Hello forum,

I'm designing a current limiting circuit as shown in the schematic. So far it is working as expected. It should limit the current to about 580 mA in a few micro seconds. (I was testing it with different parameters, that is why the directives maybe look a bit wild)

When I was about to create a PCB, I just realized that there is a lot of power dissipating in the MOSFET in when it is regulating the current. Since it has to take the complete voltage times limiting current. The power is reaching almost 5 W which might be to much in my application in terms of heat.


I don't really need the current to be regulated, so I was thinking if there might be a way to self-adjust the mosfet as soon as it starts limiting and use it as a switch.
I tried a bit with other transistors trying to detect the change of VOUT and pinning the gate of the MOSFET so it will be open, but without success.

Alternative to this circuit would be some kind of crowbar circuit, but it needs to be really fast and should not let any transients pass.

I appreciate any help and suggestions for the circuit.
Thanks in advance
Snoozy

 

So you appear to have two different requirements.
Do you want to limit the current to 580mA, or to stop the current completely when it goes above the limit?

How accurately do you need to detect the limit current?

 

Look up "foldback current limiter".

 

A crowbar circuit usually refers to an over-voltage protection circuit. one that slams a pseudo-dead short across the output if the output voltage exceeds operating parameters. Often, this is done with an SCR and a resistive divider network.

For a power supply, the two most common methods of current limiting are constant current and foldback. Constant current limiting recovers automatically - remove the too-low impedance across the output, and the output voltage returns to normal. Foldback is similar to an electronic circuit breaker - the output is latched off until some kind of reset function is performed.

https://en.wikipedia.org/wiki/Current_limiting

ak

 

... a current limit ...
Snoozy

describe your LOAD -- what should happen to ITS voltage if the current limit is not reached

describe your OVERLOAD condition -- is IT a limited set of different events --or-- might it be unpredictable

? Is there a damping/tank capacitor in parallel to the LOAD (at the V.out side)

 

Foldback circuit (but I don't know how to sweep the value of the load resistance, so I contrived it with V2)

 

Thank you very much. I definately will have a look at the foldback circuit. Looks like it shuts down as I want it.

The circuit does not need to be a regulator. It should just work super fast and does not be super accuarte, but the maximum fault current needs to be know (able to be calculated).

The load of the circuit is normally about 6 Ohm and may vary from 0 to infinity in the fault scenarios.
It might rampup or be applied suddenly with a switch.

I was also thinking to add a thermal fuse as a protection for the transistor. That would be sufficient for my application, but in any case I'm going to try to understand the foldback cirucit and see if it will pass my requirements.

That were really great answers of you all guys!

 

I think I get the function of the foldback circuit now. A problem is the voltage drop over the collector emitter and R5. I will try to exchange the bi polar transistor with a mosfet. And play a bit with the circuit.

BTW: I just realized, with the load of the circuit (6 Ohm) and the given voltage the current is way higher under normal operation than under fault conditions. I need to double check that since this will be part of a complex circuit.

 

Power dissipation in the pass transistor usually is not a problem. During normal operation it has almost 0 V across it, and during the fault condition it is off.

Measuring a current so you can take some action requires either a shunt resistor (for the fastest response, but with a small voltage drop across it) or a Hall effect current sensor (for absolutely zero voltage drop but a slower response time).

Linear Tech (now a part of Analog Devices) has hot-swap and electronic circuit breaker control chips that use a power MOSFET's Rdson as the sense resistance. A possible problem here is is that while a minimum Rdson value is guaranteed by the datasheet, the actual value for any particular device can be much less.

For a given power transistor device package, a power MOSFET will (almost always) have a lower ON resistance than the effective resistance of a saturated bipolar transistor. A possible issue is turn-off speed. You have to suck a lot of charge out of the FETs gate-source capacitance. This can be calculated and dealt with when you know the maximum turn-off time your system can tolerate.

ak

 

In the cirucit I posted above, the power dissipation is a problem, since it is a regulator and in the regulating state the voltage drop over the mosfet will cause power dissipation.

So I'm not 100 % sure if you are talking about the fallback circuit. In the current regulating circuit I can choose a small shunt, which is fine.
In the fallback, I have the shunt (R1) and the voltage drop over Q2. It would be cool, if Q2 could be replaced by a MOSFET. But I'm not sure about that yet.

An IC might be an option, but is difficult to rate since this circuit has to pass a standard were rating of complex ICs might be difficult and simple components are usually better to rate.

