an optocoupler in series between the bases of the output transistors and the output of the voltage regulator opamp. Normally it is in saturation, but when current limiting is needed the transistor is closed.

But your LM324 probably can not deliver enough base current for your power transistors. Usually the power transistors are used in a darlington or shizalsky configuration.

You didn't notice that the output transistors are Darlingtons.

If OP is using darlingtons, he should have drawn them. That is what schematics are for.

[edavid]

But your LM324 probably can not deliver enough base current for your power transistors. Usually the power transistors are used in a darlington or shizalsky configuration.

You didn't notice that the output transistors are Darlingtons.

The minimum beta of those 2N6051 Darlingtons is 750, so at 20A output the LM324 would have to sink 27mA, which is outside the datasheet worst case spec.

But all this is moot because OP said he is now going to use 2SD111 NPNs, which have a minimum beta of 10 

[Xnke]

Fortunately for the confused and irritated, most of this thread is disused now. I was able to trade with a local who has helped me out with getting some parts and a good, proven design, one he uses and has used for 15+ years, and he was able to go through the design with me so I understand it now.

Thanks for the help, and for explaining why the PNP/LDO style has it's pitfalls and snares to watch out for-I will still have to find a use for these 2N6051's at some point!