koralr33fer

I’m pretty sure the ECM sends spark signal triggered by crank sensor. The crank sensor tells the ecm what position the motor is in and when to fire the spark plug. In my case the ECM was sending the signal to the coil, it just wasn’t reaching the coil. The plug going into the ECM was the culprit. It must have contracted the metal just enough in the cold that the wire and pin were losing contact to the ECM pin. I ended up drilling a small hole through the plug feeding the wire through and soldering the wire directly to the ECM. The issue has not occurred since. This was a battle we fought every winter before so I know your pain. If you are detecting rpm on your live data then your crank sensor is fine and the ECM should be sending spark signal to coil. The coil pack has 3 pins, check for ground, power and when cranking it should receive a signal from ECM. Without all 3 no spark. See which one you are missing and this will help narrow down where to start.

Might be a good idea to check your grounds and power inputs at the ecm. I would FULLY go through this the 2 parts in the FSM. The troubleshoot section has a step by step diagnostic section for a no spark condition, it’s a life saver.

So just to be clear, you had no control signal for the coil at the pin on the ECM? After you put the charger on the car you were able to get signal? I would just double check the wires have continuity from ecm to the coil. Should have close to 0 ohm resistance. When I had the issue putting extra juice to the battery would not get the car to start. If simply upping the bolts and amperage is getting the car to start you likely have a weak connection somewhere. Maybe go to the basics and make sure you got good clean grounds and battery terminals are clean. Usually if you can get an electrical failure to work with extra juice it’s a result of high resistance somewhere in the circuit. I’m pretty sure to get the spark you have to have power and ground at both the coil and ecm. Also you must have input from crank sensor and cam sensor. Also, just a heads up, if you get a crank sensor check engine light it’s often a result when you crank the engine a lot and the car doesn’t start. It kind of sends a ghost code to computer as a result of all the cranking. I had that code when I had the issue and I got too focused on the crank sensor when it had nothing to do with the problem. Do you have any codes?

I would guess it’s more likely you have a weak connection between the ecm and the coil. That is where my issue occurred. If the car works normally in warmer weather than it’s not likely the ecm is failing intermittently. My weak point in the connection was at the white blocks at the ecm. The wire had just enough play, and contracted just enough on warm weather to lose continuity. I would trace that control signal wire back to the ecm. Check it for continuity. The connection was poor enough for the car to not start although there was no ssignal gun of corrosion or any visible damage. Just because the connection looks ok doesn’t mean it’s tight. Those little metal wire holders n the block are s***. What I ended up having to do was drill a wire through the white block in an empty spot, soldering a wire to the correct pin on the ecm and connected it to the wire. In a way bypassing the white block where it was losing connection.

The Factory Service manuals for most lots of Subaru models are available here for free https://sl-i.net/FORUM/showthread.php?18087-Subaru-Factory-Service-Manuals-(FSM)-Every-Model-USDM-EU Saved me a lot of time and headache having these available. I attached 2 in particular that I used for troubleshooting the no spark (for 2008 legacy). Do the ignition one first, then check the ecm power and grounds. You can find the one for your model by downloading the one that corresponds to your vehicle and then going to engine diagnostic section, also you can do the find in document feature (control+F) and search for "spark" make sure you are in the correct section for your engine. Subaru Ignition FSM.pdf

Confirmed! Was 10 degrees out this morning and she started! Moral of the story I guess for a no spark condition would be to check at coil connection for computer signal, and don't assume because you are getting reading on a volt meter that it is working properly. Check to make sure the ecm is getting all required inputs like crank and cam sensors (typically if you were missing one you would get a check engine light and a trouble code though). Check power input and grounds at ecm, and if all that fails, do the wiggle test, have someone crank while you jostle some wires at ecm (the one not working properly was the B137 or the one on top when in ecm). That last test saved me like 300 bucks cause I was convinced the ecm was bad after all other tests were done.

