Second Gen ECU / TCU Failure (An Overview)

[Cruiser]

Recently I experienced first hand a rather common failure for the Magna TR/TS. While heading home from class, we started to smell an unusual burning smell in the cabin, followed by some erratic behaviour from the automatic transmission.

After 22 years or 240,000 kilometres, the Transaxle Control Unit (TCU), commonly referred to as the transmission computer, was dead.

Perhaps I should have noticed the warning signs. The occasional rougher than normal shifting and downshifting, some late shifts and one occasion where the car lost power without warning, only to recover a few seconds later. Most owners wouldn't notice these warning signs, attributing them to an old and tired transmission. In some cases, there are no warning signs at all.

When the unit does fail, it shows common symptoms of a failed transmission. The car will lock itself in third gear, otherwise known as "Limp Home Mode", and moving between forward and reverse often seems rough or harsh. I've seen this misdiagnosed as a deteriorating or damaged transmission before.


My car is a TR-series Magna 4-cylinder 2.6L Automatic, assembled in October 1992. Mitsubishi used different variants of the Transaxle Control Unit depending on engine and model year, and I believe at some point in 1993 the TCU was revised to a newer 2-plug variant. Being an older model, mine still had the 3-plug variant of the 4-cylinder TCU, which is extremely prone to failure. For reference, the markings on my TCU are as follows:

ELC26M
MD750103
G1T11674E

MITSUBISHI ELECTRIC CORP.
JAPAN

These failures are attributed to a single component within the TCU - a 47uF (microfarad), 50v (volt) electrolytic capacitor, manufactured by Rubycon. After several years in service, these capacitors develop a leak from the lower seal, and the electrolyte inside leaks onto the circuit board, slowly corroding away the traces and connections on the surface. To make matters worse, electrolyte is conductive, so once it bridges several traces and connections on the board, it causes a short inside the TCU. Here's what mine looked like immediately after it failed:





Rubycon is a reputable capacitor manufacturer, so we shouldn't automatically assume that Mitsubishi used sub-standard components here. I suppose it is possible that these 47uF 50v capacitors came from a bad batch, but another likely culprit is the metal heatsink sitting directly to the left. Capacitors don't fare well in high temperature environments, and placing one directly next to a component that emits heat is an almost guaranteed recipe for disaster down the track.

I cleaned away that top layer of blackened and burnt electrolyte, but it was no use. The corrosion had completely destroyed the traces on the board. While it could be repaired by re-routing these damaged traces with wire and some electrical hackery, it was easier to find another TCU that hadn't suffered such extensive damage to work on instead.

Here's what the same component in a healthy and working TCU looks like:





This is where most people would stop. Purchase another used TCU that hasn't failed, install it and keep driving. However this unit, while it looks to be in good shape, will fail in the same manner. It's a case of when, not if.



IMPORTANT: Everything beyond here assumes some knowledge of soldering and electronics. If you aren't comfortable with this stuff, consider looking at professionally reconditioned TCUs / ECUs. They'll already have had this procedure performed and tested, and often come with a warranty. A TCU / ECU could be easily damaged if this process is not performed correctly, so some knowledge and skill is absolutely required. I'm documenting this process purely for reference.


Thankfully, capacitors are cheap. Even some decent quality capacitors from reputable retailers like Element14 are only a few cents each. So if you consider yourself proficient at soldering, repairing these units and in the process hardening them against future failures will be rather simple.

In addition to the three black and blue capacitors pictured above, I also replaced the small, yellow ones to the right:

A Tip: Taking pictures of the board before working on it definitely comes in handy, especially when trying to remember where components go and which way they should be installed.

Removing the old capacitors takes a little patience and care. I won't go through the definitive guide to soldering or desoldering here since there's already thousands of resources online showing how it's done, and honestly, 9/10ths of the technique comes down to practice. Personally, I have rather basic soldering tools, so I used a cheap desoldering pump to remove as much of the solder holding the capacitors in on the underside of the board as possible, then applied heat to the remaining solder on each pad, one side at a time, and gently rocked the capacitor out bit by bit.





