current limiting for power supply

akis

Member
2008-07-18 1:09 pm
Hello

My bench power supply (over 35 years old) has died and I have decided to build a new one.

I am trying to implement a current limiting device based on a sense resistor. In addition I would like to be able to select the current limit level with a rotary switch, eg 50mA, 100mA, 250mA, 1A, 2A, 3A.

I am trying to use a very low value sense resistor to keep the voltage drop low. Suppose that I use 0.05R, then at 3A I will have 150mV.

I would like the circuit to limit as soon as the voltage drop across the sense resistor is 150mV and I would like the limit to be quite rapid. In other words I do not want to see the output voltage starting to drop as I approach 150mV, I want it to be sudden.

So I tried to use a comparator and an op-amp, just to experiment. I do not know how to wire up a comparator so that it conducts at 150mV (or at any other arbitrary voltage difference). It seems the comparator will conduct at the most tiny voltage difference.

I also tried an op-amp but that also fails because my circuit is absolute voltage dependent, eg it behaves differently depending on the absolute voltages present at + and - inputs of the op-amp, even though the differential voltage is the same.

Can someone please post any simple circuit I can experiment with.

Thanks
Akis
 

akis

Member
2008-07-18 1:09 pm
Now that you have shown me that there are such things as "current monitors" I have experimented in simulation with the chip INA170 buffered by an OPA340.

The problem however still remains with the current limiting action being too progressive and affecting the output voltage well before the limit is reached or well after it has been exceeded. I use an NPN transistor to short the voltage reference when the current limit is reached.

Is there a way to achieve a swifter current limitng action?
 
yes,
comparator (393) as you mentioned earlier.
An opamp can be made to operate as a reasonable comparator.
You must set up a ref voltage at one of the input pins.
Then when the monitor goes above or below ref voltage the comp swings to the opposite polarity triggering the limiter. A little positive feedback (big value resistor) can insert some hysteresis to remove the tendency to flutter.
 

akis

Member
2008-07-18 1:09 pm
yes,
comparator (393) as you mentioned earlier.
An opamp can be made to operate as a reasonable comparator.
You must set up a ref voltage at one of the input pins.
Then when the monitor goes above or below ref voltage the comp swings to the opposite polarity triggering the limiter. A little positive feedback (big value resistor) can insert some hysteresis to remove the tendency to flutter.

In simulation still, I removed the negative feedback from the op-amp / buffer that follows the monitor (OPA340), used a reference voltage on its negative pin, and now it works very nicely.

Except it oscillates / flutters. Say the current limit is at 100mA. As soon as the limit is breached, the op-amp goes full positive, the limiter cuts the driving voltage and the output drops. That makes the current fall well below 100mA, so the monitor drops and the op-amp goes to the ground, causing the limiter to stop limiting and so on.

Basically the act of limiting causes the error condition to go away and we go into a vicious circle.

I have tried the positive feedback trick but that does not help as the underlying condition (output 100mA) is being completely removed when the output voltage drops.

Any way I can make it more stable ?
 

tomchr

Member
Paid Member
2009-02-11 12:58 am
Calgary
www.neurochrome.com
Some suggested using a comparator for the current limiter. I would advise against that. The gain of a comparator is rather high at the transition point and the supply will chatter as you're describing.

I suggest lowering the gain by using the op-amp as an amplifier rather than a comparator. You may also have to make the feedback frequency dependent to avoid oscillation.

Franco Chapter 8 would be a good place to start.

~Tom
 
Some suggested using a comparator for the current limiter. I would advise against that. The gain of a comparator is rather high at the transition point and the supply will chatter as you're describing.

I suggest lowering the gain by using the op-amp as an amplifier rather than a comparator. You may also have to make the feedback frequency dependent to avoid oscillation.

........
~Tom

I would beg to differ! This is precisely what comparators are FOR; never being operated in linear mode.

Hysteresis from +ve feedback is the usual remedy, and, as I suggested earlier, check out fold-back limiting which has its own useful characteristics.
 
Last edited:
What are the specifications ? pls provide following details.

1) Regulated variable voltage min - max.
2) Are you looking for continues variable current or stepped current limiting as your previous?

some regulator ICs has this built in feature. so we don't reinvent the wheel. Put your needs in detail. so a circuit can be provided to you.

just found one that might be of help to you check the link Powersupply design
 
Last edited:

akis

Member
2008-07-18 1:09 pm
As I said in the beginning I am trying to replace my age-old PSU which I am very fond off. It sits on the left of the oscilloscope in this old picture I just found on these forums.

It is very old, the rotary switches are huge and made to withstand a nuclear explosion, the on-off switches are also big, and the down position is ON!

It is dual independent channel from 0 to 30 V in steps of 100mV and current limiting as 50mA, 100mA, 250mA, 500mA, 1A and 2 A.

The output voltage selection is effected by 3 rotary switches: 0-10-20 V, then 0, 1, 2, 3 ... 9 V, then 0.1, 0.2, 0.3 ... 0.9 V.

For current limiting again there is a beefy rotary switch that pre-selects the desired value.

I quite like the way you can select the current limit accurately without actually having to reach it, and I also like the way you can select the voltage output just from the switches.

For my project I will use a rotary pot to adjust the voltage, but will use a rotary switch for the current limit with preset limit values.

