When I first got the car in October of 2021, even before it's initial service I took it to the grocery store. Hey, we needed food! When I got there I parked but wasn't happy with the parking job so I went to start the car and straighten it out - but it just turned over and wouldn't catch. I went into the store to get groceries and figured I would deal with whatever when I got out. When I got out, it started like a champ. Hot start issue? Who knows. We did a whole bunch of maintenance tasks and it never happened again.
That is until about 3 weeks ago.
I was going on errands, I backed out of the garage to let it warm up a bit and then ran inside to get something I'd forgotten. When I came back out, it had stalled. When I went to re-start it, it just spun quickly and didn't start. I waited maybe 2-3 minutes and then it started fine, and it has been running fine ever since. (we had another stall, with me driving since this)
Which is kind of annoying - it is so much easier to ID stuff that is broken all the time. Stuff that is broken for random moments here and there is tough. But I have been studying, to prepare for "next time" - or ideally to prevent a "next time".
Step one was learning a bit about T-shooting fuel injection systems and reading online forums about other 911 owners with no-start issues. There are some very smart and experienced folks out there, and I will hopefully hit them all in the references below. I'll list out the tools I picked up along the way as well.
The fuel pump came up a lot as a point of failure, as did the fuel pump relay (under the driver's seat). The coil came up a bit too. Everyone is focused on the thing that failed for them, which is probably normal. I learned about "noid lights" - you plug them into the fuel injector harness to make sure your injectors are getting a signal to release fuel. I also bought an in-line spark tester - it goes between your distributor and the plug and has a window you can see the spark through. Between that and the noid light I would be able to verify that the car was getting fuel and spark. The fact that it doesn't catch at all leads me to theorize I am missing at least one of those, and possibly both. I also grabbed a fuel pressure gauge - there is a test port on the LH fuel rail (as you face the engine) where you can check it. A weak pump can cause issues. But the car runs fine, no issues winding it out and the power feels fine too. Which makes me think it is something more basic, more boring.
So being an electric nerd, I looked at grounds. There is a very important engine ground on the LH intake manifold. One of the systems that grounds here is the DME, including the crank sensors. The Bosch DME system uses a crank position sensor and a crank speed sensor. They are both Hall Effect sensors, so they 'watch' a toothed ring on the flywheel. While the speed sensor won't stop the engine from starting, the crank position sensor will. It will cut both fuel and spark as a failsafe. Since we won't know where the crank is, we can't know when to inject fuel or try to ignite it. Makes sense.
So step one was I read the schematic, I found the ground and I cleaned it up. Glamorous & complicated work but someone has to do it. The Bentley manual has the ground diagrams as well as a chart of where they are on the car. Super helpful!
So here is our ground, lots of wires going into 2 eyes.
I removed the aux heat duct to get better access. you can see the fuel injection test point at the near end of the fuel rail on the outside. (if we need it). You basically remove it to work on anything on the left side of the engine. 2 screws.
Yes, these are pretty crummy. I'm sure they make a connection (the car works after all...) but if they are 5 or 10 ohms when they should be zero in theory, it could throw things off. Or if a little jiggle causes the car to stall (of course I did not try that...duh)
I picked a Dremel over the plumbers paper because I did not want to remove any plating and put scratches into surfaces. I also didn't want loose media in the engine compartment. Ideally I'd use a brass brush here to be gentle on surfaces.
One down, one to go. That's a lot of corrosion. 35 years worth.
Don't forget the bolt!
I polished up the manifold where this all screws down.
A friend who works in the industry recommended Nyogel 760G for protecting the ground after I re-assembled it. The idea is to keep water & air out of it so it maintains good contact. I gooped a bit of this on top of it. Nyogel is commonly used inside of push-on engine connectors as I understand it, but is also very good here.
