sanity check - window comparator

[poorchava]

Hey,

I've been designing a circuit that needs some means of protection from supply voltage being either too high or too low. I came up with something like this:



The circuit is supposed to cut off the power supply rail if it's below 20V or above 30V. High accuracy is not needed.

Obviously the mosfet is first with reasonable Rds from LTSpice library and resistor values are totally out of my ass. Spice says it will work, but maybe I'm missing something obvious?

[Jay_Diddy_B]

Hi,

I can see a few features of this circuit:

1) You have a temperature dependency. The Vbe threshold of the transistor will reduce by 2mV degree C. This may be o.k.  because you don't require precise thresholds.

2) The other feature that I see is that you could end up with the MOSFET in the Linear region. This could mean high dissipation in the MOSFET. some kind of hysteresis is required so that the MOSFET is on or off.

Linear Technology has an IC that implements this function. Have a look at the LTC4365.

http://www.linear.com/product/LTC4365


Regards,

Jay_Diddy_B

[nctnico]

I'd use at least a real comparator like the LM339 (quad) or LM393 (dual). Besides that you'll need a clamping device to limit voltages (=blow a fuse) beyond the supply range of the components.

[dannyf]

It will work, better with R3/R7 replaced with appropriate zeners and the associated resistors resized.

[poorchava]

I'd use at least a real comparator like the LM339 (quad) or LM393 (dual). Besides that you'll need a clamping device to limit voltages (=blow a fuse) beyond the supply range of the components.

That's the main problem. A TVS is already in place, but it has a downside of blowing up when input voltage exceeds clamping threshold for longer periods of time. The device has to be flame-safe and all the magic smoke has to be kept inside no matter what.

Hi,

I can see a few features of this circuit:

1) You have a temperature dependency. The Vbe threshold of the transistor will reduce by 2mV degree C. This may be o.k.  because you don't require precise thresholds.

2) The other feature that I see is that you could end up with the MOSFET in the Linear region. This could mean high dissipation in the MOSFET. some kind of hysteresis is required so that the MOSFET is on or off.

Linear Technology has an IC that implements this function. Have a look at the LTC4365.

http://www.linear.com/product/LTC4365


Regards,

Jay_Diddy_B

The temperature dependancy of junction voltage drop is not an issue. I am aware of the possibility of the mosfet ending up in linear region, but I could come up quickly with some simple way of adding hysteresis. I will work on that and see what I can come up with.

The LT chip seems like an ideal solution, but it's rather expensive. But it made me search for similar devices. What do you think about LM5060?
http://www.ti.com/product/lm5060-q1

Seems to do pretty much the same thing aside from reverse polarity protection, which is not required in this case (diode bridge before the circuit).

[nctnico]

I'd use at least a real comparator like the LM339 (quad) or LM393 (dual). Besides that you'll need a clamping device to limit voltages (=blow a fuse) beyond the supply range of the components.

That's the main problem. A TVS is already in place, but it has a downside of blowing up when input voltage exceeds clamping threshold for longer periods of time. The device has to be flame-safe and all the magic smoke has to be kept inside no matter what.

Then use a bigger TVS and a PTC (self resetting) fuse. I'd use through-hole parts for both.