Learn me troubleshooting loose electrons

I'm working on the 1990 Saab and have a couple electrical problems that I just can't quite wrap my head around.  I'm starting with a bunch of unknowns, and trying to work my way backwards.  I've got four things that I know are wrong right now:  The ignition switch, the fuel pump, the fuel level sender, and the low brake fluid indicator.  I can't quite grasp how they're interconnected - and maybe they're not - but I need some help thinking this through.  To be fair, I haven't had time to really go and do a lot of troubleshooting yet - this is a family project and my daughter hasn't had time yet this week to go over to the shop with me, so I've left it alone other than looking over the wiring diagrams and trying to think about this.

Let's start with the brake fluid light.  With all the wires disconnected from the ignition switch and kept insulated from each other as soon as the battery is connected this indicator light comes on very dim.  It's acting the way trailer lights act when there's a grounding issue on the trailer.  According to the wiring diagram, this circuit should only be powered through fuse #22 when the ignition is turned on.

The ignition switch is disconnected but I'm bypassing it by clamping together all of the leads that are present with the exception of the lead that goes to the starter.  According to the service manual, having all of those wires connected is the normal function of the ignition switch in the "run" position.  To actuate the starter, I'm touching that last lead to the clamped conglomeration of wires and it does crank the starter as expected.  In the run position, the normal dash lights come on - along with the dim brake fluid indicator - and things seem "normal".

The fuel pump does not seem to be getting any voltage.  According to the service manual, it only powers the fuel pump when the engine is turning but my test light is dark and the digital multimeter shows only milivolts across the fuel pump connector terminals during cranking. 

Finally, the fuel level sender does not appear to be sending.  Despite having 5 gallons of fuel in the tank, the gauge is all the way down and the fuel light is on.  According to the service manual, I should have plenty of fuel in the tank for the light to shut off.  Since this car is new to me, I have no idea if the fuel sender ever worked.

The fuel pump and the level sender both share a grounding point in the rear of the car.  The other things on that grounding point do work, however.  The tail lights are on that ground and they're quite bright.  I haven't traced the wires for each yet to see if they have a common point before the ground point.

Having the ignition switch unplugged shouldn't be introducing any weirdness and having those wires isolated should be keeping the entire gauge cluster dark.  So that dim warning light is puzzling to me.  I am assuming that something is backfeeding voltage into it, but I can't quite think about how to go about finding it.  My next troubleshooting step for the fuel pump is to go upstream and jump the relay to see if it runs so I can determine if the problem is that the computer isn't turning the pump on.  If that doesn't turn the pump on, I was going to try to attach an additional temporary ground to see if that gets the pump to come on and/or gets the fuel gauge to work.

Troubleshooting the fuel pump is pretty easy, and I'm not too concerned about that.  But what should my first step be with that light on the gauge cluster?  That one really has me puzzled.

First step is to get a wiring diagram and look at it.  Not really in a modern art chin-stroky way but as a road map.

European cars often have atrocious wiring diagrams.  (Looks toward the BMW based looming in the distance)

That brake fluid light: Where does it get its power and where does it get its ground?  What other devices share those? (Illumination circuit, other gauges, etc)

When I am done beating my head against the Mini for an hour I'll take a peek and see if the wiring diagrams are GM-beautiful or BMW-asinine.

First, as mentioned, find a wiring diagram and become familiar with how to read it.

Next, get a good ignition switch and connect the wires rather than haphazardly holding them together.

I would check the fuel pump relay, I always carried a spare in the glove box. 

Next would be the fuel level indicator. 

Last would be the m/cylinder fluid level indicator.

Inside the cluster, the brake warning light gets power from "15" (DIN for ignition feed) and grounds through what appears to be a pressure differential switch and then on to a ground at location 93 or 65 depending on chassis.  This ground is shared by... a lot of other things labeled only by numbers, like Volvo does, which is great for making multilingual diagrams but is annoying in every language.

Another thing is that before the ground leaves the cluster, the power goes through the light to the battery warning light, and thence on to the D+ terminal on the alternator.  So I'd unplug the alternator and see if the brake fluid light stops illuminating.  If it does, I'd chalk it up to normal behavior or an alternator issue to not worry about until the engine runs.

