90 Eclipse GSR4 repaired ECU diagnosis

Sooo, does the restoration of the 5 volts to the component side of C4 fix your problem? Did you have to replace the power transistor as you had suspected?

Nirodac

Hi,

first of all, thanks again for you help, Nirodac. I replaced C4 (with a 0.1 uF cap), cleaned the PCB underneath it and resoldered the contact points - that indeed restored the 5V to the rest of the PCB (and most of all the ICs). That did the job and the car is running again. I test drove it without any problems. After that I took the car for a longer drive at night (assuming it's fixed). At first I thought my eyes had gone bad ("Yeah, it *was* a long day, after all..."), because the headlights seemed a little dim. And the dashboard illumination did too. And even the clock looked less bright than normal... When the radio started turning itself off and refused to turn back on, I finally got the clue - the +12V now... Turns out I had been driving on battery all the way (with the lights on). Luckily, the engine seems to be the thing in the car that runs the longest with low voltage - stranding on an Interstate with no lights is not my favourite on a very rainy, dark Friday night. As you have guessed, the battery didn't get charged, because the alternator voltage regulator circuit is broke. As Mitsubishi did a

*superb* design on that car, the battery warning light is triggered by the alternator voltage regulation coil transistor - transistor broke, alternator voltage regulation coil dead, alternator dead, no battery charging, lots of load, battery about to die - and no warning light on! *sigh* As it looks, the regulator unit is not available by itself, so I've been looking around for a used or rebuilt alternator - will keep you updated.

Norman

Your welcome, glad I could help.

The service manual shows the voltage regulator as a separate item (inside the alternator) as part of the brush holder assembly. It also shows the diodes as a separate assembly. Some manufacturers actually sell rebuild kits that include these parts, and brushes and bearings as well. But certainly the easiest repair is the replacement of the entire unit. Good Luck.

Nirodac.

Yes, the IC regulator is a seperate part inside the alternator, although I expect it to be fully sealed, so it has to be replaced entirely. I suspect the main transistor to be broke, since that produces the error(s) I have now. I expect the diodes to be ok. I didn't know that they are seperated from the rest of the regulator, then again, that will probably not help me anyway.

I'd love to find a rebuild kit that's resonably priced. My current options are (US$ = 1,25 * Euro):

Junk yard alternator (working): 50 Euro Junk yard alternator (working, 3 months warranty): 80 Euro Rebuilt alternator from a part seller: 150 Euro

Now for the Mitsubishi prices: IC regulator only: 250 Euro Alternator, rebuilt: 270 Euro (! Just 20 Euro more than the IC regulator) Alternator, new: 400 Euro.

IMHO, the IC regulator is way overpriced. I'll probably go for a junk yard option. Maybe I'll take the regulator out of the junk yard alternator and put it into mine.

Norman

Ouch, that's pricey!!

Nirodac

Just finished putting the new clutch in my sons car, took it for a test drive, and the battery wouldn't charge. Dismantled the alternator, inner slip ring is all chewed up, the alternator ran out of contact brush material. This sucks, replacement cost (Canadian dollars) $276.00, plus taxes. There was no indication, other than a dead battery.

Nirodac

I feel your pain... My "new" alternator arrived today - it's a junk yard alternator for 80 Euro. Besides that it seems to me that the alternator warning light has been extremly bad designed by Mitsubushi - quite often it fails to go on by design.

Norman

Nirodac,

I can't begin to tell you how helpful the post below was.

My 1992 Expo LRV had symptoms similar to Bornish's. It cut out twice, then re-started each time after a few minutes. Then when we stopped for lunch, it wouldn't restart at all. We towed it home.

No spark. Could be no signals (TDC and or crank angle) from distributor, or could be in the MPI computer. Tried to read codes with my multimeter. No codes at all, not even the steady pulsing that indicates no codes. Checked out the control relay, and determined that the computer had power and ground.

Opened the computer up. Some crud around the 50 V, 47uF cap that's right behind the linear regulator 2SB1355 transistor. Powered it on the bench.

0 volts at all three leads of the 2SB1355. Removed the 2SB1355 and checked it out. It's good--at least it acts like a diode from base to emitter and anothe from base to collector.

I'm hoping that one of the electrolytic caps has shorted and has crowbarred the supply, but I'm not optimistic. Nirodac, have you analyzed this regulator circuit to any great degree?

Regards,

Larry

++++++++Original post follows+++++++++

Hi Bornish Just a curious question, what country are you in?

The following information applies specifically to ECU part number MD159561, from a 1991 Plymouth Laser (Eclipse), non turbo. I'm giving you this information because I believe it to be close to your ECU, if nothing else, it'll give you some insight into how the Mitsu ECU's are built.

Looking at my ECU from my 1991 Eclipse (Ply Laser) the green power transistor near the center of the circuit board is a 2SB1335A, PNP, Vceo=80V, Vcb=5V, Ic=10 Amp, and power rating is 50 Watt. It subs to an NTE

378. This is nothing fancy, it's rated an audio power amp transistor. The other four transistors are part number (all 4) 2SD1415, NPN Darlington, 100V, 7A, 30 watt. These transistors provide the ground to the injectors, when the ECU wants gas injected into the cylinder. By the way just incase you have different transistors, remember the Japanese drop the first two characters on the device number. Example 2SB1335A will be printed as B1335A on the transistor.

