Did I Get Punked?

When I left work yesterday (Los Angeles) it was around 90 F outside and my car had been parked all day in direct sunlight.

As soon as I started my car (2019 Mazda CX-3), I noticed that my OAT indicator was reading 47 F.  Rats, either the sensor has failed or a wire has broken.

But as I began driving home, the indicated temperature reading crept up in a linear fashion and after about 20 minutes, it was giving me a very plausible reading.

This morning, the OAT indicator worked perfectly from the start of my drive to work.

Did somebody stick ice on the probe yesterday to punk me???

I'm teaching statistics this week to a bunch of young engineers so maybe but this seems pretty pointless.

Thoughts

My battery blew off the chart and when I started it later it was all good.  
 

In reply to RX Reven' :

Ghost in the machine.

It has an upper limit of 100 degrees, so sitting in the hot sun, it was actually 147 degrees so it rolled over and showed 47?  The creep up was it cooling off until it was back under 100?

Total wild ass guess.

-Rob

Datsun310Guy said:

 

1.21 gigawatts.  Good thing you didn't hit 88mph.

In reply to rob_lewis :

Thanks for taking a guess Rob but I've seen triple digit readings on a few occasions.

Apparently, the probe is located in front of the radiator and my grill spacing is pretty wide to prankster is still a viable theory.

I know, I know, but starting at 47 F (not 000 or 999) and gaining a nearly perfectly continuous two degrees per minute for 20 minutes just feels like "piece of ice".

I'm going to check for evidence of recent water. 

Theory: The OAT sensor display is pretty heavily damped. I noticed this on my old Subaru. They seem to sample the temperature relatively infrequently (like, once every few minutes) and display an average of the last several (ten or so?) samples. Maybe it's a weighted average, where the most recent sample has the most weight. Via some fault, maybe a loose wire for the sensor, maybe an electrical glitch, maybe a software bug, the temperature sample that the computer for the OAT display received when you started the car was some bogus low number, maybe 0, maybe as far negative as the sensor can read, but impossibly low. That bad sample skewed your weighted average down to 47F. Over the next 20 minutes, the computer got several new, accurate temperature samples from the sensor, which pushed the bogus low sample farther out in the weighted average, thus the slow climb back to an accurate temperature that you observed.

In reply to obsolete :

Highly...

Could something have parked near casting a shadow to somehow throw it off?

Did the ghost in the machine cause it to temporarily read in C? 47C is approximately 117F.

Rons said:

Did the ghost in the machine cause it to temporarily read in C? 47C is approximately 117F.

This was my initial guess. Depending on where it is in relation to the engine and the sun, it could easily read that high. EXCEPT: the temp linearly climbed to something reasonable. So that one is out. 

The sensor reports the temperature to the rest of the car via CAN - I think it talks to the ECM first, it is read by at least the BCM and the cluster.

Interesting bug. From what I've seen of the Mazda implementation (assuming a CX-3 is similar to an ND Miata), it's not really damped at the gauge end and I've seen no indication that it has a memory. On our V8 NDs, we don't pass along the OAT until everything has booted and it's a fairly low refresh rate, so it shows -40 (minimum value) when it boots up. As soon as we send the legit message, it pops right to the normal level. That's the gauge.

FYI, max value is 158F. It won't wrap, it will just not display any higher.

Fun story: a customer asked if we could use the OAT display as an oil temp display. Not a bad idea. So we coded it up and sent some test messages. Yeah, turns out the HVAC system uses OAT as a factor, and when we told it the temperature was 200F it went into OMG SAVE THE HUMANS mode and maxed out the AC.

Rons said:

Did the ghost in the machine cause it to temporarily read in C? 47C is approximately 117F.

That was my very first thought but I triple confirmed F not C.

I then did the math...212 - 32 = 180 so 47 X 1.8 = 85 plus the intercept of 32 which gets you to 117.

Keith, thank for the "far more knowledge than I have on the subject" explanation but this seems to leave us with an enigma.

As was the case yesterday, all was fine today.

Sorry, I got into the documentation and kinda geeked out.

As Keith pointed out, an open circuit would read -40 and a closed circuit should read.... a lot.  Being off a tiny bit would indicate a resistance problem, and it's awfully WEIRD to have a small resistance issue that slowly comes back to normal.

However.  I wonder what the effects of a minor refrigerant leak that happened to be aimed at the sensor would be.  Does your A/C still work?

In reply to Pete. (l33t FS) :

Hi Pete,

Interesting idea...the A/C is working normally but that doesn't rule out your theory as I could lose some coolant before noticing a loss of cooling.

Aircooled lives nearby and is a licensed HVAC Tech which means he answered "no" when asked if he would, under any conditions, release coolant into the environment and has one of those little stick thermometer thingies.

I'll let you know what we find.