Now, I didn't read the manual first, which was dumb, but when I did (later), this is what the troubleshooting page for no-start looks like:
So I feel semi-smart - I went for the most likely issue first instinctively. I also am planning on polishing/tightening up the fuse box connections; the fuse for the fuel pump could cause these symptoms possibly as well. I have not yet had the downtime to replace the fuse panel although I did mark the wiring so I am ready to. I am not sure what they mean by "ground behind the fuel filter", I am pretty sure the ground in the manual is the one on the manifold. But I'll check this all out when the car gets back. * Update: I did check the manual and the correct schematic. The entire DME grounds on that intake runner, so that is all of the sensors, the fuel injectors and more. There is a ground behind the fuel filter (G402) but I am not sure what runs to it yet. (see below)
Here are the ground locations on the car. G407 is the biggie, and what we just cleaned up.
Page 1 - the DME is the "box" on the left. You can see G407 notated at the bottom.
Page 2, the stuff "to the right" of page 1. The only 2 things that are not grounded to the manifold at the fuel pump and the oil cooler temp switch and blower motor.
As far as the sensors, I did take a look when I had the aux-heater duct out:
Top = Engine temp sensor (installed in the head for cylinder #3)
Middle "DG" = Crank Speed Sensor
Bottom = (with missing insulation) is the crank position sensor (aka "reference sensor"). Not confidence inspiring. I can replace a bad section of wiring goop it up and cover it with shrink tubing possibly. The sensors are not expensive, but are a bit of a PITA to install. The two of them ride in a mount that screws on down by the flywheel under the car and which seems to be difficult to remove (based on online accounts). Then you need to re-set the tolerance - the distance from the flywheel is important. I will start with seeing if the resistance and then signals are in-spec and if so, then see about protecting this lead as necessary.
Next Steps:
* Check fusebox / clean contacts
* Review ground diagrams to look for other possible grounds I should clean up - for this and in general.
* Test crank sensors with DMM & Oscilloscope (luckily I have a couple of those!) That will be fun.
If the car stalls again in the meantime, I will keep the noid light & spark window with me in the car as well as a DMM. They're all small. The only issue there is that 2 of 3 are two person operations. Someone tries to start the car and person #2 looks for the noid light to light and the spark window to show a spark. If they are both "off", then it is likely the crank position sensor. Breaking both systems at once is a bit too random. I will also check the voltages to the fuel relay. If there is a spark and no fuel, or fuel and no spark that will send us in other directions. Or if the noid light just stays on that is a different issue. More fun to come!
To be continued
Sept 2, 2023
It stalled the other night coming home from band rehearsal while I was driving, which while not fun was a bit of a gift. I finally got to see what the car is doing when this happens. The tach went to zero while the engine was still turning maybe 2,000 rpm (going maybe 30mph?) and the engine was not responsive at all to throttle. It felt electrical, like the power got cut.
I looked at the schematic more and asked around other car groups to see if others had seen similar issues. The thing that stuck out on the schematic that I didn't understand before is that the DME relay (next to the DME, under the driver's seat) not only powers the fuel pump, but powers the DME (the engine management computer). Reading up on line and then looking at responses to my questions, it seems like the relay is a common problem with many cars and can cause exactly the sorts of symptoms I am having. Intermittent connections in the relay can cause the DME to lose power.
A couple of suggestions were to test the existing one by unbolting it and then with the car idling, try dropping or moving the relay around. It should stall the car if there are bad solder connections internally. Another suggestion was to purchase a solid state replacement that won't have these issues, which I did. I purchased a replacement and got the optional diagnostic LEDs that give me circuit status as well.
The Hall effect sensors seem to be less failure prone, and a quick heck of their resistance (around 970 ohms) should be enough to show them as good or bad. I am off to the garage to try bouncing the relay around, maybe do some soldering and also verify the DCR of the crank position sensor. The precise value given in the manual is 960 ohms, +/- 96 ohms (10%). That is between pins 1 & 2 (counting from the bottom of the connector) 1-3 and 2-3 should be very high - they're just looking for bad insulation. My reference sensor value measured 1,055 ohms on my Fluke DMM, which is the top end of the acceptable range. The insulation tests were both off the scale which is great. So despite shedding insulation, it seems fine.
