[hfarthington]

I think I have reached my wit's end on this seemingly unsolvable issue; my 2001 OBW VDC has a clock backlight that is not illuminating.

The background:

When I became the owner, there were additional lights out in the cluster so I decided to replace ALL of the lights in the cluster. However, I noticed the clock backlight did not come on. So, I checked all pertinent fuses and pulled the bulb. Tested bulb. Bulb is good. Examined contacts on cluster. No visible damage. Checked continuity between various points on board and between contacts when bulb was twisted in and found no issues. Reinstalled and still no illumination
I kind of gave up at this point because, hey, it isn't the most serious of issues and the vehicle needed and work to make it seaworthy.

At some point in between this I am spinning around in the snow and the clock light illuminates (no idea if spinning in snow is relevant). Okay, so now it's taunting me. Eventually it goes back out one day, who knows how long it lasted, a few hundred miles maybe if that. Anyway, now I wanted to fix it even more, especially because it's annoying to look at a dash with a dark rectangle in it. So, I pulled up a wiring diagram from a library resource (I think it is Haynes) and became disheartened because it seemed tightly integrated into other working circuits that I didn’t know where to begin. I had to do the other work on the car so I gave up on this until recently finishing the major work. Right before I parked the light had randomly turned on again for a little bit. This was after I undid everything again (with my lacking of electrician’s skills I have to revert to a caveman style of simply unplugging and plugging things back in). I do all of this with the battery disconnected so I have no idea if it was simply another random occurrence.

Notable observations:

A previous owner or mechanic has tapped into a few wires and electric taped over the exposed wires. I tried for hours to teach myself how to understand Subaru’s FSM wiring schematics (power distribution, ground distribution, etc.) and I determined that there are only two wires that more or less directly go into the instrument cluster (it seems that every wire is connected in this vehicle, which makes sense if you think about it : ). One of them is found on SMJ i1 (of which I cannot find a FSM page that details each pole’s color and what system it goes to, so that was kind of a dead end for me. The one wire I did manage to figure out is found on connector i5, which connects to “F/B” (I think this means front wiring harness bulkhead wiring harness). Anyway, B2 is the wire on schematic WI-20 and goes to FB-27 (“Combination meter, Front fog light switch, Headlight leveling switch, Illumination light, Rear fog light switch”)
The FSM has neither the bulb for the clock backlight nor the odometer backlight, but the one I found at library shows both the bulbs.
All of the other dash lights work. When the turn signal is engaged there is a slight dimming of the indicator lights, like “brake” or the check engine light, and I’m not sure if this is normal. It doesn’t do this when just the hazards are on.

Recap of what tests/work I’ve done:

Replaced bulb. Swapped bulb. Swapped base (plastic twisty). Tested all bulbs when doing so. Checked combination meter grid for damage. Checked combination meter wires and connectors for damage. Checked continuity when bulb is in slot. Unplugged rear brake, turn, and rear backup light harnesses. Checked all chassis ground locations inside. Also, all engine grounds were cleaned (this is especially easy to do when the engine is out). All of this was done with no change and left no further clues.

Final thoughts:

Perhaps I am foolish to chase after an electric gremlin for something so petty. Perhaps I am overthinking this and the solution is much simpler than I think. Bottom line I don’t know how to navigate the schematics and I don’t know where I should be checking for continuity, ground etc. I think I have done all of the basics a layman can do. I have to figure out why my transmission buzzes in gear, why the rear defogger works only on a few lines, and why my mirror heaters aren’t turning on—although there are a good posts I have read and need to read on those issues before I post a question. This backlight issue is one that I can’t find any good posts on and I really could use your help. If you made it this far, thank you for reading.

 

[plain OM]

The backlight bulbs for the clock/outside temp and the odometer/trip meter are related.

My 2001 wiring diagram shows them; it's essentially the same as in your "Library Schematic". One side of both bulbs goes to a ground. (On your library diagram, it's the upper sides of the two bulbs that come together at a vertical line. That line is the ground). The other side of each bulb goes to a "dimmer circuit".

If the odometer/trip meter backlight works, then we can conclude that the common ground (the vertical line) is good. Moreover, the common dimmer circuit, as far as the odometer/trip meter backlight is concerned, is also good. That collapses the possible area of the problem. It has to be in the combination meter (i.e., not in the wiring under the dash etc.) and in the clock backlight circuit only.

The combination meter uses a printed circuit to provide the connections between components. The printed circuit is on a flexible material. It's possible that one of the traces (the "printed circuit" wires) for the clock backlight is cracked, either between the "dimmer circuit" and the bulb, or the bulb and the ground circuit.