I do have the wiring diagrams from the Saab workshop manual and they are...  diagrams.

As much as I hate problems that just sort of go away on their own, I'm thinking that the behaviour may have been related to a very low battery.  Once I put it on the charger for a bit and hooked up the ignition, it stopped trying to light the brake fluid light.

So now I can focus on the fuel pump circuit and why that's not working.

Okay, still fighting something here and I'm way out over my skis on trying to understand wtf is going on here.

It started doing the dimly lit light thing again and the brake pump wasn't working so I started digging in to it.  The brake system is not vacuum-assisted.  It's a Teves Mark-II ABS systems which was also installed on some 88-90 GM and Ford products like the Reatta and the Thunderbird, as well as some Jaaaaaaaags and Range Rovers, some Alfas I think, and perhaps even a Ferrarri or two.  There's an accumulator that gets pressurized by an electric hydraulic pump and that pressurized fluid is used to operate the brakes.  There's a pressure switch that is supposed to monitor the system and when the pressure is below 140bar, it grounds out a pin that leads to the pump relay and when the pressure goes above 180bar, it opens that connection, turning off the pump.  When the pressure drops below 105bar, it grounds the "brake fluid" light on the dash as well as a pin on the ABS ECU to let the system know there's a problem.  "In Theory" the braking system should work completely independently of the ABS system while the ABS system will shut off if the pressure switch or the brake fluid level switch tells it there's a problem.

Things that I discovered:  The ABS system uses a little bit of a weird relay.  Turns out that it had been swapped out for the fog light relay which does not use a weird relay.  So both the ABS relay and fog lamp relay were energized at all times.  That's fixed now.

The ABS system has two diodes (303A and 303B).  I tested those with my multimeter and they were failed open in both directions.  I replaced them with two new diodes and tested those diodes for proper function beforehand.

Someone, at some point, was chasing this issue and, I assume, never found it.  Probably what caused the car to be parked.  The wiring loom for the pressure switch has been cut back, and some of the wires have been probed through their insulation.  There are a couple wires that appear to have been melted at some point (in the diagram below, the "source" side of the #57 connector that has one wire going to the gauge cluster and one going to the ignition relay).  It also appears that someone was trying to find the source of a parasitic draw.  My guess is that draw was caused by the relay swap - my tests show that with the relays in the right spot there's very little current flowing with the ignition in "park" and the lights out.  I do need to revisit that, though, to make sure it's as low as I think it was.  I know it was under an amp by a decent margin, but it is an older system, so its computers shouldn't be thinking very much when it's not running.

These repairs having been made, with the key in "drive", the brake pump should fire to life if the pressure is below 140bar and run until it hits 180.  In my case, it does not.  However, if you disconnect one of the +12V power sources - the ignition switch feeds 3 different circuits - the brake pump would run and the light would light.  The system looks like this:

I did try unplugging the alternator's D+ wire and there was no change.

I pulled out the 47E bulb for the charging light and there was no change.

I removed the turn signal fuse from the fuse box.  According to the gauge cluster wiring diagram, that seems to intersect with some things that touch the +15 circuit in the cluster.  Suddenly the brake pump comes on when you put the key in "drive".

Putting that fuse back in, I unplugged the two electrical connectors on the gauge cluster and there was a completely dark cluster and no brake pump at all.

With the gauge cluster unplugged, I unplugged the connector from the level switch (#299 in the diagram) and simulated a low brake fluid condition by jumping pins 4 and 3 on the connector.  That had the effect of grounding terminal #2 on the pressure switch (#294) to earthing point 300.  Now the brake pump will run with the ignition in "drive" and all the power leads connected.  It also runs with +15 disconnected, but looking at the wiring diagram that makes sense - the only part of the brake boost system that should be dependent on +15 is the warning light.  I have unplugged the ABS computer to take it out of the equation, since it isn't required to run the boost system.

