I think maybe TK should open the right switch housing and inspect the starter button contacts and the wiring solder joints.

Going to ask a question here.

The temperature sensor.

Am I correct in saying that if it reads that the engine is overheated that it inhibits a start by cutting either fuel or spark? Putting 16v temperature sensor in 2v engine does that. Has it been changed out?

Looking at the drawing makes me think that the fuel injection computer could deny earth to the injectors based on a signal from the temperature sensor, but that wouldn't explain why the positive Voltage drops while cranking. It behaves normally with the injector harness unplugged, so I'm thinking the problem is there. I'm thinking maybe a short in a positive. Maybe an injector has shorted (maybe the one that burned) explaing why it works normally, after the injectors are shut down, for the last two seconds. I have to say this is over my head, so this should be ignored, unless someone verifies at least a part of it.

Daveyson,
I was thinking that maybe the power to the fuel pump was dropping out even as the start button is being pressed.

Yes, I think that now too after after the results from Roberts test.

I'm trying to think of what stops when the start button is released, cause that's when it then works normally again, making me think injectors.

But then it worked in October, making me think airflow meter, I'd best drop out.

I like Roberts idea of unplugging the consumers and testing for shorts.

+1 on comments by Robmack!

even as the computer and fuses are working there is still the connections at the tank.
maybe do a continuity test of the wire going directly to the fuel pump.

Hi Olaf,
missed your comment. hmmm temp. sensor. I’m not sure. interesting.

The FI relay may be damaged in ways that defy understanding based on the symptoms we observe (i.e. the pump running for 2 seconds after the start switch is released but not running normally).  It's a cheap fix to replace the relay. Maybe it fixes the problem or maybe not. Maybe this problem is disguising another problem.

I was thinking that another experiment would be to put everything back in place except for the FI relay and then to jumper between pin 30 and both pins 87 on the relay socket to simulate the FI relay engaged. Then try to start the bike.  This jumping would be very temporary, only for the duration of the experiment.  It would prove or disprove the defective FI relay theory. This experiment is not my most favourite idea because I'd much rather see a correct replacement done.

Robert, is there any merit in this idea? Using the horn relay (no inner 87 pin)
meaning there would be no power to the pump, to check that the power doesn't drop out from your earlier test. Or cutting the Outer 87 pin Of the fuel injection relay  (or bending it over) and repeat the test?

I have done both of these tests previously, multiple times, without issues, chasing a strange "bike only runs without pump" problem.

Actually, another test I remember doing was a jumper wire between both 87 sockets and the horn relay.

Just watched your first video again. I'd still be looking at the AFM.
Maybe try to start it with, and then without the AFM connected - any difference?

I could be missing something btw, ignore me if I am 

daveyson wrote:Robert, is there any merit in this idea? Using the horn relay (no inner 87 pin)
meaning there would be no power to the pump, to check that the power doesn't drop out from your earlier test. Or cutting the Outer 87 pin Of the fuel injection relay  (or bending it over) and repeat the test?

I have done both of these tests previously, multiple times, without issues, chasing a strange "bike only runs without pump" problem.

I just threw it out there; maybe no merit to it and I'm barking up the wrong tree.  It's easy to verify the FI relay's condition which will either prove or disprove my theory.

robmack wrote:Try this experiment.

First identify the following two connectors on the FI relay socket 

•  the pin 87 that leads to the power feed to fuse 6 (Green/Red) 
  
•  pin 30 (RED). Be very careful because there is constant power on pin 30 whether the bike is on or not.
  

1. Disconnect the 5-pin connector to the injector harness
   
2. Pull the FI relay
   
3. Make sure Fuse 6 is installed
   
4. Take a jumper wire and carefully short pin 30 (RED) on the FI relay to the previously identified FI relay socket connector.
   

The expected behaviour is that the fuel pump will start to run and will continue to run as long as you maintain the pins shorted.  Remove the jumper and the pump will stop running.

This will verify continuity of wiring between the fuel pump and the suspect pin 87.  If that pans out, then the failing element is most likely the FI relay.

I intend to give this a try this afternoon, once I wrap up work. But I do have a procedural question... I don't find it very practical to get into the relay/electrical box without the tank removed. But in order to assess if the fuel pump is running, per the expected behavior, I would need to have the tank/fuel pump wired up (and either drained or the fuel lines connected). How do you recommend I proceed?

Prop the tank up at the back using a wooden block.  It should allow enough space to get to the relay box contents without disconnecting the fuel lines.

Okay, very good! I had propped it up some, but not quite enough to get in there. I'll get more aggressive with my propping

A longer hose to the front of the rail is a big help.

