[SOLVED] Looking for work-around to delay open circuit detector on power-up

[d123]

Hi,

This is a circuit in progress/in the design stage, so LED measured voltages are real, 18V supply is based on the last AC/DC transformer experience (18.7V out when loaded with a similar circuit/current load), voltage references and divided down reference of TL431 are also real and tried out; INA826 is based on calculations and simulations; the logic (CD4071 works on a breadboard, as does the simple regulator using BJT, op amp and Vref, but at 11.8V. Will check all this is still valid with a higher supply when get further with this design.

as the title states: looking for work-around to delay open circuit detector on power-up. I found the TI app note pdf at the bottom of this post, about using a Zener and resistor to keep the line functioning in the case of an open string/blown LED, but this is not something I am able to use at present, even 'though it's a great solution.

Description of problem:
The over-current part works (...using a trimpot to emulate INA826, admittedly), and doesn't trigger at power-up.

When I connect the open circuit sensing pin to the load resistor before power-up, as identified by the red circle and asterisk in the attached images, if I'm understanding what's happening (oscilloscope is basic and old so can't record any of this, and my eyes aren't as fast as a semiconductor...) it detects <~980mV and triggers the shutdown. All LEDs are off, and voltages into OR gate, and into PNP show this.

When I disconnect the open circuit sensing pin and leave it floating, the circuit powers up correctly. All LEDs are on, and as should happen, OR gate Out pin is low.

If I connect it to an intermediate LED, but not the transistor collector (as that also triggers at power-up), all LEDs are on, and as should happen, OR gate Out pin is low.

It takes about a second for the circuit to settle and it to be safe to connect the open circuit sensing pin/cable to a LED string current resistor (the Vref/Rsense resistor) without it triggering the shutdown function.

The open circuit detection seems to work, as when I pull an LED out with the circuit on, to emulate an open string/damaged LED, the shutdown function triggers.

I have tried RC power up reset type of solutions, which obviously aren't working, and have unwanted effects on the sensing resistor/required current, and went through some online POR circuits and app notes, but they were all for power, not signal sensing delay.

Does anyone know of a way to delay the open circuit detection that excludes power-on delay strategies, as using such would mess-up the over-current function, or is the method I'm attempting to employ a non-starter and not a feasible option? Thanks.

Here are the images of the circuit, and also the pdf if the picture quality in not good enough to see:

 

[BradtheRad]

It sounds as though you want the IC to test for no load and short circuit at startup. I suppose it can test for a short during the first cycle. But I think it cannot detect absence of load, until it performs a cycle or two and voltage soars high enough that it decides it has to shut down.

Come to think of it, you want it to continue to detect these conditions while running. Can you determine if it can do this?

When I disconnect the open circuit sensing pin and leave it floating, the circuit powers up correctly.

There may be some internal pullup/pulldown default value. Can you find out whether this pin produces a voltage on its own at startup? Could it be 0V?
Supply+?
Some value in between?

I have tried RC power up reset type of solutions, which obviously aren't working, and have unwanted effects on the sensing resistor/required current,

Are these the simple type of RC circuit that you've tried and didn't work?
One is for low output on startup. The other is high. (Timeframe = 0.4 sec) Did you try both?

 

[c_mitra]

I have tried RC power up reset type of solutions, which obviously aren't working, and have unwanted effects on the sensing resistor/required current, and went through some online POR circuits and app notes, but they were all for power, not signal sensing delay...

You need to use a FET to turn on the power to the ICs using a RC based delay as suggested by BradtheRad in the post above.

 

[d123]

Hi,

Thanks.

Brad, I tried the RC POR that is just 1 resistor to V+ and a capacitor to ground, the connection to the "over-performing" pin taken from the RC junction, so I'll have another go with the schematic you have posted for the "high on power-up then drops".

The IC in question is just an op amp, and the pin used to sense open circuit/no load is the inverting input. It's a quad package, over-current sensed by one op amp, and open circuit by another.

I think that I need/want to inject >1V ( as the non-inverting input has a reference voltage of ~980mV) to that inverting pin at power-up for less than a second until the sensing resistor develops 1.23V, which I am guessing is when the constant current stabilises its value.

Yes, from breadboard experimentation, it functions correctly and detects either an over-current event or a no load/open circuit event and shuts the pass transistor down, then returns to normal operation when either of these triggers/events is removed. ...So long as open circuit sensing pin is only connected after about a second (it must be much less than a second in reality) has elapsed after power-up.

I'm not sure this is a possible kind of "POR", as it's not delaying powering up an IC, it's tricking an input pin for a second and then having the >1V signal stop operating and not influence circuit operation. I tried to get an NPN driving a PNP sending a positive voltage into the inverting pin to go on at power-up then turn themselves off ASAP, but failed pathetically. As c_mitra says, I was thinking about it last night, maybe combining the RC POR and a PMOS to...turn itself off by using a capacitor from V+ to the gate so it momentarily is on, but the charged capacitor develops sufficient voltage to pull the gate to V+ might work? Probably not.

Thanks.

 

[c_mitra]

..turn itself off by using a capacitor from V+ to the gate so it momentarily is on, but the charged capacitor develops sufficient voltage to pull the gate to V+ might work? ....

I see. You then need to AND the Power on Reset with the open circuit detector signal. You still can use one of BradtheRad's signal.

 

[d123]

Hi,

Just as an end note, POR at first power-up with RC and PMOS works great, ...but it brings a previously not happening unwelcome side-effect, meaning not happening before trying to add open circuit detector part: it can't re-trigger after an over-current or open circuit event goes away, so it disables the main (i.e. the inner circuit of LEDs, after primitive voltage regulator) circuit re-start after either trigger event, doesn't matter which After trying quite a lot of ways of getting the open circuit op amp to unlatch it's output high without success - which needs the POR to happen first, it's back to the drawing board, I'm afraid.

Attached image shows/describes the flaw.

Many thanks for your help, guys.