Short circuit protection for boost converter

[treez]

Hello,

I am designing a 3W boost converter to drive LEDs. (its an emergency LED light, which comes on when the primary power source fails)

Vin is a 3V6 battery. (three 1V2 NiCd cells)
It must have output short circuit protection.

Do you know of a very quick way to turn off a series PFET in the 3V6 rail, when a short circuit is detected?


There is also a 10V internal rail which i could use to power such short circuit protection circuitry. (the 10V rail comes from a micro boost converter which draws from the 3V6 battery)

I am thinking of doing something like a series sense resistor in the 3V6 rail, and "reading" this with a differential opamp (powered from the 10V rail) , and then the output of this feeding into a comparator which somehow switches off the series PFET, so that the short doesnt drain down the battery.

I am thinking that if the short draws say a 1ms, 40A spike out of the battery before the PFET gets switched off, then thats bad but not too bad?

...anyway, the differential opamp will be inaccurate due to common mode issues....but then who cares?......i am not looking for accuracy, i am just looking to detect dirty great big currents........am i on the right tracks here?

 

[Tahmid]

You can think about these methods:

You could sense the output current and turn off the MOSFET if the current is too high.

You may also use a linear current limiter at the output (since output power is pretty low).

Also, as current flow increases, battery voltage decreases (due to internal resistance). So, maybe you could use the voltage drop (below a certain level) to detect short-circuit or over-current.

Hope this helps.
Tahmid.

 

[treez]

thanks,

though i cant use the limiter, as this is an emergency light, and i must make it as efficient as possible

 

[treez]

I am wondering if the following method is ok for short circuit protection? (pdf attached)



The Opamp will suffer from common mode effects, so will be innaccurate, but i only want it to detect a big current, and dont need massive accuracy

...do you think that this schematic is ok for short circuit protection.?

the following short circuit current spikes would flow, but not all the time.....

https://i48.tinypic.com/116ne68.jpg

 

[alexxx]

I am wondering if the following method is ok for short circuit protection?

You could use the sence resistor, small enough to avoid resistor overheating and power waste. I believe hysterisis is required to the comparator to avoid output swinging around the current cut-off value.
However, and due to the nature of the application, you should consider that such kind of protection techniques without the use of an MCU could lead to leds blinking, something unacceptable in emergency lighting.

 

[treez]

you should consider that such kind of protection techniques without the use of an MCU could lead to leds blinking, something unacceptable in emergency lighting

Thankyou for your input Alexx...though if i may i should raise issues, as you may know something that i dont here?.....

alexx.....i dont see how having this circutiry on the PCB will make the LEDs blink......this circuitry (short circuit protection) only does anything if the LEDs become short.

If there is a short circuit, i think the comparator jittering is going to be the least of the problems........as long as the comparator trips the FET to switch off when theres a short, i am not concerned if it jitters....it can jitter all it likes....surely?

 

[alexxx]

If there is a short circuit, i think the comparator jittering is going to be the least of the problems........as long as the comparator trips the FET to switch off when theres a short, i am not concerned if it jitters..

This is a choice you have to make. If it OK for you, then there is nothing more to be added by my side.

 

[treez]

..to be honest i dont know, maybe jittering comparator draws lots of current?

.....maybe the monstable needs an input low for a certain time, and a jittery input may not set it off.....so i dont know, maybe the jittering really is going to be bad(?)