Timing Circuit? Counter?

How accurate do you want it?
There are simple methods you could employ with RC circuits. (Resistor Capacitor)
You could use a 555 Timer, or other ICs.

 

haha... well a 555 is a timer, and a very famous one at that!
I will have to think something up. Have you played with transistors yet?

 

Hi Ernest and welcome to Electronics Point

You've listed four different voltages there, but there only seems to be one decision threshold - somewhere between 1.3V and 3.6V. If it's greater than that threshold, the voltage is OK, and if it's less, the voltage is out of specification, and after a three second delay you want an output to go low and latch. Is that right? A voltage of around 2.5V would be about half way between those voltages, so would that be an appropriate threshold?

What power source is available for the circuit? Do you need it to be powered from the input source? (From a battery, I assume, based on the voltages you specified.) Or do you have another power source that is guaranteed to be at least 2~3V? Because circuitry that operates from less than 2V needs some special design considerations. What would the minimum available supply voltage be?

 

Yes, it can definitely be done. I'm just considering the most appropriate way to do it.

Thanks for the clear definition of behaviour. I had realised that you want the three second timer to reset if the voltage goes back over the threshold, but it's good to see it spelled out like that.

How accurate does the threshold voltage need to be? For example, would "somewhere between 1.4V and 1.8V" be accurate enough?

Is the input current drain important? If it's being connected permanently to a battery-powered supply, you might want to specify the maximum allowable load current, so it doesn't discharge the battery unnecessarily. I guess if it draws less than around 100 µA this will be low enough that you don't need to worry about it.

What does the output need to switch? Is it just driving an LED, or is there something else?

You want the circuit to unlatch when it's power-cycled? I guess you mean when the power supply to the circuit is disconnected, not when the monitored voltage is disconnected?

What is your priority: minimum component cost? Minimum component count? Minimum PCB space? If PCB space is critical, can you deal with SMT (surface mount technology) components?

 

So there is actually another threshold somewhere between 0.7V and 0V. The timer should be enabled only when the voltage is between 0.7V and 1.3V. Less than 0.7V or more than 1.3V should disable it. Right? And you want both of those voltages to be fairly accurate? Would 0.55~0.75V and 1.2~1.5V be accurate enough?

 

OK I'm working on a design but it won't be ready until tomorrow. In the meantime maybe Gryd3 would like to suggest something.

 

These are the two building blocks I'm thinking of using for a suggestion, but I am unsure how I would 'cancel' the timer once started. I'm thinking my method would simply 'pause' the timer and I don't like that

The window comparator would saturate a transistor that will provide current to the capacitor connected to the 555. The RC delay would be set to at least 3 seconds, but if the reference voltage goes back to normal, the capacitor will still maintain a partial charge... so the next time the timer should start, there may only be 1-2 seconds until the event is triggered. I have not drawn it from scratch because I have not thought of a way to drain the capacitor in a nice timely fashion to reset the 3 second delay.

 

OK, here's what I've come up with.



This circuit must be powered from a clean, regulated 12V supply. It has an open collector output that goes low and latches low when the input voltage has remained between the two thresholds (1.3V and 0.7V) continuously for about 3.5 seconds.

It is based around an LM339 quad comparator, U1, which is an old device but still widely used. The LM339 contains four identical voltage comparators, named U1A~U1D, which are represented as triangles on the schematic diagram, each with two inputs and one output.

Each comparator drives its output low if the voltage on its inverting ("-") input is higher than the voltage on its non-inverting ("+") input. If the "+" input voltage is higher than the "-" input voltage, the output floats and can be pulled up to a positive voltage.

This type of output is called "open collector". In the low state, it is able to pull down to 0V quite strongly, but in the high state, it does not pull up. External components must be used to pull the output up towards a positive voltage.

The input voltage is monitored by U1D and U1C, using two different voltage thresholds set up by the four-resistor voltage divider RA~RD. The voltages created by this voltage divider are marked on the diagram; these voltages are only accurate to within a few percent.

