Hi all,

I just finished bolting back together my new engine setup for this season (which I will make a build thread for eventually, feel like the PO did a great job with that so I feel I should take the helm), and my CLT sensor has stopped reading. Car is a 10AE with VVT swap, PO's build thread is here in case anyone is curious or familiar: Hairdresser HP - An NB 105kpa max HPDE Build - Miata Turbo Forum - Boost cars, acquire cats. Here is kind of a rough sequence of events:

1. Assembled and dropped in engine, got everything wired up and ironed out, engine (w/ new 264/264 cams, sounds badass!!) fired up and ran successfully with reasonably good CLT feedback (as compared to my temp gauge in the cabin).
2. Perhaps unnecessarily removed ECU from car to solder in wire to flex pin (#26 on the MS3 Basic. This isn't working yet either and I wonder if I have a bum sensor; will report back after trying second sensor).
3. Reinstalled ECU and now the CLT sensor is reading 180F at all times, which I believe is an indication of failure. Car will still start and run just fine when given some gas.
   1. Have had issues in the past where a few of the parameters go haywire, and I noticed that depressing the harness connector or the female USB port towards the ECU a bit seemed to correct the issue until I guess the car shook it out of position again (for what it's worth, this never affected how the car ran or drove). This has been ongoing since I've purchased the car, was never super bothered by having to jiggle the connector once every few months.
   2. A note: the harness connectors are aftermarket, PO had the harness extended so the ECU can live where the passenger airbag used to. It's a nice touch for sure, but the connectors are VERY tight and require some force to get out.
   3. If I change the Bias Relay value in TS to something small, like ~400 ohms or less, I can get the CLT to jump around a bunch. Not sure if this means anything but maybe it's useful data to someone who is more electronically savvy than I am.

Based on 3.1 and 3.2 above, I feel as though there was a loose connection somewhere in the ECU (long L-shaped pins that are soldered to the board from the female connector port?) that I may have finally killed reconnecting the harness. So, I pulled the board out of the ECU housing and found that some of the teardrop-shaped solder points at the bottom of the board (below connector 2) looked like they were almost scraped off or something. I'm not sure how that happened, but I'm also not sure if this would really affect anything electrically either (looks like the connection to the board is still made). Picture below, do we think that I should maybe attempt to resolder the points that have this damage?



My other thought was maybe to try to find the bias resistor on the board and see if it's functioning okay. Based on past symptoms and things I've changed recently, I'm doubtful it has anything to do with anything beyond the connector, but it may be worth checking. The ECU is like 12 years old now w/ lots of track time vibration/abuse.

Even though nothing has changed with it since it became inoperable, I checked out the CLT sensor harness with a multimeter and got some results that I'm not sure how to interpret. Here is a reference picture of the connector:



So both of the "top" pins are reading just under 5V when grounded to the valve cover, which I feel is maybe not right because I believe one is supposed to be a ground? The "bottom" pin reads something like 9V -- I'm not sure if that's weird or not, but I do know that pin is for the dash so I kind of care less about it at the moment. I know that one of them is supposed to have 5V, the other two results I'm not sure how to interpret. I will determine which of the top pins is the ECU signal and verify continuity to connector 2E later today (as the aftermarket harness is definitely a variable here) as well as perform a good visual inspection of the wiring.

Has anyone had issues like this, or have any feedback on the information above?

Thanks very much,

CJ

 

failed/unplugged sensor should read 70°F.

 

Appreciate the reply! Just so I understand, that'd be regardless of calibration? As in, 70F is the default (non-configurable) return value when the sensor is out of range (e.g. resistance is extremely high)?

 

Yeah that input has a 5V ref. so if it floats up to 5v, it knows there's a failure and will drop the temp to 70°F as a failsafe. I believe it would do the same for a shorted sensor as well.

I would try unplugging the sensor and seeing what reaction the MS has (or try shorting the pins with a wire or a ~2.5K resistor).

unless there's a way to update the fallback value and yours is set to 180°F (or Rev made his Basic units do that)?



But sounds to me like maybe there's a cold solder joint somewhere on that circuit...

 

I'm pretty sure Megasquirt has a default value that it'll use if the sensor is out of range. I'd assume it would default to 180 to try to keep fueling reasonable. I know Link has that capability as well.

 

Okay, gotcha. I do get the same 180F value when I unplug the sensor, so I am thinking that's the fallback value and maybe that's something specific to the Basic like you're suggesting. For fun, I'll try to short the pins and see what happens -- a resistor is a great idea, I'll dig through my Arduino kit to see if I have anything hanging around.

I would agree (with my limited-to-none PCB experience) that the symptoms would suggest a cold solder joint, but everything looks really good as far as I can tell. I haven't really had the chance to dig into the board schematics (assuming it's the same as what Megasquirt has released -- after all, it's a Megasquirt board right?), so I'm having difficulty visually identifying all points of the CLT circuit. I do see two 2.49 kOhm resistors on the board that I've tested with a multimeter that are returning correct values, and have gone through anything nearby that looks like it might be associated with the circuit. Also, I've tested continuity at the female connector pins to the solder on the board, and it looks solid. If I didn't mention before, there is also continuity between the CLT connector and the harness connector locations 2E and 3F. My plan today is to hook up the harness and see if I can get good continuity to the solder joints associated with each pin location (on the bottom of the main board) from the CLT connector to rule out an issue with the harness connector itself (which is aftermarket).