But what doesn't make sense is that according to the service manual's description, the pressure switch is supposed to work thusly:

• Below 140bar, contacts 4-1 are closed to earth the coil of pump relay #293
• At 180bar, the contacts will open
• Below 105bar, contacts 2-1 will close to earth the 47F lamp
• At 134bar, contacts 2-1 will open
• Below 105 bar, contacts 3-5 will open, breaking connection between pins 9 and 10 of the ABS ECU (#291) alerting the ECU that strange things are afoot at the circle-k
• Under "normal operating conditions" contacts 3-5 will be closed

Nowhere in any of that do they say anything about a connection between 2 and 4.  And yet, that's the only path to ground that could possibly exist.  With pin 2 not grounded, the pump won't run and pin 85 is an open circuit.  As soon as the jumper is put in to place to ground pin 2, I have continuity to ground on the relay's 85 pin and the pump turns on.  The other odd thing that happens is that the pump runs for a little bit and shuts off.  But one or two good presses of the pedal, and the pump comes on again.  Unfortunately, I do not possess any way to test the actual pressure of the system other than to tell you that after the pump runs, I have a decent brake pedal so there's *some* pressure in there.  Also, if the pressure is up around 140, it shouldn't be lighting the light, but it is.  When I test the pins of the switch, 4-1 is always open.  I've tried pumping the brakes about 50 times to depressurize the system as much as possible and I've also checked it right after the pump runs.  It does not appear to ever be closing that circuit.

Also, when I plugged the gauge cluster back in, I still have (almost) no lights or gauges on the cluster.  There's a chance that I didn't get the connectors on correctly, I suppose - I'm working through the speaker hole in the dash and have to do the whole thing by feel - but when I put the right turn signal on, I get the indicator light on the dash and the fuel gauge needle starts dancing.  That sort of screams bad ground to me.

So...  bad ground in the cluster that fried the pressure switch and the ABS diodes?

Bad pressure switch and the swapped relays fried the diodes?

Bad pressure switch and me jumping the wrong pins on the brake pump fried the diodes?

The pressure switch is one of those hard-to-find parts that appears to be about $500 brand new or $200-some-odd to rebuild.  Ordinarily, I'd just try a switch and see what happens, but if some other problem that I haven't already fixed took out the switch, then that would be a pretty expensive troubleshooting step. 

I think my next thing will be to pull the fascia off so I can get the gauge cluster out and poke around on it to look for anything broken and to make sure that I have the connectors properly seated.  It's kind of odd that I suddenly have zero gauge cluster after "turning it off and back on again".  Am I wrong to think that if the problem was a bad ground on the cluster - and it was grounding through the pressure switch - that once that alternate ground was re-connected the gauge cluster would start working again?

That is a lot to try to process.

I am going to focus on one salient point: Removing one power feed from the ignition circuit will allow it to work.  Putting that power feed back in makes it stop working.  Yes?

If that is the case, it is going to be a ground issue.  There is a poor ground somewhere, possibly/probably at the comb at 300.  With a circuit disconnected, the pump is finding ground through that circuit's other components.  When you supply power again, the ground is no longer there.

Bad ground does not necessarily mean a dirty bolted connection, it could be a burnt/cracked/relaxed terminal in a connector.

Yeah, it's a lot.  I've been shuffling about with this giant sheaf of printed-out wiring schematics and pages with handwritten notes scrawled all over them.  I'm basically a half-step away from having a corkboard with yarn strung from one pushpin to another...

But, yes, that is exactly what happens.

I'll go back and verify the connections to ground on the connectors for the fluid level switch and the pressure switch.  According to my notes, it looks like those were showing no resistance to the battery terminal, but I'll make sure of it.  It shouldn't have a path to earthing point 300 - the fluid level switch should only be closing that connection when the float sinks and according to my continuity test, it's working correctly.

Continuity tests will work if there is one strand of a 30ga wire intact.  On a circuit that sees any kind of load, a voltage drop test needs to be done, with the system loaded (or an attempt to load).  The difference in voltage between the power side *of a device* and the ground side *of a device* should roughly be system voltage, minus maybe a tenth to a half of a volt for transmission losses.

Or if you want to be saucy, and I had been reading the diagram wrong anyway, the 2 pin of 297 (the fluid pump, right?) should be no more than about .5v from battery ground.  And when I said the 300 comb I think I meant the 93/65 comb.

Pete. (l33t FS) said:

Bad ground does not necessarily mean a dirty bolted connection, it could be a burnt/cracked/relaxed terminal in a connector.