In an old post I've also seen one tilted almost 90 degrees to the left, Suitable for your earlier model,  with the electrical plug on the left. Probably more difficult to support the tank though.

TrailKlaus wrote:Okay, very good! I had propped it up some, but not quite enough to get in there. I'll get more aggressive with my propping

Be careful. In old posts I have commented that old fuel lines go hard and crack so if you have never replaced them, do so. They use 8/13mm fuel line same as Hondas but make sure its for fuel injected engines. Make them about 1 1/2" longer and you can move the tank much better.

92KK 84WW Olaf wrote:

TrailKlaus wrote:Okay, very good! I had propped it up some, but not quite enough to get in there. I'll get more aggressive with my propping

Be careful. In old posts I have commented that old fuel lines go hard and crack so if you have never replaced them, do so. They use 8/13mm fuel line same as Hondas but make sure its for fuel injected engines. Make them about 1 1/2" longer and you can move the tank much better.

I appreciate the useful reminder! Fortunately I had replaced all of the fuel lines previously, so they're fresh and flexible. 

I got the tank situated, had the 5-pin connector disconnected, had Fuse 6 installed, and I pulled the FI relay and bridged pin 30 to the identified pin 87 (I think the relay shows it as being 87b). The fuel pump did turn on and run sustained until I disconnected the jumper wire. Which makes the FI Relay look like the prime suspect at this point.

If you're near a wreckers (junk yard in the US I think) look under BMW bonnets (hood in the US I think)

If you see a light green relay, check if it has five pins. If it does it's good for your brick, which can be confirmed on the net or somewhere.

Edit: The part number should be 6136 8 373 700.

Make sure the relay has this picture on the case:



I have had a little bit of luck getting replacements at the local auto parts store. They might have to get it from the warehouse, but that only takes a day. I think the last one I got cost me about $22 or thereabouts at Casrquest/AutoZone.

Hang on a sec...

The injectors were working in October, but that was with the rail away from the bike. The injectors should be isolated by the o rings, and the rail by rubber washers too, I thin. Spose an injector is shorted and it's metal retaining clip touching the rail (or through the bolt)

The fire caused by a loose injector makes me wonder if the previous owner has installed the rail wrong. Maybe steel washer on wrong side. The Voltage drop might disappear with the the rail off again.

Might this be a problem?

I have had injectors off a couple of times and it can be tricky to get them correctly sealed putting them back. I have had a leak and had to take them back off and in 2 cases had to use a spare O ring to get it right.

As soon as you have them reinstalled I think pressurise the fuel system before connecting them up, check the O rings on the rail to injector side, then connect everything and do a retest checking both the rail side seals and injector to head seals

If you miss a seal, as in one has fallen out as you are installing, or have a leak, be sure to find it at this stage. Not too difficult to get a fire if you have fuel under pressure.

Good point Dave! Let me add that I would expect the coil in the injector to be insulated from the body. I doubt the o-rings that seal the injector to the rail or the cylinder head are depended on to provide any electrical isolation.

An easy test would be to check the resistance between each of the two terminals on the injector to the metal shell of the injector. If everything is okay, I would expect the resistance to be very high, essentially an open circuit. I could see a short circuit through an injector body to ground causing problems for the ECU.

While a failure of this nature is really unlikely, it won't hurt to check.

TrailKlaus wrote:... Success? When I pressed the start button the voltmeter showed about 10.2V, and when I released the start button it climbed towards 12V for a second or two.

So, I quickly attributed this to a successful test and started down the path of trying to isolate the failure in the FI relay.  But the results have been nagging at me ever since.  There's something not right about this picture.

When the OP did the test, the FI relay was removed so there should not have been any appreciable load on the circuit supplying current to the FI relay coil (which is essentially what we're testing here).  Yet there is an appreciable drop in voltage of 1.8V.  Voltage just doesn't disappear and reappear as if by magic.  Voltage gets divided across elements in series under load (i.e. when current flows). A voltmeter puts no appreciable load on a circuit (i.e. it's a very high value resistance) so there should not have been any current of significance flowing in the circuit when the start button was pressed.  Yet there is evidence of voltage drop.  Across what element in the circuit was this voltage being dropped?  The only logical place is inside the ICU in my mind.  Why did the voltage rise once the start switch was released?  The FI relay has no knowledge of the state of the start switch.  However both the ICU does have that knowledge and controls the FI relay based upon that knowledge. Again there is evidence pointing back to the ICU.

I'm beginning to suspect the ICU as being defective.  I'd encourage the OP to test the FI relay for proper operation before doing a part swap (unless you can get a known good relay for cheap). I suspect that we'll see the OP replace the FI relay and the same behaviour will be exhibited. Then we're back at square one chasing other theories.