 

You should be able to find the fallback value in the tune, or upload the tune here and someone can probably find it for you. A log might not hurt either in case there is a tune/setting issue, although that sounds unlikely given what you've said so far.

I was going to say, aside from grabbing a multimeter and testing for continuity I'm not sure what you'd be doing at this point.

 

Yeah, you're probably right -- I'll check that out a bit later today. No opposition posting a log, but I'm not sure why anything in the tune would change randomly. I will say, the ECU was unplugged to solder in a Flexfuel sensor (and the mechanical pieces were done by your write-up actually, so thank you! No leaks!), but that connector is on the complete opposite side of the ECU so I can't imagine there'd be any interaction.

I would really love to know the full layout of the CLT circuit so I could chase it around with the multimeter, but all the documentation that I can find is pretty enigmatic (to me, the layman).

 

Glad it was of use to someone! Do you have pictures of those solder joints? IMO if you were just in there that would be the first place I'd start investigating. Also, considering you said that's not working either I'd be even more suspicious.

Can you get a top down photo of the backside of the ECU so everything is in a similar focus?

 

So, the pin I soldered (probably poorly, which may be the reason I'm having issues with the FF sensor) was on the DB37 connector at the rear of the ECU, where all of the non-OEM I/O lands. I think this corresponds with what they call the "expansion board" -- it's a small white board that is oriented towards the front of the ECU. The signal in question lands on the front of the ECU with the rest of the factory stuff (ECU is PNP), so nothing has been modified around there -- the only action it's seen is being plugged/unplugged since this issue started.

That said... I got my hands on some schematics from the Megasquirt MS3 hardware manual and was able to identify the resistors associated with the CLT circuit -- all checked out. I then put the ECU all back together, plugged the harness connector into it without installing it into the case, and successfully proved continuity between the CLT sensor harness pins and the solder joints on the bottom of the board associated with pins 2E and 3F. So it's making it to the ECU, harness connector is ruled out. As a last test, I dug around and was able to find a 1K resistor (not quite what @Braineack suggested, but it's all I had) that I installed between the two CLT pins with no change. I think the last thing I can do without shipping it out somewhere is to fully define the point-to-point path the CLT signal takes and test every connection and component... but, my inexperience is failing me a bit here. I think maybe I just don't speak the language fluently, but I am not able to discern that level of detail from the schematics.

As you had requested, here are a bunch of pictures with kind of crappy captions. Please let me know if you see anything suspicious:


MSLabs board 1

MSLabs board 2

MSLabs board 3

MSLabs board 4

MS3 Main board. Trying to show solder joints for resistors.

MS3 Main board 2

MS3 Main board 3

 

Did you clip the solder joints with some wire cutters or something? The ends don't look quite right. I doubt that would cause issues though if the soldering joints are solid.

Yeah, you'd basically be tracing the paths on each side of the board visually to figure out where they go, but if there's an internal fault there I think you'd only be able to find it with a multimeter.

If this is an MSLabs board your best bet is @Reverant

 

Realizing that Reverant is Dimitrius w/ MSLabs… I have been emailing back and forth with him. On the solder joints – he mentioned he ground some of them down to prevent contact with the case.

We tried a few things that didn’t work, but then I mentioned to him that there is 5V on both sides of the CLT connector (which I think I said in the OP) – he confirmed that is NOT normal. So, I took a look at the other sensors that share that 5V/sensor ground, and every single one is actually not working and all have 5V on both pins.

So, since all the grounds seem to be in place happily… I guess I’m going to tear the harness apart to make sure that nothing is contacting.

 

Can you show us what you did for the flex sensor?

 

I can, but I've learned that it's moot recently. Apparently for this generation of MS3 Basic (V2.3, she's pretty old), the Flex pin is used for VSS1 or something and you actually need to land the signal wire at the PE0 pin w/ some resistors added on the board -- I have a question out to Dimitris on the last part, the thread I saw that on has a dead attachment.

But, for the sake of answering you: I pulled power from some EVAP system plug that is near the passenger front shock tower, sent signal back to pin 26 (Flex pin) on the MS3's DB37 (which I soldered in; my solder is a little ugly because I did it in the footwell of the car with a pretty crappy iron), and I've actually tried a handful of ground locations, including the instrument ground (pin 3 on DB37). All of the wires are installed into pins at the connector (tried not to crimp anything unless it was absolutely necessary. I can include pics if you want to see, but it's pretty similar to @SimBa 's install.

Also, as a general update, it turns out pin 3F is dead on my ECU. Dimitris suggested landing the ground wire at pin 3C instead (which looks like it'll just route it to the ground near the front of the IM), so we'll give that a shot today and see if everything resolves.

 

Just as a data point, have you tried disconnecting the flex sensor and does that change anything?

 

Yes I have, and nothing changed as a result -- the thought definitely crossed my mind that I either fried something or got lines mixed, so I made sure to triple check my work. For what it's worth, we were able to determine ground pin 3F was dead with the ECU disconnected from the car entirely (although I also performed the same tests through the harness just to make triple-sure).