Once it's out, clean the board. Seriously, it may look perfect to the eye, but if a capacitor has started to leak and left a thin film of electrolyte there, it probably wouldn't be noticeable but could cause damage down the track. Some cotton swabs and some methylated spirits, electronic cleaner or pure alcohol works wonders here.

Another Tip: The board is covered in a lacquer of some kind, and it's common for this to burn off while soldering or rub off while cleaning. This shouldn't cause any issues, but cans of circuit board lacquer can be had from Jaycar Electronics if you want to touch it up afterwards.

The small Plus symbol (+) on the board denotes the positive side of the capacitor. On traditional electrolytic capacitors, there is a stripe running up one side of the capacitor. This is the negative side, so the stripe should be on the opposite side from the positive / plus symbol on the board. Because I was replacing the existing capacitors with the same type, I simply referred back to a photo of the board I took earlier and positioned the new capacitors the same way.


Capacitor technology and miniaturisation sure has come a long way in 20 years. Pictured on the left is the original, on the right is the modern replacement. Both are rated for the same 47uF 50v and come from the same manufacturer, Rubycon.





The leads coming out of the capacitor are differently spaced on the smaller one, and this does present an issue when inserting them through the holes in the circuit board. Thankfully this is a simple fix - bend the leads to make them fit. Slot the capacitor into the holes in the board, and then solder it in.

Something I do have to stress here is that the quality of the solder joint in this application is rather important. The last thing anyone needs is to have a solder joint break and the TCU malfunction while driving.

The Magna TCU uses a double sided PCB, meaning the leads of the capacitor are connected to traces and components on both sides of the board. To ensure these connections were made correctly, I made sure to use a clean soldering iron, then I used a flux pen to apply flux to the holes on both sides of the board. Then I inserted the capacitor, touched the iron to the solder pad and capacitor lead, and fed a small amount of solder in. If done correctly, the solder easily flows through the board and makes a solid connection. Trim the excess length from the capacitor leads sticking through the board, and move on to the next capacitor.

Seriously, pick up some flux for a job like this. It makes this process so much easier and helps to make solid, reliable solder joints. I wouldn't trust these units with these repairs half as much if I'd performed this job without it.

After I'd finished replacing all of them, I probed around the board with a multimeter to check that the newly soldered in capacitors were connected correctly, touching one multimeter probe to the new solder joints and the other to another component on the same circuit board trace to check for continuity between them. Call me paranoid, but I did this about three times over to be absolutely sure. Thankfully, they all came up good.


Soldering can leave some residue on the board, so break out the alcohol / metho / cleaner and cotton swabs again and get cleaning.

I mentioned earlier that some of the lacquer on the board had burned off and rubbed off while soldering and cleaning. I touched up the affected areas on the underside of the board with a can of Circuit Board Lacquer from Jaycar. The board will still work just fine without this, but it's an added layer of protection.




Put the circuit board back into the TCU casing, and we're done!





With the reconditioned unit reinstalled in the car, it feels significantly better to drive. The upshifts, downshifts and moving from drive to reverse are all silky smooth. The shifts are timed exactly where they need to be, and it does feel like it's putting that power and torque down to the road a little better on acceleration. It's still an automatic, but it feels like less of the "slushbox" we've become familiar with in these older automatics.

The total cost came to $40 for another TCU from a self-serve wrecker, about $1.10 + Shipping for all six capacitors from Element14, and a couple of hours of my time.


So, how does this apply to the Engine Control Unit (ECU)? It turns out they have the same problem. I have a Mitsubishi AW319647 ECU in this car:





Lifting the covers of that unit revealed the same issue - a capacitor that has started to leak onto the circuit board. This unit hadn't failed yet, but eventually the ECU will fail in the same way as the TCU did.





See that dark patch to the right of the right-most pictured capacitor? Yeah, that's no good. I went ahead and performed the same repairs as I performed to the TCU to the ECU as well, and it's running beautifully - at least it's running as well as a 22-year-old Astron II can.

So, Second Gen owners that are keen on holding on to their cars - get in there and check those capacitors. You'll save yourself a lot of hassle down the track.