If the input voltage is higher than 1.3V, U1D will pull its output down to 0V. If the input voltage is lower than 0.7V, U1C will pull its output down to 0V. In either case, their commoned ouputs will pull to 0V and this will hold CT discharged, because QL is (initially at least) conducting like a short circuit.

Only if the input voltage is between those two voltage thresholds will both outputs go into the floating state, and allow CT to charge up through RT.

This configuration is called a "window comparator" because it compares the input voltage to a "window" between 0.7V and 1.3V. If the input voltage is outside that window, the commoned outputs pull low; if the input voltage is within the window, the combined output signal becomes "open" and no longer pulls down to 0V.

While the input voltage remains within the window, CT charges up through RT. If this condition remains true continuously, after about 3.5 seconds the voltage on CT will reach the 9.6V threshold voltage on pin 5. When this happens, U1A will pull its output to 0V.

When U1A's output goes low, QL turns OFF, and becomes an open circuit, and the commoned outputs of U1D and U1C can no longer discharge CT, even if the input voltage goes outside the input window.

Also when U1A's output goes low, U1B's "+" input goes lower than the fixed 9.6V threshold on its "-" input, so its output goes low as well. This output is designed to sink about 5 mA maximum.

Once the output goes low, it latches into that state. This is done by QL, a small-signal N-channel MOSFET, which is controlled by its gate, driven by the output of U1A. While the gate is high, i.e. +12V, QL will conduct when the window comparator's output goes low, and this will discharge CT. Once U1A's output has gone low, however, QL has no gate bias and does not conduct at all, so the window comparator cannot discharge CT. CT charges up to +12V and the circuit's output stays low until power is cycled.

I'm not too happy with this arrangement. If anyone can suggest a more reliable way, that doesn't require too many components, please do.

When power is removed and the +12V rail goes to 0V, DD conducts and discharges CT ready for the next run.

U1 is powered through pins 3 and 12 which are shown attached to U1A. CD is a 100 nF (also called 0.1 µF) ceramic capacitor called a "decoupling capacitor" and it is needed to ensure reliable operation of U1. It must be connected as closely and directly as possible between pins 3 and 12 of the IC - tacked on underneath the board is good.

DIN and RIN provide some protection against excessive voltage at the input.

CT needs to be a good quality low-leakage capacitor. Electrolytics might be suitable but I recommend a film capacitor such as http://www.digikey.com/product-detail/en/R60DI4470AA30J/399-5908-ND/2571343

The delay time is determined by RT and CT. The formula is roughly
t = 1.6 × RT × CT
where t is the delay in seconds,
RT is in ohms,
and CT is in farads.

With the values given:
t = 1.6 × 470,000 × 4.7×10^-6
= 3.5 seconds.

 

Ra, Rb, Rc, Rd create a voltage divider.. if you change the 12V to 14.5V then the device will trigger at a different reference point. So instead of triggering between 0.7 and 1.3V it could end up shifting to 0.9 and 1.7V. That is the first thing I notice, perhaps Kris has another reason.

 

Sorry about the slow reply. Real life rudely intruded into my little online bubble

As Gryd3 said, the input voltage thresholds are set based on a 12V power supply voltage using RA~RD. Also the 9V6 threshold voltage affects the timing; it should be 80% of the supply voltage, whatever that is. I could change the circuit to use a regulator to set the thresholds; this might make it easier to deal with the latching logic as well. I'll look into this option. If you're buying components from Digi-Key, add a TL431C (http://www.digikey.com/product-detail/en/TL431CLPM/296-26717-1-ND/2256072 for the through-hole version or http://www.digikey.com/product-detail/en/TL431AIDBVR/296-15237-1-ND/566512 for the SMT version) to your list. But there will also be some new resistor values.

Those all look good. You only need either a BS107 or a 2N7000, not both. There's no preference so you might as well get the 2N7000; it's about 20 cents cheaper!

Regular 1/4 watt metal film 1%. These are better quality than carbon and the price increment is negligible.

My pleasure

I will see if I can improve the design to be less fussy about the power supply voltage, and to latch more reliably.