I don't know what the dimmer circuit itself contains, but suspect some electronic components such as resistors and/or semiconductors (transistors). Again, if these components are soldered to the printed circuit , there could be a cracked or deteriorated solder joint that's causing the on/off symptom, or it could be a defective component (a transistor can develop an internal bad connection, often heat sensitive). If the problem originates in the dimmer circuit, it would be a trace or component that's related only to the clock backlight and not the odometer backlight because the latter works.

A cracked trace or solder joint might not be obvious but sometimes when viewed with strong magnification the crack can be identified.

 

[eagleeye]

The backlight bulbs for the clock/outside temp and the odometer/trip meter are related.

.

just a FYI on 2000-2004 cars,

In my H4 2002 car the temp is located there with the clock in the gauge cluster.

but in my H6 2002 with the auto climate control, it is inside that auto climate panel. (on a button you have to hit to make it show for a few seconds, or hold to leave on).

 

[hfarthington]

The backlight bulbs for the clock/outside temp and the odometer/trip meter are related.

My 2001 wiring diagram shows them; it's essentially the same as in your "Library Schematic". One side of both bulbs goes to a ground. (On your library diagram, it's the upper sides of the two bulbs that come together at a vertical line. That line is the ground). The other side of each bulb goes to a "dimmer circuit".

If the odometer/trip meter backlight works, then we can conclude that the common ground (the vertical line) is good. Moreover, the common dimmer circuit, as far as the odometer/trip meter backlight is concerned, is also good. That collapses the possible area of the problem. It has to be in the combination meter (i.e., not in the wiring under the dash etc.) and in the clock backlight circuit only.

The combination meter uses a printed circuit to provide the connections between components. The printed circuit is on a flexible material. It's possible that one of the traces (the "printed circuit" wires) for the clock backlight is cracked, either between the "dimmer circuit" and the bulb, or the bulb and the ground circuit.

I don't know what the dimmer circuit itself contains, but suspect some electronic components such as resistors and/or semiconductors (transistors). Again, if these components are soldered to the printed circuit , there could be a cracked or deteriorated solder joint that's causing the on/off symptom, or it could be a defective component (a transistor can develop an internal bad connection, often heat sensitive). If the problem originates in the dimmer circuit, it would be a trace or component that's related only to the clock backlight and not the odometer backlight because the latter works.

A cracked trace or solder joint might not be obvious but sometimes when viewed with strong magnification the crack can be identified.

Thank you for your response.

I have had the cluster out two times at least. I will take it out again and again re check for microscopic cracks. . .a simple swap of a spare cluster could answer this aspect of it, but, I don't have a spare lying around. I can also do a comparative resistance measurement between the contacts of the odometer and the clock lights, instead of simple continuity test. Or better yet, a resistance test between the ground point on the chassis and those two lights . . . I think?

Isn't the "dimmer circuit" essentially the components that are dimmed when the control module (combination switch) actuates a change? If it was the dimmer circuit would the dc be in play when the headlights are switched off? Remember that these two backlights are on when the headlights are switched off and then when headlights switch on are subject to the "dimming" setting. Ref: http://www.subaruoutback.org/forums/80-electrical-electronics/93505-dash-lights-don-t-work.html
@eagleeye yes, thanks, that is important to state. Mine is like that too. It's in the climate control section, I think it's called "Auto A/C". Plain OM is referring to my attachment diagram's nomenclature I believe. Interestingly though, as far as circuits go, according to the FSM copy I found online 2001MY(H6) the outside temperature display system is connected to the combination meter. . .

 

[plain OM]

Or better yet, a resistance test between the ground point on the chassis and those two lights . . . I think?

Yes. Also, because the odometer light works fine, the only ground verification that's actually needed is that the ground side of the clock light has continuity to the ground side of the odometer bulb. Remove both bulbs and measure between the ground sides of their contacts on the board.

Isn't the "dimmer circuit" essentially the components that are dimmed when the control module (combination switch) actuates a change? If it was the dimmer circuit would the dc be in play when the headlights are switched off?

Yes. The two backlight bulbs should be on whenever the ignition switch is turned on. The clock and odometer are liquid crystal displays, and have to be backlighted to be seen.

When the headlight switch is Off, the bulbs are supplied the full system voltage (nominally 12 V) through the dimmer circuit; The bulbs' opposite sides being connected to ground, they would be at full brightness.

When the headlight switch is turned to either of its On positions, the dimmer circuit is switched by voltage from the tail & illumination relay so that it limits the current to the bulbs in accordance with the setting of the variable illumination control on the headlight switch.

