On 01/19/00, ""Dr. Bob" <email@example.com>"
All this statistical analysis of the brain failure syndrome has me wondering what it is in the controller that gives out. I know there are some 'smart' parts in there in the later cars especially, the ones that supposedly 'learn' the habits of the driver and the car.
>>Bob, I had the pleasure of interfacing with the Bosch engineering groups that designed the Motronic and Jetronic controllers some 15 years ago. The LH-Jetronic control modules are of the 8-bit generation, probably use an 8032/8031 (ROMless 8051/8052) microcontroller. You will probably find a Motorola 68XX die mounted on a ceramic substrate in the mass air LH-Jetronic hotwire controller. Bosch often engineered custom ASICs to interface to the analog side of the circuits. There may be some power FETs.
The L-Jetronic brains were totally analog, used RCA chips and a lot of custom Bosch stuff.
The first generation ABS also used a full custom chip, code named Bayruth.
The "tuning" to the chassis was accomplished by an engineer with a "golden butt" who had the pleasure of living in the extreme north of Sweden every winter and replacing hydraulic valve orifices to compensate for system dynamics. His "tush" could feel how the ABS worked and he would specify a certain set of fittings for each OEMs cars. Short wheel base cars also have a 68000 CPU to manage excessive yaw, I don't think our 928s require this control but I know the Corvettes did.
>>>I'm not sure, it is possible that by the time the S4 generation came to be, the ABS might use a pair of 16-bit 8096/80196 processor to integrate the ABS and PSD, if not there is an 8031 used in the PSD system (along with
an accelerometer under the driver seat).
Dr. Bob... But the typical failures may just as easily be a driver transistor or one of the FET output devices that actuate the injectors. Looking at the construction while upgrading the chipset, those output FET's don't appear to be anything exotic, and since they stand up off the main board they are easily replaceable.
>>> Yep, quite often the devices near the interface are fried first.
They're the items exposed to over voltages, spikes and sparks. Do you have a SEM and XRAY set-up in you work bench too?
Charter Member 928 Owners Club
Central AZ PCA
first a collective thanks to all who helped to collect a reasonable set of data and also provided additional information. Since I can't post the curves, I'll put the summary in words:
I feel that with all data at hand, including the information from those who sell new brains, the most likely conclusion is that indeeed BOSCH had a manufacturing problem:
The LH brains from the 87-89 cars started to fail after 9 years. I have no report on 90 cars ("L") with a failure and only one 91 car whose LH brain failed already after 4 years (and was replaced by Porsche at no cost!!!!) so I consider this as an accidental failure, with a different failure mode.
Thus, the problem was most likely fixed late 89 or 90.
There was no report on failed brains on pre-87 cars, so, if they start to need new brains now, it is most likely a different failure mode as well.
The symptoms of a dying brain don't show a clear pattern.
They can fail with a sudden "death" as you would expect if a central control unit fails, but apparently it as can start with simple idle problems.
It is definitely not mileage related, as brains failed between 20 and 140 k miles.
This makes solder fatigue or bond wire problems less likely, as well as degradation of the semiconductors themselves, as both usually is greatly enhanced by temperature/temperature cycles so the mileage should matter.
Semiconductors normally don't die from just sitting somewhere.
So my guess is that penetration of humidity and corrosion/ oxidation is the key. This can occur by bad sealing of the chips, or by using insulating material that turns out to be aggressive to metals in the long run. Who knows ...
It would be great if some of us had the know-how and the equipment to analyze a failed brain, or by some other means identify the failure - and maybe even find a "cure" ... (So don't throw away the failed units ;-)
It is encouraging to see specialists are getting interested and share their knowledge to find the root causes and solutions.
David, (thanks for the part numbers of the brains!) you mentioned that you saw IT in the microscope, could you describe what was visible, maybe you already have the solution ?
As I mentioned before, I try to get something out of the BOSCH people, but this remains to be done. I'll keep you posted. If anyone wants more details just mail me off-list. I'd be also highly interested in any ideas on the potential failure mechanisms and fixes.
Keep those brains alive !!
90 S4 Euro Auto
you're right, I haven't heard of dead replacement brains.
But all the replacements I have heard of were all bought in the last years, so if they also life for only 9 years, they will start to fail not before 2006. But I think its reasonable to assume that they don't have that problem anymore.
AND beware of used ones - like some unlucky people who bought a used brain out of 87 cars ...
Some other information I forgot to mention:
From the data I've got, one out of three 87-89 cars have needed a new brain.
