1
\$\begingroup\$

I want to detect if the electronic locker in my differential is functioning. The differential locker is essentially a solenoid that engages the axle locker, however since the wire runs under the vehicle it is exposed, and a damaged wire would go undetected. I have no way of mechanically detecting it (limit switch, plunger, etc), so I need to do it electrically.

The locker's solenoid operates on 12V and draws between 5-8 amps, so assuming the higher value, the locker has an effective resistance of between 1.5 and 2.4 ohms.

I have found some 10A shunt resistors with a voltage drop of 75mV (I'm guessing that means 7.5mohm), which will have minimal impact on the operation of the solenoid. My calculation is that my actual voltage drop will be between 37.5mV and 60mV.

I'm not trying to measure the current, just detect it by lighting an LED (most likely a lighted switch which expects 12V. Most of the articles or information I've found reference OpAmps to measure current, but I just want to detect it, so presumably I would use a PNP transistor, but which one?

I have done some searching, but I don't even know if I'm searching for the right thing.

\$\endgroup\$
  • \$\begingroup\$ Are you wanting some kind of light bulb that lights up somewhere if the wire is damaged and the solenoid is no longer able to be activated? What are your goals here? What does "detecting it" mean, exactly? How do you get notified? Also, there are common ways that faults in the headlights are identified and reported in vehicles. Have you read about how these are achieved? \$\endgroup\$ – jonk Oct 18 '17 at 18:15
  • \$\begingroup\$ Sorry, I edited out the part where I wanted to light an LED. The goal is to light an LED when current is flowing (ie: the locker is activated) such that if the light doesn't come on, we know there's a severed connection somewhere. Edited the post. I haven't looked into headlight fault detection. Will research. \$\endgroup\$ – Prdufresne Oct 18 '17 at 18:18
  • \$\begingroup\$ So, is it correct to say that the light should always come on when there is 5-8 amps flowing in the solenoid and that the light should otherwise be off? \$\endgroup\$ – jonk Oct 18 '17 at 18:21
  • \$\begingroup\$ Correct. However, my goal isn't to do over-current detection. \$\endgroup\$ – Prdufresne Oct 18 '17 at 18:23
  • 1
    \$\begingroup\$ @Trevor I was trying not to assume anything here. (I've had cars with positive ground and this site covers the world.) But true enough. \$\endgroup\$ – jonk Oct 18 '17 at 18:41
5
\$\begingroup\$

Just to toss in another thought:

schematic

simulate this circuit – Schematic created using CircuitLab

None of the transistors need to be power BJTs; they can all be small TO-92 types (or smaller.) Optimally, it would be nice if they were matched. But I added \$R_4\$ as a potentiometer so you can make adjustments, instead. This also covers variations in your \$R_1\$ value. Nominally, it should be about 20-30% of the indicated value. But your mileage may vary depending on the matching of \$Q_1\$ and \$Q_2\$ and your actual \$R_1\$ value. (I thought about suggesting a BCM62 for matching but then there is \$R_1\$, so I think the potentiometer is useful until you know what you actually need here.) \$Q_3\$ is totally non-critical. I'm assuming there is a high-side switch here, so \$SW_1\$ is there to represent that detail. (Probably fused, too, but I didn't add that.) If you get BJTs that can stand off some decent voltage (like 2N5401), then there's probably no real need for load-dump protection.

I've not gone through this for all the practical details of an automotive situation. It's more a behavioral approach that avoids opamps. It might work okay, as is.


Just for grins, I tossed the above circuit into LTSpice and used the standard BJTs and \$R_4=22\:\textrm{k}\Omega\$ on a lark. The DC sweep came out this way:

enter image description here

You can play with either \$R_2\$ or \$R_4\$ to change the location where the switching takes place.

If \$Q_1\$ and \$Q_2\$ are matched, it won't even budge even if you vary \$\beta\$ and \$V_{BE}\$ over a reasonable range. Of course.. that's if they are matched. Of course it will work fine then. So if you can, pick \$Q_1\$ and \$Q_2\$ as matched pairs such as BCV62 or BCM62. They won't stand off a load dump well. But they might survive it.

Unmatched, and messing with \$\beta\$ and \$V_{BE}\$ (as a function of \$I_{SAT}\$), I get the following spread:

enter image description here

That's with a fair amount of variation on the BJTs. So it's not horrible, so long as you get the same part number. But it may require some tweaking of either \$R_2\$ and/or \$R_4\$ to make it work well for you.


I've added \$R_5\$ to prevent a mistake in setting the \$R_4\$ potentiometer to an accidentally harmful value. The value of \$R_5\$ should be the same as the value chosen for \$R_2\$, roughly speaking. I had imagined using a potentiometer until the right value was measured, then replacing it with an actual resistor. But if the potentiometer is kept, then the value of \$R_5\$ could be a little less than \$R_2\$ (as little as half) but probably shouldn't be more than 25% larger. Somewhere in that range should be okay if the same exact value isn't available.