This is shown in the "library" diagram, but perhaps is better reflected in the attached diagram. On the first page the dimmer control at the upper right is powered from the line extending between A and B. "A" is powered when the ignition switch is at On. The 12 V is passed through the dimmer circuit to the bulbs, with their lower sides connected to E, which on the next page goes to ground, so the bulbs are brightly lit.

When the headlight switch is at one of the two On positions, 12 V from the tail & illumination relay comes to C on page 1. This causes the dimmer circuit to switch from a simple pass-through to a current controlling function in accordance with the illumination control (page 2), to which the dimmer control is connected via "D".

 

[TheBlueSubaru]

Depending on the nature of electrical gremlins, the troubles they cause may not be as innocent as you believe them to be. They may result in larger systemic issues or be the result of larger systemic issues or some combination thereof.

Modern cars have a local area network for power and control signals, and a lot of that is serial in nature.

The theory that you've got intermittent circuitry via a crack in a board or a thermal issue in old components is a good one to check. Your best bet (though it sounds hard to get the part) is to replace the cluster with a (presumably) known-good one. As old as your car is, likely that'll be a scrapyard part. It sounds like you've done as much as reasonable outside of that.

 

[hfarthington]

Thanks. I'll update later today with what I find.

 

[hfarthington]

"Remove both bulbs and measure between the ground sides of their contacts on the board."

Done. With bulbs removed I get no continuity/infinite resistance. Connected I get continuity at 24 ohm. A bulb's resistance is 12 ohm, so doesn't that make sense that by connecting the bulbs I bridged the open circuit?

Other measurements on the same points (perhaps pointless):

With clock bulb only in I get 1.5 MOhm ha
With odometer bulb only in I get infinite.

I await your wisdom.

 

[hfarthington]

Should also note that I don't know if I was actually measuring ground....

 

[hfarthington]

Please view the video I made that shows the points I measured.

 

[plain OM]

Good video, thanks!

Done. With bulbs removed I get no continuity/infinite resistance.

If I can summarize what I'm seeing near the begining of the video:

(In the following "continuity" refers to the meter buzzing. Depending on the meter, it will buzz when there's a resistance of less than a preset amount, such as 100 or 200 Ohms. For this purpose, it's just an indication of a connection; the actual resistance reading will be more significant.)

0:02 bottom to bottom, no continuity.
0:07 bottom to top, continuity, "no resistance pretty much"
0:18 top to bottom, no continuity
0:20 top to top, no continuity

So the only measurement with very low resistance was bottom to top. If that was essentially zero Ohms, then that's probably the common ground. But, granted, it's not certain, yet.

On a board like that the printed circuit around the outer edge is usually ground. At around 1:12 we can see a wide band of green at the upper left. That should be the common ground and it will be continuous, as a wide or narrow trace at many places on the board. Find a solder joint that's on that expanse, and you should have a ground reference. Now, with both bulbs out, there should be continuity to that ground from one of the two bulb contacts at each bulb. If it's the bottom, and top, points as at 0:07, then that's the ground and both bulbs are connected.

If the ground checks out, then the next step would be to start out at the non-ground (i.e., high) bulb connections at the board. According to the wiring diagrams, each of these connection points goes, apparently separately, to the dimmer control. Consequently, starting at each "high" contact, follow the traces and compare the two as you go along, e.g., measuring the resistance from the bulb contact to the next point in each case. They should be identical. (At some point they might reach an integrated circuit that can't be traced further, but let's see.)

That board appears to be two-sided, and it might be difficult to follow the traces. (I suggest against disassembly of the cluster to access the other side.) However, given the time (and it appears, interest) you've already invested, there probably isn't much to lose in giving it a try.

I've attached some annotated clips from the video that might help clarify what I'm suggesting.

 

[hfarthington]

Thank you very much. I will post my findings. I have been musing about this and had an idea---what if I were to wire the odometer light bulb directly to the contacts of the clock bulb and do it in such a way as to insulate any contact on the circuit board on the clock side's bulb socket and the wire connections? 1st concern: Fire<Heat from excessive, unchecked resistance. To which I say, wouldn't the risk be relatively low because all I'd be doing is mimicking the circuit anyway? Also, wouldn't the circuit still be protected by a fuse (actually two fuses I think) since it's being mimicked? 2nd concern: Causing other parts (much more critical ones) of the combination meter (or car) to fail/perform incorrectly. Obviously a fire would realize this second concern.