One thing I kind of love running into is the same tech over and over in different places. Speakers & microphones are the same basic device, and with a small tweak you get solenoids and relays. Tweak some more and you get electric motors. Here, the Hall effect sensor is a kissing cousin to a guitar pickup except in the guitar the magnet is inside the coil of wire and you detect the steel string moving. The Hall effect uses a coil of wire around steel to detect a magnet moving by. Same three things, combined differently but getting to the same place. Kinda neat.
September 5th
I did some basic maintenance over the weekend, replacing the air & fuel filters. I hadn't done the fuel filter before but once you accept that the aux heat duct has to come out for most tasks on that side of the engine, it's pretty easy. And once it is out, it is easy to test the engine sensors. So I did. You need to remove the retaining clip - a square shaped wire that clips around the plug. I use needlenose and a dental pick to get it off without dropping it. Then the plugs pull off. I also checked my grounds.
Head Temp Sensor (top): This measured 1.42K in 80 Deg F ambient. From 60-85 it can read from 3.6k to 1.4K so we're fine here. (R lowers with temp)
Crank position sensor: Mine measured 1,055 Ohms. Spec is 960 +/- 96 which is 864 to 1,056 Ohms so I'm still ok here. Measurements from pins 2-3 and 1-3 were off scale (also good).
I measured the intake manifold ground I fixed up at .1 Ohms (to another point on the engine) So that looks good.
When I had the fuel filter out I also found that other ground connection behind it. The only things grounded there are the rear wiper, the rear defroster, the heater & the oil level sender. I measured it at .1 Ohms to my cleaned up ground so I left it alone.
So my sensors and grounds look ok, and the more reading and talking to folks I do, the more it looks like a Fuel Relay/DME relay issue. Especially when i think about the tach going to zero while the engine was turning. The tach signal is generated by the DME, so for it to go to zero, I think that means it almost had to have lost power (or have some internal fault, which would stink!!) So I bought the solid state DME relay and a URO replacement as well since it is inexpensive & I could get it overnight. The Bentley manual says to take the driver's seat out; the 911 people said pish posh, move the seat all the way back and then use the power adjusters to move it up as high as it goes. And wouldn't you know it, that worked great. It's a 10min job. The only tricky bit is finding a 10mm socket the right height; regular is too short and deep is too tall. But you undo the one nut and washer, the relay lifts off its stud, you remove the plug (which is polarized with a plastic pin) and connect the new one and bolt it down.
I swapped in the Uro relay until the solid state one gets here. I wanted to pop open the old one and see what is going on in case I could do soldering magic or something.
The case pops off fairly easily (I used a dental pick)
Lots of room to work. Normal seats have a plastic bit of trim there (you can see the 2 clip nuts). Just missing on mine. But it makes no difference here. You can see the DME (left), Fuel Pump/DME relay (middle) and altitude sensor (right). A single 10mm nut, like every key piece of every car...
I popped out the relay circuit and the far end (away from the pins) looks pretty crispy. They are the same relay doing the same job. Why would one be that hot? No bueno.
Another angle - that 2nd relay has discolored the plating due to temp.
I could remove and re-flow all of the solder, but I honestly don't see any bad connections. I could re-install this loose and poke it and prod it looking for issues. The high temperature leads me to think I should just leave it and replace it. If I can't diagnose why it ran hot I am just putting a problem back into the car.
I used a DMM to check the diode (it's fine, .544V measured) and check both coils; their DCRs are 71.4 ohms and 80 ohms. So if the spec was say 75 ohms, +/- 10% they are well within that. Even if the spec is 80 ohms, 71.4 is pretty close to 72 which would be ok. Nothing seems really off here. The switches each have a low (.1 ohm) closed resistance when I closed them manually, and an infinite open resistance. So that all looks fine.