For reference, here's the capacitor list from both of my units:

Mitsubishi ELC26M MD750103 Automatic TCU, 2.6L ENG. (3-PLUG)

1uF 50v (85c) Radial Electrolytic x3
47uF 10v (85c) Radial Electrolytic x1
47uF 25v (85c) Radial Electrolytic x1
47uF 50v (105c) Radial Electrolytic x1 *

* This one is a common failure, and is the one that failed on mine.

Note: Just replace all the capacitors with ones rated for a maximum temperature of 105c. The price difference is a cent or two, and they'll last longer. I ordered all capacitors rated for 5000 hours @ 105c. (Since the TCU doesn't run anywhere near 105 degrees, the capacitors will last significantly longer than 5000 hours, something insane like 500,000+ hours of use.)

Note 2: Instead of ordering one 10v 47uF capacitor and one 25v 47uF capacitor, just order two 25v ones - it will fit and you can use a higher voltage, only the capacitance (47uF) has to be the same.


Mitsubishi AW319647 Engine Control Unit, 2.6L 4G54 EFI

22uF 50v (85c) Radial Electrolytic x1
47uF 50v (105c) Radial Electrolytic x1
100uF 16v (105c) Radial Electrolytic x1

Cheers,

~ Michael (Cruiser)

[MadMax]

Brilliant writeup!

Failure along these lines isn't restricted to the Mitsus of this age either.

There's one place in Adelaide that can test and repair these units if you find you have this problem and you can't DIY.

[ac1176]

I'll second that sentiment - nice work.

FWIW, electrolytic capacitors do have a service life, and will eventually fail irrespective of what brand or quality they are. The in-service life of an electrolytic is dependent on how much voltage it is subjected to relative to its voltage rating, and the temperature it is operated at. Generally it is better to invest in high temp (105 deg or better), and voltage ratings higher than the capacitor is anticipated to be operated at (25V+ in automotive applications), as you've suggested in your writeup. A good quality electrolytic cap if not operated too close to its ratings should be good for 15-20 years - a timely warning for us 2nd gen owners whose cars are now past their 20th birthdays.

[MadMax]

Computer motherboards used to have failures long ago due to bursting capacitors, current ones have solid versions - worth knowing if you want to do a remove/replace on car ECU capacitors.

[Cruiser]

FWIW, electrolytic capacitors do have a service life, and will eventually fail irrespective of what brand or quality they are. The in-service life of an electrolytic is dependent on how much voltage it is subjected to relative to its voltage rating, and the temperature it is operated at. Generally it is better to invest in high temp (105 deg or better), and voltage ratings higher than the capacitor is anticipated to be operated at (25V+ in automotive applications), as you've suggested in your writeup. A good quality electrolytic cap if not operated too close to its ratings should be good for 15-20 years - a timely warning for us 2nd gen owners whose cars are now past their 20th birthdays.

I wouldn't be surprised if several of the running anomalies associated with these older cars could be attributed in some way to electrolytic capacitors. With the conditions these capacitors operate in, even if they aren't leaking, they would certainly be dried out and operating out of spec. The ESR would have increased, which in turn increases heat dissipation, which accelerates the failure of the capacitor.

If the capacitors only dried out, then repairing these TCUs would be a simple process. It's the nature of how these capacitors fail by leaking electrolyte and corroding the circuit board this failure particularly awful.

I tend to agree though. A couple of tips for anyone taking on these repairs - you need to find the best capacitors you can get your hands on for applications like this. I would always opt for 105c or higher rated capacitors, with a higher voltage rating whenever possible (if replacing a 10v capacitor, use a 25v and so on), and stick to reputable capacitor manufacturers like Rubycon, Nichicon, Panasonic or Nippon / United Chemi-Con. As cheap and convenient those off-brand Jaycar capacitors are, they really aren't up to scratch in the long-term in an automotive application like this.

Also make sure to check out the capacitor data sheets if you can. I had a few Panasonic capacitors in mind until I looked at the data sheets and saw "Not Suitable for Automotive", so I went for the Rubycons instead.

Computer motherboards used to have failures long ago due to bursting capacitors, current ones have solid versions - worth knowing if you want to do a remove/replace on car ECU capacitors.

Those would be Aluminium Polymer Capacitors, and while they cost more than regular electrolytics ($3 - $4 per capacitor), they are well made and durable. If you're after a truly long term replacement - I'd be surprised if they ever failed again to be honest - then those are what I would be using.


