0
\$\begingroup\$

I'm building a device that uses a microcontroller to control 15 solenoids via a preexisting multichannel DC solid state relay board. When all is working as intended, at most one solenoid will be actuated at a time, and each actuation consists of a 100 millisecond 24V/3.5A pulse.

I'm looking for a way to insure that a solenoid won't be overheated even if there's a software fault that causes an output pin to remain high, and consequently cause the SSR to remain open, driving the solenoid.

So my question is this: is using a a slow-blow fuse in series with the shared 24V supply a prudent way to protect against burning up solenoids?

If so, what specs should I use for the fuse? Should I just take the the I^2 * T value (1.2) and double it?

If there's a better way to do this, I'd love to hear it.

\$\endgroup\$
4
  • \$\begingroup\$ How much can the solenoids take before damage occurs (assume worst-case, they're hot, the 24V is at the high edge of tolerance, it's a particularly hot environment and the software freezes..)? Fuses can die over time from being hit repeatedly so caution is called for if you're calling it close. \$\endgroup\$
    – Spehro 'speff' Pefhany
    Commented Aug 20, 2018 at 4:51
  • \$\begingroup\$ That's a good question, but one I'm not sure I can easily answer. The solenoids are in old, hard-to-find, pinball "score reels", so I'd rather not test to failure. I have already burnt one up, but I don't know how long the fault was present for. \$\endgroup\$
    – Fustercluck
    Commented Aug 20, 2018 at 4:54
  • 1
    \$\begingroup\$ If you have no idea, I would suggest looking at what original fuse was used to protect the solenoids. The hairy-eared engineers of old would have done their homework. There are ways to test a coil but it's involved and still requires some guesswork. You could add a 74HC123 and do the 100ms with hardware but that wouldn't protect against a bad driver or other hardware issue. \$\endgroup\$
    – Spehro 'speff' Pefhany
    Commented Aug 20, 2018 at 5:16
  • \$\begingroup\$ You might be able to simply put a 470 - 2200uF uF cap in series with the score reel solenoids. \$\endgroup\$
    – Henry Crun
    Commented Aug 20, 2018 at 5:50

1 Answer 1

Reset to default
2
\$\begingroup\$

Drive solenoids through a hardware monostable, so that they simply go off.

It can be as simple as RC coupling the drive fet.

schematic

simulate this circuit – Schematic created using CircuitLab

An other approach is to sense the current drain into all solenoids, and if it stays on for too long, force a hardware reset on the CPU, thus (hopefully) releasing the drive.

schematic

simulate this circuit

The reason for this circuit, was that it was an easy retro-fit to impact printers that would burn out the printheads when the cpu locked up. This circuit was added in series with the common wire, and connected to the reset pin.

\$\endgroup\$
4
  • \$\begingroup\$ Thanks Henry. I was wondering about coupling the inputs capactively. Is the intention above that all the components above would be connected to the input of the SSR, or does M1 represent the FET at the input of the SSR? \$\endgroup\$
    – Fustercluck
    Commented Aug 20, 2018 at 4:58
  • \$\begingroup\$ All depends on time constant and currents you need to get. If you have to put 10mA into an opto-ssr, then a direct C isn't going to do it. \$\endgroup\$
    – Henry Crun
    Commented Aug 20, 2018 at 5:05
  • \$\begingroup\$ Gotcha, I guess I would have to measure the input impedance, since the SSR board doesn't specify that. Any thoughts on the fuse? Since I only drive one channel at a time, I thought that might be simpler. \$\endgroup\$
    – Fustercluck
    Commented Aug 20, 2018 at 5:09
  • 1
    \$\begingroup\$ A fuse is not right. They do not have a specified designable time. A PTC thermal fuse, in the solenoid coil is a viable choice. Personally - capacitive coupling/monostable is the correct choice. If what you need is a timeout, design a timeout. \$\endgroup\$
    – Henry Crun
    Commented Aug 20, 2018 at 5:14

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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