1
\$\begingroup\$

I'm building a device that regulates the atmospheric composition of a space using a gas sensor and a control loop to actuate a solenoid valve to release compressed gas into the space. Due to the nature of this task, it should be clear that such a device could easily create a serious safety hazard if it malfunctions.

Regardless of how reliable or fail-safe a device like this is designed to be, any install of such a device must of course be accompanied by the install of an external gas monitor and evacuation alarm system. However, I don't want to rely on that as the only safety measure. I'm planning for an independent watchdog MCU from the main control MCU for to exclusively perform fault detection, and it's not hard to add an additional interlock relay to let the watchdog MCU cut power when it's in a resettable "soft" fault state.

However, for other fault types (i.e. those that could indicate damage to the device, or where one more fault could allow the device to start reading itself as 'OK' again while operating unsafely), I need a way for the system to cut off more permanently, such that the device will need to be serviced and re-inspected in order to function normally again.

I'd also rather have this happen in a way that's more robust beyond just writing some bits into nonvolatile memory, and put it completely beyond the scope of a mere software glitch or electrical failure to re-activate things.

Right now, the best I've been able to come up with is a crowbar circuit for letting the watchdog MCU intentionally blow the solenoid valve fuse:

schematic

simulate this circuit – Schematic created using CircuitLab

I'd have to imagine there's already some kind of device available to do what I want more directly though, and that doesn't depend on having the 24VAC input energized at the point the fault occurs.

What type of device should I be looking at to perform this function instead?


Here's another version that's probably better, fusing the input coil to a single relay and blowing the fuse with the controller's power supply rather than the solenoid valve's power supply; that way the gas controller can present just a single relay output for driving whatever solenoid from whatever power supply.

schematic

simulate this circuit

\$\endgroup\$
11
  • 2
    \$\begingroup\$ You are describing a crowbar circuit protection scheme. See the link for the etymology of the term. \$\endgroup\$
    – Transistor
    Jul 12 at 17:25
  • \$\begingroup\$ @Transistor Ok, that makes me feel a bit better about the idea of blowing the fuse intentionally like this. Is there a device or method for implementing it that would allow the watchdog MCU blow the fuse even when the 24 VAC input isn't energized? Since in an actual implementation, ideally I'd like this device to only need to provide a single set of relay contact terminals that can be wired with any external power supply and external solenoid required. \$\endgroup\$ Jul 12 at 17:32
  • \$\begingroup\$ @Transistor Thanks for the etymology link, I never knew that. I always thought it just referred to a brute-force method of opening the fuse as if by a crowbar. \$\endgroup\$
    – John D
    Jul 12 at 17:56
  • 2
    \$\begingroup\$ If you do use a crowbar circuit, make sure that your supply will actually blow the fuse under all conditions. We once owned a piece of test equipment from a well known manufacturer that had a power supply crowbar. One day it tripped, and the supply impedance was so high that the fuse didn't blow, and the next we knew the lab was full of smoke from a cooked mains transformer. \$\endgroup\$
    – Neil_UK
    Jul 12 at 18:20
  • 1
    \$\begingroup\$ Yes, a crowbar circuit will work -- but you probably really want a latching relay that's reset with a pushbutton that you supply, or an NVR switch. See this post about latching relays. \$\endgroup\$
    – TimWescott
    Jul 12 at 18:55
3
\$\begingroup\$

A fuse will probably work, but that's not how it is done in industry. For starters, you have to deal with replacing a fuse each time there is a fault. I worked on steel annealing furnaces where the possibility of explosive gases building up and exploding could potentially happen if it weren't for all the safeties being designed into the system. I can tell you some of the ways they did it, and some of the ideas can be incorporated into your design. First, they used a Maxon valve, which is basically a valve that must be opened manually with an actuator arm and then is held in the open position until there is an emergency condition or power loss. This gets a human involved to see what happened. Maxon is the trade name, but basically it is manual-open, power-loss on the coil to close. The coil must already be energized to allow the arm the valve to open. You don't need to blow a fuse for a safe shutdown, just de-energize the holding coil.

Secondly, there was a "valve integrity check". At the start of a cycle, the system would close two valves that were facing the furnace, and then open a third valve that would pressurize a section of pipe with natural gas. The feed valve would close and then a pressure sensor would watch the gas trapped in the pipe. If the pressure held, then the integrity was proven. If it decreased, as would happen during a leak, then the system would indicate a fault and require a reset. You might consider this method of making the system safer.

There were other checks as well, such as presence of oxygen when there should/shouldn't be. I don't know if this would be applied to your system.

Using electrical contractors that latch electrically for the purpose of logic was highly frowned upon, as how I was told by the instructors in the classroom and should be designed out. The reason for this is because the system is in an unknown state at start-up. I want to be clear I am not talking about a latch that resets upon loss of power, but the kind of latch that holds its state upon power loss.

I don't know all the details of what you need, so have to end it here, but it sounds like you will also need a pressure release valve.

\$\endgroup\$
1
\$\begingroup\$

I would design a current limited shunt that would trip a breaker with an indicator or alarm signal in worst case acceptable time but would not damage or overstress any components. A Triac Crowbar can be triggered easily at zero crossing with with optical isolation and maintain power to the control system but guarantee interrupted power to the compressed valve solenoid until serviced. This can be easily tested under simulated conditions to see if the result causes any unintended EMI to other systems.

\$\endgroup\$
-1
\$\begingroup\$

Fault sensitive equipment should not use OPEN LOOP stop circuits. In case of wire failure your circuit will not stop the power supply. Use CLOSE loop stop which incorporates UVT equipped circuit breaker. UVT= Under Voltage Trip. https://www.relectric.com/blog/2017/10/27/circuit-breaker-under-voltage/#:~:text=An%20under%20voltage%20release%20is,breaker%20does%20not%20power%20it.

So basically you provide power to your critical circuit via circuit breaker which has an UVT. The power to the UVT coil is provided by the safety circuit. If the safety loop is broken the CB will trip. Any fault in safety loop will cause the CB to trip, wire breakage, power lost to circuit, or actual fault. Making this design very reliable and robust even for highest of standards of safety. As its function does not rely on power supply being present.

\$\endgroup\$

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.