I've added some possible protection schemes to the circuit. The diode-only would need to be something with perhaps \$200\:\textrm{V}\$ reverse voltage and at least \$10\:\textrm{A}\$ capability. The downside of it is that the flyback current dies out slowly. And in this case, it might die out way too slowly. The diode+zener combination provides a high voltage so the flyback current pulse can die out more quickly. But both parts must be capable of the same high current potentials and I'd want about \$30\:\textrm{V}\$ across the pair -- so select a zener in that area, if possible. Finally, there are also automotive MOVs (Littlefuse makes them, for example) which can repeatedly absorb solenoid flyback energy as well. They come in a variety of voltage ratings. Again, I'd be looking for about \$30\:\textrm{V}\$ here.

\$\endgroup\$
  • \$\begingroup\$ +1 nice.. my turn to be picky though ;).... Winding R4 to the wrong end = smoke... \$\endgroup\$ – Trevor_G Oct 18 '17 at 21:13
  • 1
    \$\begingroup\$ @Trevor Well, this whole thing was just tossed out as a random thought and more behavioral than real. Now we've got the suggestions for adding some fixed resistance to the pot and inductor flyback. I guess it's starting to get real. I will plug in a few things into the schematic (though I would not have minded you editing it.) \$\endgroup\$ – jonk Oct 19 '17 at 17:11
  • 1
    \$\begingroup\$ @Prdufresne Added some solenoid flyback protection thoughts. Just so you know. The diode by itself is probably my last choice because it might take seconds to do its job. The diode+zener or an automotive MOV of sufficient size might be the best choice here. Something that will yield about 30 V across the coil during flyback would be good. \$\endgroup\$ – jonk Oct 19 '17 at 18:34
  • 1
    \$\begingroup\$ :) Yes Jonk.. see what you started. But then again.. this is what makes trolling this forum fun. \$\endgroup\$ – Trevor_G Oct 19 '17 at 18:57
  • 2
    \$\begingroup\$ @Trevor I'd wanted to toss out a thought that would be junk box parts and accommodate discrete BJT variations using a pot. Then someone (who shall remain nameless) mentioned matched BJT pairs. Which is of course better, though boutique and less likely as junk box parts, and might mean that a design without a pot could be achieved. So I had to go Spice the darned thing to add more. Then someone (still nameless) mentioned the need for flyback protection -- good point by the way. Still more. So... maybe I toss these out as a minor comment to your answer next time... ;) \$\endgroup\$ – jonk Oct 19 '17 at 19:08
2
\$\begingroup\$

Since this is automotive I have to assume the wire is on the +12V side. As such you need to use high side current sensing. There are numerous devices that do this like the LTC6101.

enter image description here

Whichever sense method you chose, you would then need to feed the analog value into an appropriate comparator circuit and have it turn on an LED when current is detected.

Like this perhaps.

schematic

simulate this circuit – Schematic created using CircuitLab

Choose values for R2 and R3 to set the threshold voltage to turn on the light when the current is say over 1A.

NOTE: Since this circuit takes it's power from the solenoids power line it consumes zero power when that is turned off.

ADDITION:

You have to assume the car manufacturer did not add a fly-back diode at the switch or the solenoid, so when it goes off the top of the circuit will spike up large. As such I added a 10A fly-back diode. Note I chose a ZENER to also protect against large positive voltage spikes from the car's environment. The Zener part does not need to be 10A though so two diodes, high current rectifier and low current Zener, may be cheaper.

\$\endgroup\$
  • \$\begingroup\$ I'm mostly worried about directly exposing opamps to a vehicle's power system. May want to add a zener (or something) for some kind of load dump protection (excess of 100 V and gnarly when it happens.) \$\endgroup\$ – jonk Oct 18 '17 at 18:39
  • \$\begingroup\$ @jonk yes perhaps a zener limiter would be prudent. I don't think the diode will care about the noise much. Decouplers are probably even overkill with this LOL. \$\endgroup\$ – Trevor_G Oct 18 '17 at 18:42
  • \$\begingroup\$ I tried simulating the circuit you shared above with a Load to the left of Rsense and a 12V battery to the right, but I couldn't generate a voltage differential across the LED. I'm sure I'm doing something wrong in selecting my R2 and R3 values. I set Rsense to 7.5mohms and my load to 2.4ohms. \$\endgroup\$ – Prdufresne Oct 18 '17 at 20:56
  • \$\begingroup\$ @Prdufresne unless you have the simulation model for the LTC part you wont be able to. It is just drawn as an op-amp because they did, and it has the right connections.. it is more complicated than that internally \$\endgroup\$ – Trevor_G Oct 18 '17 at 20:59

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.