Check out the circuit trace at both ends of capacitor C4. C4 is located at the anode (non banded) end of the big diode in the center of the PCB. One side is the ground feed the other side is the 5 volt feed. There should NOT be a break in the trace on the +5 volt side. C4 is a tiny surface mount cap. Check very closely with a magnifying glass. Scrap off any protective coating if necessary. If you can power up the ECU with the cover off, you should be able to measure +5 volts at the center pin of the green (2SB1335A) power transistor, the negative of the meter is grounded on the anode side of the big diode in the middle of the PCB. This is the five volt regulator output. BEWARE, if you slip off this pin with your probe and short an adjacent pin,you will most likely toast the ECU. Remember that there may be a conformal coating on the PCB. You should be able to follow this trace right to the other end of the PCB and around to pin 42 of IC2, the second biggest chip on the circuit board. I think also that pin 43 of IC 4 (the big one) is also at 5 volts from this rail. There should be 5 volts at all points on this trace

If your processor is not getting the 5 volts it needs to run, it will not be able to pulse the injectors, or do anything else for that matter. You should replace ALL your filter capacitors just on spec. This is a real issue with Mitsu ECUs.

Remember also that this information refers to a specific ECU, while it may also apply to your ECU, it also may not.

Good luck, and keep us posted.

Nope, didn't spend any more time on this ECU than I needed. 12V is provided to this ECU on a number of different pins, have you powered all of them. Another poster had the same problem and thought it was the transistor, but I believe in the end it was an open trace. I found open traces on two of the three ECU's I've repaired for my son. Did you follow the instructions I posted. Check specifically for the break in the fat PC trace near the center of the PCB. That one broken trace caused the greatest problem. Mitsu (and other electronic manufacturers) place what looks like contact cement on some components to hold them to the PCB. My broken trace was under one of these patches of "glue". Make sure that a voltage measured on one side of this glue is also on the other side, most notably near the cap at the center of the ECU PCB. Try measuring the resistance with an ohm meter. I only reversed engineered the DC input far enough to fix my problem.

Good luck

Comments interleaved.

Larry

Just the main pair so far. You can see that they go to the fat trace that goes up the middle to the 2SB1355 used as a linear series regulator.

Wow! First for so many circuit card failures, and second for so many computer failures. I've had this 1992 since about 1994, and have 160K miles on it, and this is the first problem I've had.

Seems to be just fine, unfortunately. I can go from the pin to the

2SB1355 and way beyond it. I think I've checked +12V and ground, and +5 going out, but will re-check.

This is standard practice for surface mount technology (SMT). You have to hold all the tinned, fluxed components in place after they are placed, until the populated board goes on the conveyor belt through the reflow oven (which, of course, reflows the solder).

I've been using an ohmmeter. I'll try again and also trace the voltages with power on. It would be great to merely span some break with a few jumpers.

Do you remember back in the mid 1960s when PC boards just went into wide use, and hairline fractures were common? They didn't use conformal coating then (as auto manufactures do now) and the radio service techs simply laid a coating of solder along every trace on the card.

Well, right now my problem seems to be close to the power terminals, too.

Thanks, I'm counting on it.

Larry

Well, I was baffled by this for quite a while. The ohmmeter (one of those business-card-sized digital units) indicated continuity (beeps) on all the major traces, but I wasn't getting any voltages much beyond the where the connectors enter the card, when I powered it on the bench.

It finally dawned on me that the ohmmeter will beep with some resistance, perhaps up to 10 ohms. The resistance of the entire card (through all the ICs and discrete circuits, etc.) is lower than that, so I wasn't measuring along the +5V trace--I was measuring between 5V and ground.

It appears that the failure mode you observed is in play here, too. The trace seems to have failed under the big electrolytic cap nearest the connector, where all the brown crud was. I will put on new electrolytics. I'm ordering high-temperature (industrial/military grade) caps with -40C to +105C temp range, just as were on it. It doesn't make sense to use consumer grade caps in this harsh environment. And they only cost a few pennies more.

I'm reserving judgement until the new caps come and I do the replacement, but it looks like your description of the failure mechanism is right--the heavier components are cantilevered off the board, held by cement. Ultimately vibration and bumps cause them to pull the foil from the board and open it.

It amazes me that Mitsubishi was able to make this on a two-layer board. I expected a four-layer board.

Larry

Hopefully thats your problem. Although if the caps aren't leaking or bulging, you could repair the trace and fire it up for a quick test. How would you know it was working ? Check the diagnostic output pin for pulsed signal.

Good Luck

I replaced the four electrolytics today. There was brown crud under two of them. Either the adhesive they use is corrosive (as you suggest), or the electrolyte is leaking and corrosive.

While I had the big one out that's near the connector (50V, 47uF), I looked further for the open circuit. I never could see it, but I located where it had to be and repaired it. The main trace comes in with 12 volts on two adjacent pins. This trace goes through a T-filter. The next via on that 12-volt trace is a reverse-biased diode to ground, to protect against reverse polarity (car battery installed backwards).

Using a real analog VOM (not the beeper) I found continuity from the 12 volt connector to through the filter, but not to the diode. It looked like there was corrosion around the filter lead. So I took a short piece of AWG 18 wire and bridged from the filter to the diode and suddenly 12v and 5v appeared in the right places.

Had I thought to look here first, I would have also checked the diagnostic output. It's back in the package now, so I'll just install it and try it tomorrow and report back.

Thanks a lot, Nirodac.

Just to finish the story, I installed the computer today and the car started right up and runs fine.

It is more than wishful thinking that the car actually runs better than before the failure. I expect there may have been some noise due to caps going bad that was affecting engine performance above 4000 RPM. It now winds smoothly to 6000 (not something I do very often--I'm nearly 60 years old).

Nirodac, thanks a lot. If you see this, drop a note to the Hotmail account listed above.

Larry Marks

GSR4 =GSX?