The TR has been out a couple of times following these repairs, and I've noticed some improvements following the replacements, even when comparing it to a working factory-spec unit. The still-working unit had capacitors that still looked fine, and when tested in the car it seemed to function correctly - shifts were on time and weren't too rough, and downshifting didn't seem too bad.

However with this overhauled unit installed, there's a noticeable improvement in how the transmission operates. I can clearly hear the gear changes and downshifts, but I can't feel them. There's still some shudder at idle, something I would attribute to an old engine and worn mounts, but in motion it feels brilliant for a car of its age.

It wasn't until I was able to compare a working one with factory capacitors and one with replacement capacitors side by side that I realised how degraded the old unit was. Despite being removed from a 2.6L TR with over 500,000 kilometres on the odometer, with replacement capacitors, this TCU works like a brand new one.

[ac1176]

Good points.

Anyone who has done electronic servicing will tell you that electrolytic cap failures will account for a generous proportion of their work. It's what the failing cap takes out along with it that determines how cost effective the repair will be. About 10 years ago a mate of mine gave me a faulty set top box that would power up and operate, but the picture was garbage. As he knew I was an electronic nerd he said I could have it for free if I could get it working again, and half-jokingly said that it was "probably the electrolytic caps, or something". Turned out that was exactly what it was and I got a new set top box for the cost of a couple of caps.

Interestingly the TCU in your photos also appear to have a number of tantalum caps installed too (the orange egg-shaped components). Tantalum caps have an, at times, exciting mode of failure. When they die they go tend to go short-circuit. Depending on where they're used in the circuit the results can be catastrophic, expecially if they're used to filter a supply voltage.

[Cruiser]

I have another ECU that I'm working on at the moment, an AW335189.

This unit was removed from a working car, one that occasionally had starting issues. I removed the AW319647 in the TR SE and tested this unit in its place and was able to reproduce the same starting issue, so I'm assuming there is a fault within the ECU. However these same issues don't happen when testing this ECU in a third car.

I've searched these forums and determined that the AW319647 and AW335189 ECUs are interchangeable, so that shouldn't be it.

I'm not aware of any common issues with hard starting or stalling immediately after startup that can be attributed to the ECU, such as bad or intermittent connections at the ECU connector, so I suspect I'll have to probe around inside the unit and see what I can find.


I'll also check the ratings of capacitors inside both this ECU and a spare 2-plug TCU I have and report back. I'm sure someone will find it helpful having some capacitor listings here considering how common these failures are.

[Cruiser]

I can now confirm that the AW335189 ECU has the same electrolytic capacitors as the AW319647.

Mitsubishi AW335189 Engine Control Unit, 2.6L 4G54 EFI

22uF 50v (85c) Radial Electrolytic x1
47uF 50v (105c) Radial Electrolytic x1
100uF 16v (105c) Radial Electrolytic x1

In both the AW319647 and AW335189 ECUs I've refurbished, I've found that both the 47uF 50v and 100uF 16v capacitors were leaking. When replacing these capacitors, replace all of them, not just the suspect ones.

I've also found my spare 2-plug TCU so I'll update this thread with those capacitor listings shortly.

[Tpwagon]

Thank-you for a great write up...I have a first gen TP Wagon,it's a keeper and my daily transport and is in good condition,relatively fresh engine and trans....strangely I have noticed a very rare ( couldn't even call it intermittent) cut of power for a second,not a miss,just a total cut,it's a split second,many probably wouldn't even notice it,then back to normal,usually on hot days at cruising speeds. Yesterday after it had been sitting in the sun for some time I noticed it stalled at idle,I could restart instantly,the car runs faultlessly apart from this issue. It has me thinking I may be well advised to check both ECU and TCU capacitors. With you experience do you consider this may be a possible cause? I'm happy to grab a spare couple of units and get the capacitors changed,not a bad idea to have a few spares for 25 year old electronics! Thanks

[ac1176]

Possibly. Can't hurt to at least crack the top of the ECU and have a look.

Do you have an immobiliser fitted to your car? Check a thread I posted some time back regarding a similar-sounding fault I experienced.