Latching Short-Circuit Detection Circuit

[sean6541]

Sorry to bother, I have a quick question regarding the LATCH functionality of a comparator (LT1711) that is being used in an overcurrent detection circuit (see attachment). The circuit is part of a hard-switched full bridge inverter that will be operating at about 200VDC with a frequency of ~100kHz. In the circuit, GND is the 0v power rail. BRIDGE_GND is connected to the sources of the lower NMOS transistors and to GND through the linear hall-effect current sensor. SGND is the isolated 0v rail for the logic side of the MOSFET drivers and microcontroller. I_SENSE is the analog output from the current sensor and is connected to the ADC of the microcontroller.

The functionality I am looking for is to have the microcontroller monitor current and if it exceeds a programmed value for a programmed time, stop the switching output to prevent overheating of the MOSFETS. For safety, I want the short-circuit protection handled without requiring the microcontroller. To achieve this, I have come up with the attached circuit, where RV1 sets the delay and RV2 sets the trip threshold. Once a short-circuit is detected, I want the DIS(able) output to latch and remain latched until the power has been cut.

My question: Can I feed the comparators output directly into it's LATCH enable pin to keep the output latched until the power is cut? Something about it seems to easy. 

[Ian.M]

The datasheet says the latch setup time is typ. 1 ns.  Unfortunately it doesn't specify upper or lower limits, but a small delay between the output and the latch control would be advisable.  However you have a bigger problem - What guarantee do you have that the comparator output won't glitch during powerup, resulting in it starting up latched?

The easy fix for both is a logic gate between the output and the latch control input so you can unlatch the comparator on powerup, and to provide a small delay.  I'd suggest a NOR gate, driven by the comparator's inverted output to isolate it from loading on the non-inverted output.

[sean6541]

That makes sense. It will be fairly easy too since there is already a 7402 with an unused channel in the logic part of the circuit.

[Ian.M]

Its a *VERY* fast comparator.  Running long traces to/from it could be problematic.   Unless its conveniently very close to your spare NOR gate, I'd use a 74LVC02 single gate NOR right next to it.   Your spare NOR elsewhere could be used to generate the active high RESET pulse the gate needs on powerup, if you don't want that to be MCU controlled.

[SiliconWizard]

Yes, it's a fast comparator. The datasheet talks about that in the "High Speed Design Techniques" section.

In the OP's case, I suspect they chose it not because it was fast, but because it has this output latch feature? If so, a separate latch and a more tamed comparator could also be advisable. The latch itself can just be a set/reset flip-flop. Connect the "set" input to the comparator's output, and the "reset" input to a simple reset circuit (RC) that'll guarantee resetting it upon power-on. And with this setup, there won't be any possible glitch.

[sean6541]

An LM393B and separate SR latch was my original plan, but I though I could cheat and end up with one less IC by using the latching function of the LT1711. If I'm going to add another IC, I'll just go back to using a separate latch.

[Zero999]

Any comparator can be made to latch with positive feedback. A diode from the output to non-inverting input will cause it to latch on. If the voltage connected to the non-inverting input, has a low impedance, add a series resistor, so the comparator's output isn't loaded down. Care needs to be taken that the circuit will always start in the unlatched state, which can be done by connecting a capacitor from the inverting input to +V, or non-inverting input to 0V, depending on which is the reference. It might also be a good idea to add a diode across the capacitor and a resistor in series with the reference.