I have a digital signal called FAULT that is active-low. It will be high 99% of the time and low in rare cases.

I'm trying to use a latch so that the majority of the time the output of my latch will be low but when FAULT goes low, I would like my latch to output high AND stay there even if FAULT goes back high. The only time I want my latch to reset is on a power cycle.

In a nutshell - "Output of my circuit should be low until FAULT goes active (low) at which point output goes high and stays there no matter what until a power cycle"

I came up with this initially (I know it's wrong)

enter image description here

The only issue with this is that it assumes that the initial state would be zero and I can't make that assumption since initial state of a D latch is indeterminate.

I was thinking about using an SR latch instead but again I would need to send a reset signal initially and I don't know what's the industry standard way of sending an initial reset in a circuit. Do any ICs exist that will auto reset my latch on start up?

How should I approach this so that my circuit practically realizable in the simplest fashion possible.

  • \$\begingroup\$ D FF Rising edge input to rising edge output. ( any pulse width ) Power up reset with a cap to V+ and R to gnd. Will you want to control reset at any time? You can also use two NAND gates as SR latch with active low input and high output. \$\endgroup\$ – Tony Stewart Sunnyskyguy EE75 Jan 7 at 4:45
  • \$\begingroup\$ @Tony EE rocketscientist All I want is it to reset on a power cycle. Ah! So you're saying that I use a capacitor to give an initial pulse on start up with this circuit. Won't the resistor to ground just be a pull-down resistor and keep my reset at zero? Could you just clarify a bit more on that? \$\endgroup\$ – BJTsAreMyFriend Jan 7 at 5:01
  • \$\begingroup\$ THe RC product in ms ensures the RESET stays high for that duration. e.g. 0.1us 100k to 0V is T=10ms=0.01s \$\endgroup\$ – Tony Stewart Sunnyskyguy EE75 Jan 7 at 5:02
  • \$\begingroup\$ @TonyEErocketscientist Awesome. I'll look into that. First time I've heard of power-on reset. Thanks a lot. \$\endgroup\$ – BJTsAreMyFriend Jan 7 at 5:07

BJT, My schematic is a bit disjointed due to my unfamiliarity with the drawing tool. Here are the basics:

A. The flip-flop is a 74HCT74 or 74HC74 D-Type Flop.

B. The diode is any signal type diode similar to a 1n4148.

C. Capacitor can be tantalum, ceramic, electrolytic.

D Resistor is 1/4 watt or smaller.

E. Supply voltage is 3.3. or 5, it depends on which family of 74 logic you use for the flip flop.

F. The "1" on the D input to the flip flop denotes a pull-up (Logic 1) to the supply voltage. A 1K resistor or direct connection.

G. The clock input of the D-flop is connected to your FAULT signal. Which I assume to be high-active.

H. The RESET- signal is connected to the RESET-input of the D-flop (74HC74, etc.

I. The 74x74 also has a SET- input, you should connect this to a Logic 1 (just like the D input described above).

J. The 74x74 IC has two sections. If you are only using one section, all of the inputs on the unused section must be connected to circuit ground (DC common). Just the inputs, not the outputs! Leave these unconnected.

The D,R & C form a rudimentary power on reset circuit. You can find an explanation of how this operates on the web. The essential operation is that the RESET- node is held low when power is first applied and eventually (i.e. within a fraction of a second) rises to a Logic 1 level due to the time delay of the R & C. The effect is to hold the flop in Reset while power is applied. Once it is reset like this it will stay reset until the FAULT signal goes active momentarily, which will SET the flop where it will remain until the next power cycle (OFF then ON).


simulate this circuit – Schematic created using CircuitLab

This very popular power-on reset circuit has a number of issues. A few are listed here:

A. The power off time between power down and repowering up can be finicky and somewhat erratic. Meaning you need to ensure a long enough power off time between repowerings for it to act reliably. Say, a few seconds. You'll experiment and find out.

B. The behavior of your FAULT signal during power-up can thwart the operation of this circuit. You must make sure the FAULT signal is indeed inactive (assumed to be LOW) while the power is applied.

C. The selection of the R & C can improve performance. I suggest making the value as large as practical to get as long a power on delay as practical for your application. However, longer power-up delay (Reset pulse generation) also produces a longer turn-off time requirement. Don't make the Resistor much bigger than 47K, the cap can go to 100 uF, depending on the type of cap, ceramic and tantalum a re best for this application. Electrolytics will work as long as capacitance is "not too large" - meaning something like 100 MFD.

D. The "profile" of the power/voltage application tot he circuit can cause erratic operation of this circuit. For example, whether the power "snaps" on very quickly. Or, it ramps up relatively slowly. E.g. power is applied from a battery thru a mechanical switch. This is a snap-on case. In contrast to power is supplied thru a solar cell, where the voltage rises very slowly into the operating range of the circuitry. The latter is a problem for this reset circuit.

The most common wiring error for beginners with this circuit is getting the polarity of the capacitor and diode correct. Double check these when you assemble the circuit.

You can observe operation of this circuit by monitoring the RESET signal with an O-scope or analog voltmeter (needle type). A DVM might work, but the readings may change to quickly to follow visually. Remember the RESET signal will start at GND and rise towards your supply voltage as power is supplied to your complete circuit.

The alternative is to use a dedicated power-on reset IC such as those made by Maxim & Dallas. However, these require a certain level of engineering expertise to apply properly to your circuitry.

Good Luck!


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.