Even though this is probably an overestimation, due to the nature of such a poll (people with problems tend to respond with a higher probability than those without) but it shows that there REALLY was (and is) a big problem
built into the brains of that era.
The question is now - how can you check if your brain is prone to fail?
From the part numbers it seems not clear, but there might be a chance of narrowing things down. Here are some facts:
Porsche part numbers for the LH brain: 928.616.123. xx
89/90 13 (S4) 14 (GT)
91 25 (S4) 26 (GT)
One question now is whether all 89 cars have the .13 or .14 brains if yes, the number can't be used to tell a "bad" brain from a "good" one, if no, it might be a starting point.
But it would be only useful for original brains, as I suppose when buying a replacement brain, the part number e.g. for a n 87 hasn't changed, so I doubt that this is of much use to tell a original from a replacement brain.
The individual serial number of each brain certainly tells a lot more, but since it is virtually impossible to get hold of that numbers of the failed brains, the only chance is to get something out of Porsche or Bosch. They certainly know...
90 S4 Euro Auto
>I assume no one reported two failed brains on the same car? (i.e. do the replacement brains have the same problem or not?) Glad I have an 86. :) Maybe this should push the value of it up a little bit...
On my 928S2 (with LH-jet)
I recently experienced an ECU failure. The engine cut out during level cruising. It would not restart.
One tow later (with the trusty Jag XJ-S :-) and its back in the garage.
A quick check indicated that the fuel pump wasn't going.
The relay checked out OK, but was not being turned on.
I traced this back to a fault in the ECU. The ECU turns on the fuel pump relay with a transistor (which was OK).
The transistor is driven by a TTL 7402 chip, which has a dead output.. I replaced it with a 74HC02, and it works again.
my 87s4 with python computer runs like a champ
rhoderick t foney <firstname.lastname@example.org>
Cheapest thing I've ever fixed on any car : cost all of 50c !!!!
Moral of the story : Always carry a wire link around to replace the fuel pump relay if your ECU dies.
'84 928S2 A/T
I think this message could be important for any S4, GT, GTS driver. These cars use the Bosch LH-Jetronic fuel injection controller.
After 10 to 15 years lifetime the first controllers are beginning to stop working correctly. One reason for this is the Hybrid-chip which is no longer available. A new controller is more expensive than 2000 ? (Euro).
PFForum-members [a German language forum similar to Rennlist] wrote about a company (ACM Elektronik) which is able to repair all controllers. This information comes from a discussion with Mr. Müller from the ACM Elektronik company:
If you have problems with your LH-controller, possible symptoms are:
- The fuel-injector-valves are ticking if ignition is on.
- If you remove your foot from the accelerator, the revolutions are higher than 2000 rpm and the car is coasting, the fuel-calculator didn't show 0,0 liters/km.
- No smooth idle
- Air/fuel-mixture: too rich
- The engine won't run at higher than 3000 rpm
- Automatic Transmission: no RPM regulation when switching from "D" to "N" or "P".
- Engine doesn't run. The Spark plug is dry and is sparking.
It is not a shellac, it is actually a gel covering over the ICs on the hybrid circuit, but I suspect the real problem lies with the IC passivation not being up to snuff.
As Johnny said, the damage is has begun .... the most you can hope for is stasis. I've had the "made in Spain" replacement LH modules in for rebuild as well as the originals. I've had several customers send in their LH for rebuild prior to failure to ensure that they're not left at the side of the road at the most inopportune moment. This is the reason to never purchase a used LH, it could have gone "stale" on the shelf, or could fail the following week.
It really isn't fair to call the LH a "look-up" table. It also keeps the flappy in the proper position, you wouldn't want to be controlling that yourself!
[Note, for those of you that have L-Jetronic or early LH with 25 pin LHs,
1984 - 1986 EURO/ROW or 1985 - 1986 US models, your LH has no hybrid circuit, you can ignore all of this banter. Your LH fails at the "normal"
I get all types for repair - right up to GTS. Anything over 12 years and you're on borrowed time.
The failures appear to be partly contamination failure and partly static damage. The LH 2.3 was unique to the 928.And the "tile" is unique to the LH2.3 as well. All the rest of the LH electronics is very rugged, and failures due to other factors are rare. There is no reason a properly rebuild LH ECU will not past for many years to come.
The LH can now be rebuilt by www.928-ecu-repair.com at a very reasonable price. And yes, the problems with the "tile" happen on all LH2.3 ecu's staring with the first S4 models.