Hello from USA - I need some advice to detect loss of heartbeat signal in safety of life application

[uWave.com]

Hello from California!  I'm a microwave/RF engineer by trade, I've installed Ademco systems for myself and for friends, as well as the 900MHz (USA) Alarmnet digital polled radio backup to dialup or multiplexed leased lines.  Alarms are NOT part of my job, but from what I learned programming the Ademco equipment, I never saw a configuration for a heartbeat pulse signal monitor from other 3rd party equipment.

 

I have a dilemma on how to accomplish this - my customer has 3 electrical devices that by law here must be monitored for failure of the safety device itself.  The device does this by providing a roughly 1 blink per second green LED to communicate its status.  If the device were to "seize up", it would put human life at risk, thus the need to make sure it's functioning, and the green light is blinking just like it always is.  I can replace the LED with an optocoupler and get the emitter&collector of a phototransistor output.  It can be made to resemble a contact closure to a panel zone, but it's normal (i.e. no alarm) state is opening/closing about 1x per second (this can be stretched longer in cases of maintenance, like a firmware update that may take 30 seconds - I'd set the alarm condition to be zero blinks in 60 seconds)

 

The problem is most low-end alarm panels see this open/shut/open/shut behavior (heartbeat led) as a "swinger" or a door that isn't shut right and keeps re-triggering that zone in the wind.  

 

Does anyone know of a panel with a zone configuration specific to a pulsed on/off signal representing a "closed" state (like a door) and the lack of a signal, or a stuck on/stuck off representing the Alarm condition?  If not a panel, has anyone seen a pre-conditioner that would take the heartbeat blink and convert it to steady closed / steady open on alarm?  I'm trying to find a commercial off the shelf solution rather than a custom design just to monitor a blinking LED / transistor logic output on 3 systems.  Thanks for your thoughts here!

 

Fred

 

[MrHappy]

unless the items have a volt free relay which changes state when they go into fault....

 

I think your wasting your time ?

Edited October 17, 2023 by MrHappy

[al-yeti]

1 hour ago, MrHappy said:

unless the items have a volt free relay which changes state when they go into fault....

 

I think your wasting your time ?

Why? He could hook it up a panel with inputs and programmable outputs 

 

?

[sixwheeledbeast]

Can't rely on automatic detection of a flashing LED for a safety system, as above if the equipment doesn't have a fault interface your peeing in the wind.

 

[james.wilson]

Agreed does it not have an output. What is the device you are trying to monitor? 

[uWave.com]

Thanks for trying to lend a hand; the more people the better!  The flashing LED *will be replaced (by me)* with an optocoupler, which is an 8 pin part that looks like an LED to the device (input 2 pins), but provides an open collector phototransistor output (short to ground opposite 2 pins) on the alarm device providing optical isolation between the two.  The LED is fully encapsulated and won't be visible once I change this one part - if a dry contact closure is needed, I can drive a relay with the optocoupler phototransistor and it will click on/off once per second, though the wear&tear on the relay is undesirable, so I'd much prefer to keep it all silicon devices.

 

The people who designed the safety device put the blinking heartbeat LED in so that *IF* it's blinking, it's working.  It's an output line from the microprocessor inside meant to be human readable, but due to this application, it needs to be monitored and supervised, so my optocoupler circuit has a built in periodic self-test and will signal a trouble line if anything else goes wrong.  That part is easy to do in hardware with a soldering iron.

 

Now the problem I have is a contact closure who's "normal" state is opening/closing 1x/sec and "alarm" state is stuck open or stuck closed.  There could be a hardware failure, or even a software failure causing it to freeze up.   Both of those 2 conditions mean something died, and it needs attention.  That's the part I'm having trouble figuring out how to program a panel to connect to.

 

My more hardware centric idea was to take a capacitor, use a large value (1M-ohm) resistor to keep it charged slowly and use the periodic signal to short it to ground, along with another capacitor that is charged (tied to 5v) by the periodic signal with a "bleeder" resistor to 0V.  Alarm condition would be either of these 2 capacitors reaching their level set by the always on resistor (high or low) because it can get "stuck" with the LED on or off, so I have to handle both cases.  

 

Lastly, as this is a safety device, it has its own ALARM output dry contact, but the designer did not put in any sort of "trouble" output, just the blinking LED - without that a human must make rounds looking for blinking lights.

 

thanks for your help!  I really appreciate it.   Fred

[sixwheeledbeast]

As we say above, if a blinking LED is the only think provided by the manufacturer this should only be used as a visual human indicator not to control something.

We do understand your problem and while it can be done with engineering something; either from discretes, semis and logic or even software on a SBC, we are looking at a different level.

Being involved in safety and security systems we are looking at the risk of something like this, maybe this isn't how it works over the pond but modifying kit like this would put the liability on you if it failed.

If the manufacturers design it to have a failure output to use then this would be tested to the standards required and if that failed you would take it up with them.

Your kit wouldn't be tested to the same standards and while you are understandably trying to keep your kit electrically isolated from the manufacturers it would still be classed as a modification removing any conformity the equipment has.

 

Over here it happens, companies will build projects using all parts that are 'xy' rated believing the end product they supply has that conformity. It's not how it works all the components may have never been tested with each other before.

For things that are not critical maybe this is never discovered, but if something fails in a critical system questions will be asked, most likely in court.

Even something as simple as mixing manufacturers of electrical breakers within a consumer unit, if a fire occurs this maybe to blame. Manufacturers are not going to be testing conformity of there kit with every other manufacturer just to make sure you can use what you have on the shelf...

[james.wilson]

Not knowing the application 'if' I just wanted to monitor an output oscillating I'd either use 2 trans relays with timers or esp32's

 

However if I was life I wouldn't I'd want an output from the device because if I modified it and something happened I dont think I like prison