Pete's comment reminded me of a time when I had an issue ultimately traced to a burnt pin in a bulkhead connector (152), driver footwell, left side. You may want to take a look at it after ruling out the instrument cluster. I can't recall for sure, but to get at the connector I think the knee bolster has to come out before you can peel back the carpet. The bolster is a bit of a PITA to re-install -- make sure the metal tongues slide up into their receiver slots, and be careful not to strip the weird middle bolt. In addition to the middle bolt, the bolster is held in place by a 10-mm bolt at either end, accessed from the engine bay; they may be covered with black waterproofing goop, and hidden behind wiring.

Oh yeah, testing for voltage drop!  I knew there was a better way to check the grounds as opposed to resistance, but I flaked on that one.  Thank you!  Something that diagram doesn't show that I'm going to have to figure out is where 93/65 is actually located and if the pressure switch is actually grounded there.  According to the FSM, earthing point 65 for the 1990 model is located under the back seat.  Seems a little odd to be earthing the whole system all the way back there.  And it seems like there's a lot of opportunity for that connection to cease existing.  If I'm reading things right, it appears that they run a separate wire to that point for each of the various components.  Including a wire to connect earthing point 300 to 65.  And there's a 2 pin connector (#59) in there as well which is an opportunity for failure.  While the diagram says that it's earthing point 93 for 1989 cars and 65 for 1990 cars and the pages describing the model year 1990 earth point 65 calls out te ABS and brake system, the textual description in the brake section of the manual says that things will get grounded to earthing point "63/93".  The last time I thought the diagram was wrong, the problem was actually that I had a funny relay, so I'm open to contrary positions on this, but there is no mention in the "earthing points" section of the manual of a 63.  There's a 93 and a 65, as the diagram shows, but I think I'm safe to assume the "63/93" is an error.

Thanks for the tip on the bolster.  They seem pretty good at hiding fasteners so knowing that I've got a couple that I need to dig around and find on the other side of the firewall should save me a bunch of time and a lot of frustration!

Just adding this to the conversation in case it helps anyone out there. Good electrical troubleshooting from Carl: Zap | How to easily diagnose most any automotive electrical problem.

I haven't had the time I wanted to get over and work on this, but I'm dropping about 6 volts between the ground on the pressure switch (that would be terminal #2 on item 294 in the schematic) and earth point 93/65 (which is actually 63).  We pulled the dash fascia off in order to get better access to the gauge cluster connections.  Since I was there anyway, I went ahead and checked the various traces on the cluster and verified that the diodes in the path to the alternators D+ terminal were in good order.

Continuing to stare at the diagram, though, terminal 2 is supposed to provide ground TO the gauge cluster in order to light the brake fluid light when the pressure is too low.  But instead, it appears that the switch is somehow connecting terminal 4 and 2 to try to get a path to ground THROUGH the gauge cluster for the pump relay (293).

I'm breaking my own rule...  when stuff isn't working and you know you have something broken, fix the known broken thing before looking for zebras somewhere.  I think I need to step back, get a pressure switch that isn't connecting the wrong terminals together, and then see how things look.  I'm sitting here trying to figure out what could possibly be grounding itself through the gauge cluster other than the part that I know for a fact is grounding through the gauge cluster.  That isn't particularly smart!

Okay, the new brake switch is in and a continuity test shows that it is connecting the correct pins on low brake system pressure.  So that part is good.

Voltage drops from the gauge cluster to the chassis ground, the alternator D+ terminal, and the parking brake switch are all negligible.

I unplugged the parking brake switch but the parking brake light still comes on.  Which is a little weird, because at one point in this journey the light wasn't coming on at all, now it won't go out.  With the cluster disconnected, the parking brake switch circuit is open, so the light isn't grounding through that wiring but through the diode and its connection to other lights in the dash?

Just to see if I could isolate it at all, I disconnected the ABS ECU, the parking brake switch, the pressure switch, the fuel level sender, and the engine ECU and pulled all of the fuses except the one that feeds the right half of the cluster.  When I turn the key, the right-side lights come on and start to flash, alternating with the right turn signal indicator.  My assumption is that this is telling me that something going through the cluster is looking for any path to ground it can find. 

I've got the knee bolster out (glad I was able to try that on a junkyard car first!) and I'm going to go inspect all the bulkhead connectors for damage.