 

In the cirucit I posted above, the power dissipation is a problem, since it is a regulator and in the regulating state the voltage drop over the mosfet will cause power dissipation.

So I'm not 100 % sure if you are talking about the fallback circuit. In the current regulating circuit I can choose a small shunt, which is fine.
In the fallback, I have the shunt (R1) and the voltage drop over Q2. It would be cool, if Q2 could be replaced by a MOSFET. But I'm not sure about that yet.

An IC might be an option, but is difficult to rate since this circuit has to pass a standard were rating of complex ICs might be difficult and simple components are usually better to rate.

Q2 can be MOSFET or bipolar.

 

There are even "electronic fuse" circuits that will switch off if the current exceeds some value. So an adequate description of the functionality is what will decide the best choice for an actual circuit.
That means describing 1.what happens below the limit set-point current and 2. what happens AT the set-point current, and 3. What happens if the set-point current is exceeded by some defined amount, and 4. the condition after the set-point current has been exceeded and things are back to where the current would not be exceeded.
ALSO, we need to know how stable the supply voltage is, and will it vary, or stay constant.
It could also be useful to know if whatever the load device is, if it's common connection is isolated from everything else. Or is the load common side connected to "ground"?

 

I tried to replace it, but it didn't work. I will check that again and see if I can find a type that is working.

 

The circuit is battery powered and I need to assume that a maximum voltage of 8.4 V is present. The supply voltage can go down as two lithium cells (and the battery monitor) allow. That might be around 7 V.

In normal operation, below the set point the circuit should no do anything and the voltage drop should be low as it should not affect the battery live time.

At the setpoint the current should not exceed the set limit. It may shut down. Important is, that the transient energy passed to the load is very low. So it should act fast.

If it exceeds, the setpoint, the same should happen. Actually the circuit needs to be tested with all kinds of ohmic loads from 0 to infinity.

It would be good, if the circuit might return to normal operation after beeing in triggered. Since the device cannot be reset (the battery cannot be disconnected). Anyway, this is not a requirment, since the circuit should only be triggered by a fault event and than the device should not be used anymore.

The load is a little motor that is working with a PWM under normal operation. It has freewheeling diodes in parallel, so I hope inductive effects should not reach back. So the load is the motor with a MOSFET controlling it. But the microcontroller circuit is supplied by the same source (LDOs are following), so there is a certain capacitance present as well (~4.4 µF).
There is a common ground for all the circuit.

 

Be careful what you wish for!
A current limiter that is too abrupt will never start up if presented with a capacitive load.

 

Hm, you mean because during charging the capacitors a high current flows?
But that is a pretty normal capacitance that you would have in a circuit. And as far as I understood, the foldback circuit is normally used for battery protection.
Any suggestions how to deal with that? Speed is really the matter here. The requirement says not more than 20 µJ shall pass in case of a fault.

 

Hm, you mean because during charging the capacitors a high current flows?
The requirement says not more than 20 µJ shall pass in case of a fault.

That's the amount of energy stored in 560nF @ 8.4V.
You have to be able to distinguish between the 20uJ that charges the output capacitor and 20uJ that goes into a fault.
Perhaps you could give it a bit of a time delay to get the capacitor charged, but that doesn't deal with a fault at switch on.
Always keep at the back of your mind that people who can't do engineering get administrative jobs (especially if they go to the pub with or play golf with senior management) Then they are in a position where they get to write specifications.

Interesting to note that 580mA @ 8.4V exceeds 20μJ in 4.1μs - that would be a mighty quick fuse!

 

As far as I tested the first circuit I posted, it is fast enough. I'm not 100 % sure if I did all correct in the simulation, but I checked the integral of voltage times current and it was below 20 µJ. I can adjust the speed with the gate resistor.

Anyway, I know it is a really short time and probably even difficult to measure with an oscilloscope since you also need to have a really good switch to add the load (like a mercury switch).

With the first circuit posted the start issues are not given, since it just might to limit the charging current.

So maybe I should screw the idea of a foldback and add a thermal fuse to deal with the heating issues.
I found this kind of fuse from Bourns which is really nice because of the low cut off temperature and size. Anyway, it seems to be difficult to handle and I never used something like that.
SA72SB0

 

EDIT (Seems like there is only a small time for doing that):
I think the circuit passed, because the voltage drops super fast at the output in the simulation. I probably need to add the capacitance to get a realistic simulation.

 

@Ian0

I don't know how to sweep the value of the load resistance

Set the resistor value as R=<expression>. e.g. R=v(x)*time . (R can be positive or even negative, but zero can give odd results).