Sorry...can't think of a succinct question for the subject line!

I am working on a project on my car. The car's existing ECU is connected to a pressure switch, which opens when the A/C refrigerant pressure is low. This causes the ECU to switch off the A/C (via a different circuit). By opening this circuit myself, I will be able to switch off the A/C using a logic output from an MCU. The attached picture shown an NPN transistor in the place that I intend to insert...something. My testing shows that opening the circuit works as intended, and that a simple NPN does the job.

ECU circuit

I want to ensure the component is immune to or protected from the dreaded load dumps and other transients/events I know to expect in the environment. However, I don't want to use a relay, as inside the cabin the clicking will be annoying. My MCU is running off an automotive LDO voltage regulator (LM2937), which is protected from overvoltage, reverse polarity, load dumps, etc.

I have two questions:

1) In this specific case...do I have anything to worry about anyway? I pretty much assume the ECU's circuitry should be insulating this line from these events. Maybe the NPN I'm using for testing will be good enough - perhaps with a big diode in reverse?

2) If I DO need to worry...what would be the best solution?

  • \$\begingroup\$ How much voltage and current are we talking about here? In other words, what voltage do you measure when the circuit is open, and how much current flows when it is closed? \$\endgroup\$
    – Dave Tweed
    Feb 21, 2014 at 0:44
  • \$\begingroup\$ It's at 12V when the switch is open. The current is <1mA. It's a CPU input connected through a series of switches (temp, pressure, and on/off switch) to ground - in the diagram above they are simplified as one switch. I just want to insert another logic-operated switch. Another unconnected CPU output switches the various relays to turn on the AC. \$\endgroup\$
    – Jon
    Feb 21, 2014 at 1:16
  • \$\begingroup\$ May be of interest, although in theory you're isolated from the worst: littelfuse.com/data/en/Application_Notes/an9312.pdf \$\endgroup\$
    – John U
    Feb 21, 2014 at 10:05
  • \$\begingroup\$ Thanks for the useful doc, will have a look through - but sounds as though you're saying a simple transistor would probably be sufficient anyway? \$\endgroup\$
    – Jon
    Feb 21, 2014 at 11:44

2 Answers 2


You can't go far wrong with an opto-isolator or opto-MOS relay.

At < 1mA, something cheap like this one.

enter image description here

  • \$\begingroup\$ I have a few small ones in the parts box...I see how it would protect my circuit from the effects, but would the component itself survive an event? \$\endgroup\$
    – Jon
    Feb 21, 2014 at 11:45
  • \$\begingroup\$ The ECU will have its own protection, so no load dump will be coming out of the ECU. If your circuit cannot feed more than 10's of mA to the LED, it won't be damaged no matter what voltage is between the two points (due to a bad ground or whatever). Note the one I suggested is a transistor (not darlington) and has a high CTR @1mA, \$\endgroup\$ Feb 21, 2014 at 13:07

Since you say that your added switch will be in series with several existing switches on an input to the ECU, I suggest you use a small reed relay - they are quite quiet, and inexpensive, and will not introduce any unexpected voltage drop.


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.