Early ICU have an issue of internal condensation causing problems and its not impossible is the case here. I believe a little issue on my RS is being caused by this and as soon as its warmer I plan to change in a spare one to see if it resolves the problem. The bike acts as if it has fuel starvation and everything else is good/known good and there is no current draw.

But I feel if this fault is in the ICU perhaps it should possibly be activating the relay and drawing 12v.

92KK 84WW Olaf wrote: ... But I feel if this fault is in the ICU perhaps it should possibly be activating the relay and drawing 12v.

The ICU controls the FI relay by manipulating its ground signal.  It joins Pin 7 of the ICU to ground when it wants to engage the relay and breaks that connection when it wishes to disengage the relay.  What if the ICU has an internal fault in the ground control circuit that causes an internal voltage drop while the start button is pressed. The fault would rob the FI relay of voltage needed to meet its minimum pickup voltage so the FI relay would fail to engage.  When the start button is released, all of the available voltage suddenly becomes available to the relay and it engages. But it will do so only for the 1.5 - 2 seconds programmed in the ICU and then the relay drops. That theory meets the observed behaviour as we currently know it.

I can not remember and just a thought, have we ruled out the fuel pressure regulator? Ive read that it can cause similar issues as described here.

I was thinking the fault is somewhere related to the injector wiring cause the Voltage drop disappeared when it was removed. I just thought the 2 Volt drop was because the engine was turning with the button pressed, but OK it's spose to keep turning, so maybe I'm missing something.

Paddy the Voltage drop is a problem and there's no wires to the fuel pressure regulator.

But it's that two second bit that does a blokes head in, or does mine anyway.

Am I over-thinking this??  Tell me if I am.  I just can't explain why the fuel pressure test showed the results it showed if we assume that the FI circuit is working properly.  Can the results of the fuel pressure test be attributed to a failing FPR??  How would the FPR know the exact second you release the start button so that it allows the pressure to rise and fuel to flow? I don't know enough about mechanicals to answer that question.

Well I promised myself ages ago to stop posting on this thread cause I'm underthinkin something way over my head.

Off we go, this is a good one.

I think the fuel injection circuit failed the test cause the Voltage drop disappeared when it was removed.

Spose something depending on the injection circuit is shorting so then  the pump won't work. When the button is released, the fuel injection computer shuts off something (computers is a bit I know nothing about) so then the short has disappeared, then the pump works for the last two seconds.

Could it be?

Maybe its consumers can be tested for a short, maybe the AFM,  like Chris mentions.

I just noticed that the feed from fuse #6 goes to the temperature relay which has a connection to the infamous BMW anti-theft system connector.  Is it possible the temperature relay is faulty?

Or is it possible that this bike has some remnants of this system on it somewhere???  I can recall several starting problems that were traced back to the anti-theft system a few years back.  That system killed the engine somehow to immobilize the bike.  I think most of those systems were removed when they started failing and preventing the bike from running. Unfortunately, I can't find any documentation on the anti-theft system to tell how it worked.

I'll guess a short there would still be there for the last two seconds.

Klaus, I think, I'd try Point Seven Fives injector test. It's a quick easy check. Although the injectors coil should be isolated, it might not be anymore, since it's been in a fire.

The anti theft plug is white, four pin, and usually nothing is plugged in it. It's in the relay box.

I found this as a section of a thread on the BMWMOA site.

because of the age of the bike and the unfortunate fire situation. I found these comments Interesting enough to warrant exploration.

in a nod to Chris and others... I thought maybe... temp sensor, afm, side stand switch etc.

K75s fuel pressure, etc

Because I bought this bike having been stored not running for several years as "pig in a poke", I had no knowledge of its mechanical history. As I earlier explained, I had major flooding when attempting to start it so spent much of my time (a couple of weeks) trying to diagnose its origin primarily suspecting fuel pressure regulator and/or injector problems. I was wrong and when I discovered that a multitude of wires in both of the harnesses were severed, I began to check schematic diagrams in an attempt determine where these conductors originated and terminated. Lee helped me with this quest for routing and predicted that wiring problems were the likely source of my fuel and ignition problems. Evidently, portions of the main and injector harnesses were compromised partly as a result of being shredded by the radiator fan. Specifically, the fan motor, temp sending unit, communication circuits between the fuel injector control computer and the ignition control box as well as both wires to the air mass sensor were compromised. I bought a used injector control harness and spliced non essential wiring for things not directly related to its ability to run such as the hazard flasher relay, horn, gear position indicator, etc. Eureka, it now runs and everything seems to work with the exception of rear turn indicators, hazard lights and the horn but I'll keep working on those thing.  

Thanks VERY much to all of you for your time and tech assistance with special thanks to Lee who should write a book.