Pin 87 looks to be the issue - this is how power gets to the DME from pin 30 (battery) When the relay is engaged, those switches at 87b and 30 swing over to make contact. This connects 87b to power and power to 87 turning on the fuel pump and the DME.
Net net, I will re-flow the whole board and keep this as a spare along with the Uro replacement. I think I am going to grab a strip of foam to 'shock mount' the new solid state relay. While it should be tougher with nothing heavy moving around inside, why abuse it?
Sept 12th
Well, I got the solid state unit in the mail and installed it and it also works fine; all the LEDs light up and the car starts. I have the spare from Uro in the glove box with a spare 10mm crescent wrench. But that leaves me with the old one that "usually" worked. What was really wrong?
I started by taking measurements (above), so now I removed all of the old solder using a de-soldering gun. I like the ones that Hakko makes, I am using a Hakko FR-301 (link below) and it made short work of the board.
The new solid state unit displaying status. it fit perfectly
Having run the Uro relay for a few days with no issues and then plugging in the solid state unit - also with no issues I decided to try my re-flowed stock unit. Well, the car did not start. It just spun quickly. What gives? My soldering is not that bad. (!) I deserve better!
I brought the relay back up to the bench and checked everything again. What I found was the diode had an intermittent connection - if I tested it just touching the leads it read dead (open) but if I pushed on the negative terminal just the right way, it would work and show the correct reverse voltage.
I used the Hakko to remove the stock diode so I could read off the part number and see what I had around the house that was close. The stock part is a 1N4003. (specs below) this is just there to take the flyback from the relay coil [as far as I know?] There is no polarity on the schematic although we know from how diodes work & then testing that the negative end (the cathode - where the band is) connects to the coil and the other end of the coil is connected to ground via pin 85. With the diode "open", the coil would never energize, which would mean the DME would not power up. This is consistent with our issue.
I don't have any 1N4003's around but I have a whole drawer of 1N4007's - they are the standard replacement in old tube guitar amps like Fenders. The whole 1N series has the same basic specs; the chief difference is the 1N4003 can handle 200V, the 1N4007 can take 1,000V. Not an issue. I might now have the beefiest DME relay on the planet earth. <grunting noises>
Here we are about 2/3 of the way through removing the old solder and re-soldering the board. Those holes up top are what the Hakko does - it leaves them really clean in a couple of seconds. parts will usually pull out or fall out when it's done. I'd never buy one to fix cars, but if you do a lot of PCB work, they will pay for themselves quickly.
Installing the new 1N4007. I use fine needle-nose pliers to carefully bend the leads. You don't want stress on where the leads connect, and you want the device to hold itself to the board - solder is not there as glue. So I like to bend the leads to hold the part in place, then solder it and clip the leads with a pair of dykes.
All better. new 1N4007 in place, soldered and leads clipped. A casual observer might never notice it had been replaced.
If you have a diode test setting on your meter, clip your negative lead to relay terminal #85, and your positive to relay terminal #86. You should read the .548V bias of the diode. Reverse the test leads and you should get OL. (out of range). With my bad diode i was getting OL both ways and the car would not start. 85 & 86 are the pins to energize the relay coil that connects power to the DME.
Now that the diode has been replaced and the solder connections on the board have been cleaned up and re-done, the DME relay works just fine. I am fairly confident we found the root cause of the issue and addressed it, and now I have a 10mm crescent wrench and spare relay in my glovebox.
References:
Great tech info on DME Relay: https://911chips.com/dmerelay.html
Hall Effect Sensor: https://en.wikipedia.org/wiki/Hall_effect_sensor#
Noid Lights: https://dannysengineportal.com/noid-light-how-to-test-fuel-injectors-using-a-noid-light/
Solid State DME Relay: https://www.ftech9.com/new-products/911-ssr
Check a diode with a DMM: https://www.fluke.com/en-us/learn/blog/digital-multimeters/how-to-test-diodes
1N series Data Sheet: https://www.mouser.com/datasheet/2/849/1n4001-2577232.pdf
Comments