(Simple?) Transistor latch not working!

[Christopher]

Hi All, Above is my _simple_ transistor latch which should just work.

The comparator and inverter part of the circuit is working OK, Q4 is switching on and off as expected.

The idea is when the comparator goes out of spec, the LED stays red. It doesn't. When the comparator goes back into spec, the LED turns green which it shouldn't! The feedback thru R9 should keep the LED red....

It simulates fine, on the bench however it does not work. I have tried reducing R9 to 10K & 4k7 to increase base current, but no cigar... Any ideaS?

[Christopher]

Removing Q3 made the red LED work & latch!

Any "simple" way of making the green LED on when the red LED is off and vice versa?

[onlooker]

Why do you have the outputs of IC3A and IC3B tied together?

If I understand the purpose correctly, one needs more or less a standard bistable circuit (2 transistors or a 555) with the LEDs as part of the  loads. The two reset lines of the bistable should connect (use diode and capacitor to trigger on edge) to the globle reset and the "out spec"(?) signal respectively. Even if this signal needs inversion, one should be able to use no more than 3 transistors for the cicuit.

 You can Google for all kinds of bistable circuit examples and go from there.

[Andy Watson]

Removing Q3 made the red LED work & latch!

Are you sure that you have the red and the green correct? I can see how installing Q3 would hold the red on (because Q2 is biased on via R9 and R8).

Any "simple" way of making the green LED on when the red LED is off and vice versa?

Simple bi-stable like in this thread:
https://www.eevblog.com/forum/beginners/i-have-a-couple-of-questions-about-this-burglar-alarm-circuit/msg673437/#msg673437
Obviously you would need to invert the polarity of the circuit - and swap NPNs for PNPs, but the principle should be good.

[T3sl4co1l]

Easier on the eyes,



What's missing is a strong pull-down on the PNP.  Your logic styles are inconsistent, so the voltages/currents are as well: you've got collector-to-base style TTL on the first three transistors, but then there's the PNP in there.  An additional NPN in the same collector-to-base style format would work.

It might be better to use PNP for everything, so you get strong drive into the LEDs, no threshold violations, and you can omit the first inverter since everything will be complementary.

If you're not against TinyLogic (and equivalents), I'd just as well suggest a dual NAND/NOR (e.g., 74HC2G00) to build a basic R-S flip-flop.  You also get complementary outputs for free!

A dual NAND/NOR can also easily be built from a couple of transistors.  Example:



A pair of transistors can even be omitted if you manage the input-to-base and base-to-ground resistor dividers carefully, so that two logic inputs (and individual resistors) are wired into a single base (and B-GND resistor) to produce either NAND or NOR behavior.

Tim

[Andy Watson]

Why do you have the outputs of IC3A and IC3B tied together?

They're OR-tied open-drain outputs. I assume it's forming a window detector.

[Christopher]

Thanks for the replies...

Yes the comparators are wired up as open collector ors


I've already been a stupid idiot and because the simulation worked I went ahead and got 20 boards made up so have to bodge up what ive already got....

For now the easiest thing to do is keep the green led on all the time, and let it latch orange on a failure! But I'd like to keep it red and green simply if possible.

[Andy Watson]

But I'd like to keep it red and green simply if possible.

Could you substitute Q3 with a P-channel mosfet?