My first choice is doing it the right way unless I find a functioning, safe hack. Maybe I'm crazy and foolish for even thinking this idea could accomplish what I desire BUT if this were to be relatively low risk then what would be the safest way of setting it up? Just tell me I'm crazy and to shutup and I will snuff this idea. Don't worry, I'm going to get you that data before I resort to replacement or hacks.

 

[plain OM]

Well thought out.

My concern is not knowing where the power for the odometer light is originating. If it's in some discreet components that could handle, essentially double the current, then sure. But if it's some sort of large scale IC, where each bulb has it's own distinct internal circuit with relatively low power capability, I'd be reluctant to do the hack and risk damaging the IC and other functions. That, in part, is why I suggested trying to trace those lines back to their sources. They're drawn separately in both the "library" and Subaru diagrams, so there must be a good reason. And, in doing so, it's always possible that at the source we find that the two are the same; there's just two traces etc going out from one point to the two bulbs, in which case, hack away!

 

[hfarthington]

"On a board like that the printed circuit around the outer edge is usually ground. At around 1:12 we can see a wide band of green at the upper left. That should be the common ground and it will be continuous, as a wide or narrow trace at many places on the board. Find a solder joint that's on that expanse, and you should have a ground reference. Now, with both bulbs out, there should be continuity to that ground from one of the two bulb contacts at each bulb. If it's the bottom, and top, points as at 0:07, then that's the ground and both bulbs are connected."

Okay, I followed the solder points as far as I could on both sides. Again I made a video to help see.

https://youtu.be/_oheHQP6ETk

 

[hfarthington]

I couldn't trace very far on the supposed power side for similar reasons to the ground side. I just kinda came to "dead ends" that forced me to guess the next connection point. I attempted to disassemble the cluster but stopped at pulling the gauge needles. I figure I probably need a special tool.

I did my "hack" anyway--**DISCLAIMER** Do not attempt this without accepting the possibility that you or your loved one may suddenly find their instrument cluster or dash burst into flames while driving--
Mostly because of curiousity and in hopes to move on to the next issue. Of course illuminated the clock light. I directly soldered wires from one bulbs poles (the bulb to go into the clock side) and took extra precaution to insulate the poles of the bulb within its plastic socket with shrink tubing. I just soldered wires underneath the copper contacts that contact the board on the odometer bulb.

plain OM you take me for someone who wants to get to the bottom of things so this may disappoint you. I definitely am so appreciative of your help here.

I haven't conceded entirely yet, so if there is another test I can perform I'll try it while I still have cluster out. I imagine I'll have it out another few days until my bulbs arrive for the radio illumination--Amazingly tiny incandescent bulbs! I'm having a professional install those as they are soldered in place on the board.

 

[plain OM]

In the latest video, around time 1:26, there's an identified ground point ("GND") about an inch above the test probe tip. Didn't spot that earlier; it would have been helpful initially when checking the ground side of each of the bulb connectors (could have saved the last step/video). But no matter, your previous tests and the multiple points of continuity in the second video confirm that the ground side for each bulb is intact.

The power side is different because we wouldn't expect the power side of one bulb to have continuity to many other points (as is the case with a ground side), especially if each bulb is supplied power separately as suggested in the wiring diagrams. This does make tracing the bulb power side back to it's source more challenging, but it should be doable.

As for the added wiring, actually just the power side wire would have been needed; we know the ground side contact of the clock bulb is good. I'd suggest fastening the wire(s) along their lengths (e.g., tape to the back cover) so that they are less subject to vibration.

I haven't conceded entirely yet, so if there is another test I can perform I'll try it while I still have cluster out.

I found a frame in the first video (around 1:16) that shows where the clock bulb power contact goes, at least to a point. I've attached an annotated version to provide an idea of how this could be delved into further. In this regard it looks as if the power for the clock bulb (off the right of the frame) comes from one or two transistors, the ones immediately above and below solder point "A". A failed transistor (if that's what they are) is a possibility.

When searching the first video at the far right of the board, there was one frame at around 1:48 that was clear enough to trace the odometer bulb power contact to a pair of three solder joints each that could also be transistors; the similar configurations of two (possible) transistors related to each bulb could be the dimmer controls. But that's just my hypotheses at this point.

There are ways to use the DVM to identify transistors and check for defective ones (open or shorted). Because there seems to be a similar pair for both bulbs, there's also opportunity for comparison between the working and non-working circuit. Granted, it is "getting to the bottom of things", and I guess it all depends on how far you want to go with this, or if you're satisfied with what you have now. (I have to note that even if a defective component were to be identified, there's no assurance that a suitable replacement could be found.)