Reading above and in general I have learned to watch out for many things:

1 what happens when you power up, do you get spikes at the output ?
2 what happens when you power down?
3 what happens when you short the output (big output caps) ?
4 how much voltage drop do you get as the current increases / how stable is the output voltage
5 how much ripple when current increases
6 how good is the current limiting action
7 how stable is the output voltage wrt temperature changes (zener operating in 0 temp. coefficient mode???)
8 can you cascade two channels to make a dual rail supply, and can you then additionally ground the "0" terminal ?
9 how hot does it get and does it need a fan to cool the pass transitor (and how noisy will it be then)
10 do we need a slow start circuit (already got that one made if needed)
 

Attachments

  • guitaramp_pic4.jpg
    guitaramp_pic4.jpg
    68.4 KB · Views: 216
I would beg to differ! This is precisely what comparators are FOR; never being operated in linear mode.

Hysteresis from +ve feedback is the usual remedy, and, as I suggested earlier, check out fold-back limiting which has its own useful characteristics.

Yes, comparators are designed to not operate in linear mode. So, if what the original poster wanted was a hard cut-off (current limit reached = disable output) then maybe it would work. But that's not what he wants. A traditional "current limiter" is actually a current regulator, and must operate in linear mode. No comparators need apply.
 
I'd suggest finding the service manuals for some good quality bench supplies and looking at the circuit diagrams therein. Agilent, for example, makes (and has made) many nice supplies and all the service manuals are freely available on their website. The circuits can usually be easily be adapted to different voltages and/or currents, by changing just a couple resistors to set the new upper limits for current and voltage ranges. Try finding a model simiilar in specs to what you want then look at the service manual.
 

akis

Member
2008-07-18 1:09 pm
A hard cut off is something I have considered : you exceed the limit, power supply switches off, you have to press "reset" to get back up again, like a fuse.

For softer limit I will try to put some negative feedback today, so the op-amp will work like an amp and not like a comparator, maybe this will avoid the oscillations. I also do hope the simulation software is true :)
 
Hi,
the comparator allows the sharp transition. When it triggers it should bring in the current limit, not switch the current to zero. Let's say the current is 99mA, then due to temperature effect the test circuit starts to draw extra current, maybe passing through 100mA to 101mA.
The comparator could see the excess above 100.01mA and limit the output to pretty near exactly 100mA. With hysteresis added the limiter will not come out of limiting at 100.0mA but at some lower current of say 99.99mA or 99mA or 91mA. You design the hysteresis to suit what you want the limiter to protect against.

An alternative is to have two different regulators in series. That was my earlier suggestion and probably the simpler to construct.
Put a current regulator in the line between the supply and the voltage regulator.
While the voltage regulator maintains set voltage at a current less than the set limit, the current regulator is wide open and passes current with virtually no Vdrop. As current hits the set limit, the current regulator takes over and starts regulating, i.e. increases the Vdrop to achieve the current limit.
The voltage regulator now has insufficient input voltage to allow it to work as a regulator and falls to wide open. The net result is that you have a fixed voltage below and at set limit and a lower output voltage when the current regulator is active. I strongly suspect your PSU has been assembled this way.
The knob settings are feeding the correct feedback ratios to the two different regulators so that normally the voltage regulator is doing all the work.
 
Last edited:

MJL21193

Disabled Account
2007-03-10 1:20 am
A hard cut off is something I have considered : you exceed the limit, power supply switches off, you have to press "reset" to get back up again, like a fuse.

Hi akis,
How about circuit breakers? Size according to your transformer, giving a bit for a safety margin.
Adding complexity and cost, you could have 2 pairs of breakers, one pair for high current and the other for low and switch between these.
Dead easy, cheap(ish) and when mounted on the front panel, convenient.
 
Hi,
the comparator allows the sharp transition. When it triggers it should bring in the current limit, not switch the current to zero. Let's say the current is 99mA, then due to temperature effect the test circuit starts to draw extra current, maybe passing through 100mA to 101mA.
The comparator could see the excess above 100.01mA and limit the output to pretty near exactly 100mA. With hysteresis added the limiter will not come out of limiting at 100.0mA but at some lower current of say 99.99mA or 99mA or 91mA. You design the hysteresis to suit what you want the limiter to protect against.
It's quite obvious that you've never actually designed a current limiter. Your thought experiment above is interesting but impractical.
Replace "current" and "mA" above with "voltage" and "V" and see if your explanation makes sense for a voltage regulator. I thought not. I'll state it again: a current limiter is a current regulator. You don't just turn it on, and magically the current is limited to a pre-defined value. What happens when the load changes? How does your described current limiter keep the current constant at 100 mA? Current limiting is a dynamic linear process just like voltage regulation. Hysteresis is neither needed nor desired in a current limiter. Oscillations or instability of the control loop are no more (or less) of an issue than in any voltage regulator.
 

akis

Member
2008-07-18 1:09 pm
I cannot get the limiter to work smoothly using op-amp withnegative or positive feedback. It seems to want to drive the output voltage really low (as the action is to short the voltage reference) and then it picks up again etc.

So I tried your other suggestion, a PNP transistor acting as a constant current source before the voltage regulator, and this so far seems to work very well, no oscillations, it just lets the current rise until it hits a limit and then lowers the voltage.

The problem with this approach seems to be the voltage sacrifice we want to make at the current limit level. With the current monitor approach we would drop, say 100mV across the sense resistor. With the constant current source I am not sure how many mV to drop across the emitter resistor. We know the base-emitter junction is 0.6V, and intuition tells me that we'd need a multiple of that to have a stable performance. Currently I am dropping approximately 1.25V across the resistor - not sure if that is enough, or if we can get away with much less.