Here y'go......

https://youtu.be/kwkpuXxzjhk


Edit: okay, it's only the madass, but the FI wiring/system is the same, trust me 

Any progress?

1985 k 100 ecu available on ebay.

My apologies for the delay! I had promptly ordered a BMW FI relay off of eBay back on the 16th, but it was coming from Cali and eBay seems to think it might be mired down somewhere due to the mass of foul weather between there and here. So... today I ran up to a nearby junkyard, as had been recommended, and I stuck my head into a bunch of BMWs and I found a couple of those light green relays (specifying the color was quite useful, as there were blue and orange relays that had the same or similar connections, but different internal wiring). I just stuck that relay into place and gave it a try. Video shared below. I think this constitutes an improvement, but not a full resolution of my problems. I can hear the fuel pump kicking now as soon as I engage the starter, and the fuel pressure gauge confirms as much. I'll take a look at the fuel injectors and test for grounding. 

Yes, the replacement relay is working now.  The fuel pump is doing what it supposed to do.  One problem solved.

I unclipped the fuel injector harness from each fuel injector and I checked continuity in a few ways. I checked across the connections of each injector, and I got a nominal value in the 14 or 15 ohm range for each. I then checked between each injector lead and the fuel rail and I did not find any continuity. I additionally checked between each injector lead and the frame ground, and there was not any continuity there either.

Electrically, your injectors appear to be okay. Time to run some carb cleaner through them and replace the filters and o-rings if you haven't done so already.

You might want to do the screwdriver test again.

If you don't get the tick, tick, tick, here are a few tests you might want to do to check the positive and negative supply to the injectors, using a computer safe 12 Volt LED test light.

With everything back again, unplug an injector harness from an injector and connect the test light clip to battery positive. With ignition on, engine off, the light should glow at both sockets. While cranking the light should remain off at one socket (positive to positive) and should flash on and off at the other, that would be the negative. The negative is the right side in the photo (with the probe inserted in it)

It might sound unbalanced during this temporary test, since one cylinder is temporarily out of action.

DO NOT do this test with a normal test light.

daveyson wrote:
With everything back again, unplug an injector harness from an injector and connect the test light clip to battery positive. With ignition on, engine off, the light should glow at both sockets. While cranking the light should remain off at one socket (positive to positive) and should flash on and off at the other, that would be the negative. The negative is the right side in the photo (with the probe inserted in it)

From the previous test I still had all four injector harnesses disconnected, so I figured I would test them first. I clipped the test light (computer safe) to battery positive. I turned the ignition to the on position. I then probed both sides of all four connectors. For each one, one side lit green (indicating negative) and the other side didn't light at all.

When I did this test a few months ago, with ignition on, engine off, the light went on at both pins. Looking at the photo which pin didn't glow, left (positive) or right (negative)?

If your test light is like mine the colour indicates the direction of flow, so if you reversed the probe and clip, it would glow red instead of green.

Photo documentation!  Looks like, with the ignition on, engine off (obviously), test light connected (via red clamp to battery positive) I am getting a "negative" light in the left side of the connector, which was the positive side, and no light at all on the right side. 

Looks like you have an earth failure.

Does your injector harness have an earth connection just in front of injector number four? Mine does.

Another attempt to attach photo.

Again

I'm uncertain what you're trying to prove with this test. Even with the ignition on, you're only powering fuse 1, the ICU and coils, and not the ECU.  There are no grounds on the injector except through the internal components of the ECU.  The ground fault is intentional because the ECU is powered down and there is no path to ground through it.  There might be a path to ground through the mass airflow sensor and that is what you are measuring.  As for the positive side, there is also no continuity to battery positive at the injector plug with the ignition on but the engine not cranking.  

The only valid test of functioning injectors is the screwdriver test with the engine running or bench testing each injector.

Post #1: "The air flow meter is new me,"


What do you mean by this?

I wasn't thinking of these as injector tests, but power and earth to the injectors. When I did these tests I was surprised by the ignition on, engine off results, but not surprised by the results at idle. But those were the results I got with a functioning brick. Since Klaus's results at earth were different, I was guessing they're wrong.

The test results made me guess that the ECU denies earth intermittently while it's functioning, and has constant earth while off, but again computer stuff is way over my head.

Another thing I'm vulnerable to is wishful thinking. When the injector harness was replaced, I'm hoping the earth connection was overlooked. The earth wire meets the injector harness at the same point as the plug wire for injector three. Hopefully it's that simple.

chris846 wrote:Post #1: "The air flow meter is new me,"


What do you mean by this?

I'm not saying the AFM is defective. I'm saying that the tests might not actually be proving the injector circuitry because of leaks to ground through the way the AFM is connected possibly.