That said, you've done well so far, and I would certainly understand the interest in just getting it functional again, and moving on.

Incidentally, I believe the four main cluster illumination bulbs are 161's. Are the clock and odometer bulbs the same?

 

[hfarthington]

As for the added wiring, actually just the power side wire would have been needed; we know the ground side contact of the clock bulb is good.

I understand, but the way the metal contacts rest in the base made it more likely to short out on itself, as using the board would necessitate leaving a contact in. Simpler and safer was my thought there.

I'd suggest fastening the wire(s) along their lengths (e.g., tape to the back cover) so that they are less subject to vibration.

:29:

There are ways to use the DVM to identify transistors and check for defective ones (open or shorted). Because there seems to be a similar pair for both bulbs, there's also opportunity for comparison between the working and non-working circuit. Granted, it is "getting to the bottom of things", and I guess it all depends on how far you want to go with this, or if you're satisfied with what you have now. (I have to note that even if a defective component were to be identified, there's no assurance that a suitable replacement could be found.)

That said, you've done well so far, and I would certainly understand the interest in just getting it functional again, and moving on.

Yeah, I realized when I started on disassembling the cluster that I wasn't willing to take the increased risk of destroying a 98% good cluster (assuming it is the cluster that's bad and not something farther upstream), so I chose to risk the cluster, dash and in extreme circumstance the entire car itself later instead of now hehe. Obviously if I were to commit a proper fix I would have to remove the CB.

Is it possible to ID the transistor’s poles from the back side of the board with just contact points? It's evident that this thing has circuitry on the other side unbeknownst to the outside word. I'm all about getting to the bottom of things, but as you know you are sort of living vicariously through my actions right now on this troubleshoot, which is totally fine by me as I am learning. I do have a diode test function on my meter which I think we'd use to test transistors too.

Incidentally, I believe the four main cluster illumination bulbs are 161's. Are the clock and odometer bulbs the same?

I found some blue tinted 194s for the illumination bulbs and all of the other lights are Toshiba V-2 T5 Black Socket with TSD 74 bulbs, both of which I found at my local NAPA. The toshibas require a little bit of perfect aligning for most of the slots, I think it depends on the microscopic differences of the board and the plastic moldings. . . and by the way at the beginning I tried swapping known good OE bulbs around the clock spot, in case anyone thinks that may be something.

I've attached an annotated version to provide an idea of how this could be delved into further.

I've attached a reply to your annotations. To summarize, there is continuity from the "power" contact through the first point you marked up to "C" but NOT through to "A", which measured 33 ohms.

 

[hfarthington]

I also wanted to note that the dimming function with my hack in place makes the two lights (clock/odo) much dimmer than normal (my guess would be exactly two times dimmer ), but with dimmer circuit not in play, so to speak (headlights off), it appears as normal brightness. And I know normal from owning this exact same model before. Could this be related to how the transistors govern the dimmer current? It must show at least that each bulb has an independent dimming circuitry, right?

If I understand Ohms Law, decreased illumination is simply because when, from normal operation, the dimming circuit lowers the voltage and also the amperes resulting in lower light output (less power). But now there is literally double resistance resulting in doubly less amperage on that side of the power circuit, which is why I think that the increased dimming shows that each power circuit has its own transistor gang, at least for these two bulbs, and also supports your hypothesis.

 

[plain OM]

The 33 Ohms could be a resistor in behind.

And, your interpretation of the dimming effect is essentially correct. The one circuit is handling double the current load (two bulbs, in parallel). (I believe the brightness is varied by controlling the current. For each setting of the driver's variable brightness control, the circuit delivers a specific current, or at least within a range. based on the original design and specified bulb. With two bulbs, that pre-determined current is being divided between them, so both are dimmer.)

The DVM diode test function can be used to identify a likely hidden bipolar transistor. It's more complicated when the subject "transistor" is in-circuit, and any measurements at its terminals can be affected by the other components connected to those terminals. Usually, when in that situation, I would remove the solder at two of the three connections, and measure at the then free wires coming through the board from the component on the other side. With two of the three connections of a transistor isolated, measurements between all three can identify a bipolar transistor, or determine that what's behind is not a transistor. (In the latter case, as elsewhere, comparison with the odometer light control circuit could be helpful.)

 

[hfarthington]

@pOM

Thank you so much for all your help in this. I learned not to be so afraid of circuit boards

I am not skilled enough with an iron/pump to remove solder from a board without scorching it. It's right up there on bucket list, right with getting my amateur radio technician license.

I think it's fair to say we need to move on. I'll be seeing you on the boards. Thanks again.