Tried starting the car this weekend in the 30’s when it would usually start and couldn’t get car started at all Friday or Saturday. So I figured I had jostled something loose when checking wires on coil and at ecm and now had a complete failure. Which is almost better if you ask me. Got it towed to my dad’s who has a nice garage instead of working out in my cold driveway like I have been. Checked all the stuff at coil connection, got power and ground back to ecm. Checked for signal using an oscilloscope and was only getting a very weak waveform on 1 and 4 the control signals so figured we had no signal, just seeing noise from cranking the motor. Checked for continuity for all wires back to the computer and load testing them with ECU disconnected. Plugged ECU back in and I checked for cam and crankshaft waveforms while cranking with oscilloscope at ecm. Both looked okay. Checked all ecm grounds and voltage inputs. All ok, at this point I was certain it was the ECU. But just for shits and giggles we tried wiggling the wires and connections at ECU while cranking, and she fired! Moving the wires would cause to start and than not. Ended up being a loose connection within the plug itself, although the plug didn’t appear damaged or corroded. The wire that was loose was one of the wires for the control signal. I ended up doing a rather unorthodox repair in lack of having a spare connector handy and being unable to get the wire out even with a set of terminal keys to use. So....we forced out the 2 signal wires for control signal and drilled out the pin holes in the connector. Then we soldered 2 small wires to the pins of the ecm itself, then fed the wires through the connector and spliced them in to the existing signal wires.....vroom vroom! There is a chance that this connection became loose during my probing. But it would explain why I had no spark, a new coil, and no engine codes. If I was missing an input like cam or crank it would likely send a trouble code. But if there was an issue with the signal output that would not be detected as they are sent by the computer and not monitored afterward. Looks like Monday and Tuesday should be cold, I’ll know by then if I got it *fingers crossed*

The battle still wages on for us, I tried cleaning in between those connectors and have already cleaned the other grounds and the terminals. Today I tried troubleshooting using a FSM for no spark condition. Probed the connecter to the coil pack, got 11.6V on pin 2. Checked 1 and 4 and got voltage bouncing from 0 to 1.5 volts while cranking (I’m assuming an ok signal). Pin 3 goes back to the ecm, I got down into that and probed the connections there per FSM. 1.0 ohm resistance so got continuity back to ecm ground. Checked resistance on coil pack, seemed ok. FSM said to replace coil pack anyway since voltage on pins 1 and 4 of coil connector varies less than 10 volts. Got a coil just to eliminate it and still have no spark on any cylinders....idk what’s going on. Also checked for crank signal and cam sensor signal on ecm while cranking, can see varying voltages so assuming they are working ok and resistances for about 1.5k ohms on each circuit. Can also read rpms on scanner while cranking. No codes or history while scanning on snap on Solis. I’m stumped at this point... Anyone got ideas at this point? PS also tried to jump for extra juice, no effect

No success with my repair at the dealership. They wanted 375 for the diagnostic and the repair, I waited for a cold morning to pick up the legacy and it wouldn't start in their parking lot. So I ended up paying for the wire repair $200 and got out of the diagnostic charge. Still having the issue though. I'll have to try repairing that ground you mentioned. I cleaned the contact itself but never took it apart to do a real through cleaning.

Just got a call from dealership, they had the no start occur today and yesterday and came back with a decent diagnosis (for a change, last time I was told to clean throttle body for a no spark condition lol). They determined the wiring for the crank sensor had corroded and frayed inside the wiring harness, between the sensor and the connection for the computer. Not far back from the sensor itself. They said they were able to reproduce the issue by moving the wire around (don't know if I trust them on that tho). They wanted 375 for the work plus the diagnostic. I told them to go ahead and do it (usually I do my own work, first time at a shop/dealership for anything). One reason I'm not is I'm sick at the moment, and two if the issue occurs again I can go back for another free diagnostic. I'll let you know how I make out. I kind of figured it was either a bad wire/connection or the computer. I hate chasing wires though.... I wont say its solved yet though until I see it start in cold weather a few times.

Mines at dealership still, I’ll let you know what they find out. I’m really starting to suspect that the PCM has gone bad. Often times the only electronics that fail in cold are the ones with circuit boards with cold solder joints. Often times the cold causes the components to contract and the solders come apart. This is a common issue with the passenger airbag indicator on the roof in front of the car. I’m not gonna replace an expensive computer on a whim though which is why I went for the diagnostic. We’ll see what they say.....

I feel ya buddy, I'm experiencing the same exact issue with my girlfriends 2008 legacy. Wont get spark when its cold, I had a code for a crank sensor at one point and replaced the sensor and still have the same issue (possibly a ghost code from excessive cranking and no start). The battery is about a year old. This issue has been occuring since last winter. I tried finding a control signal pulse from the PCM at the connection for the coil and wasnt seeing anything from the computer. Only had 12v power on 1 of the 4 pins and no signal from the other 3 (im not sure what pin(s) its supposed to be on but I figured I see something out of the other 3). Which tells me the PCM isnt sending the signal to engage the coil. So either my PCM is bad or a bad connection somewhere or PCM isnt getting the required inputs to send the signal. I would think if it was missing another input like a cam sensor that it would trip a code. BUt I havent had any codes other than the 1 crank sensor code. I really hope you find a solution as your issue sounds identical to mine. If you find a solution please let me know. My car has been at dealership for 2 weeks waiting for a diagnostic.