Relating FSD temperature history to MTBF

[FinSub97]

Questions for ucdavis, BETA, BILL H, GARY PAGE, StephenA, gmctd and other top gurus.

Let

[MTTwister]

If it's the transistors that generate the most heat, and that seems to be related to the expansion / contraction 'breakage' of the soldered joints internally, should additional the temp 'probes' be mounted to each transistor and record their cycles? ( never have enough data pooints, ya know).

[FinSub97]

That should be asked from those who have deeper understanding on thermal results inside FSD. If it is about cracking solderings, could that location be monitored and does the cracking take place always at the same spot?

My gut feeling is that it is enough to measure temperature as close to those transistors as possible. That means the spot should be either
a) between the transistors
or more easily
b) in the middle the side surface as close to the junction between FSD and cooler (BETA, IP or whatever) as possible.

'a' -alternative imposes problems in terms of instrumentation. There is a risk of affecting the mating FSD to cooler, which could lead premature failure.
In 'b' -case pt100 or pt1000 doesn't interfere FSD functionality and cooling in any way.

If you meant that temps should be measured in multiple locations, sure, why not. Main idea is just to get one number describing what kind of thermal profile FSD has gone through in it's sordid history.

Another gut feeling is that if we could standardize the location and glueing method, we would get amazingly coherent results around the globe. This is based on assumption that Stanadyne et al has some sort of QC and all these POS are equally flawed. It is probably not in the components but in the design itself. Again, if the thermal history results in cracking or other mechanical flaws, CTT is an appropriate quantity to describe the treshold to failure (with a certain probability).

All comments are welcomed.

[autocrosser]

I think we should get everyone we can to file a complaint with the Nat Hiway Transportation Safty Admn.

http://www-odi.nhtsa.dot.gov/ivoq/

We can document the danger that the failure presents when it fails while driving resulting in the loss of steering and braking after the minimal reserve pressure is lost in the brake system.

I filed one awhile back after mine failed and I was going around a curve towing a 8,000lb load. I was able to control it but it was a surprise due to the increased steering requirement that resulted in my crossing over the center line.

One or two won't do much but a lot of filings might get some attention.

I happened to be coasting at the time and the engine had momentary power steering but after the engine shut down completely it was gone. The brakes were gone after a a couple of applications trying to navigate to the side of the road on a 4 lane hwy.

[tom.mcinerney]

FinSub...
>Another gut feeling is that if we could standardize the location and glueing method, we would get amazingly coherent results around the globe. This is based on assumption that Stanadyne et al has some sort of QC and all these POS are equally flawed. It is probably not in the components but in the design itself.

Three observations argue against the expectation of a homogeneous 'issue', readily quantifiable:
1.) These vehicles are often as reliable as the 6.2L engines for the first 100K miles. 2.) Many FSD/PMD failures have been 'cured' by attention to either of two "maintenance issues", namely fuel supply and electrical connections. 3.) Both Beta and JD independently discovered that some FSD/PMDs had transistor mtg nuts which were loose, dooming them and the associated FIP. These 'loose nuts FSDs' have been found by others as result of communication on this page.

[FinSub97]

Tom,

Judged by lack of comments in this thread I probably failed to convey my idea (or plainly the idea just sucks). Nonetheless, if the initial assumption (that is, heat and it

[ucdavis]

Jari-
I'm pretty sure there are only two qualifications for "Top Gurus" here on TDP. First, have lotsa opinions, doesn't matter if they're any good. Second, spell check ;)
I'd say you were right on both counts, great theory, and idea sucks. I fully believe your hypothesis of the MTBF as applicable to the FSD's various failure modes. Unfortunately, I think two major variable pools would polute the outcome, absent substantial testing of supposedly dead FSD's.
First, there appear to be multiple ways this box can muck up our lives (nuts untorqued or de-torqed allowing uncontrolled expansion, sloppy soldering, an occasional unit literlly burns up internally for GKW reason (I got one off Ebay that had a burn hole originating inside the FSD & others on TDP have reported similar melt-downs), and of course there's other internal component failures that surely occur among other possibilities...). Second, there are a lot of potential snafu's outside the FSD itself (bad grounds is common, poor contact w/the heatsink, misdiagnosis (which could easily run 50% or more of the reported multi-FSD-failures), lift pump intermittents, etc.) that get blamed on FSD trouble. After all, w/a known trouble maker, it is easy to jump to conclusions.
Soooo, adding all this noise into the model, you'd easily drop the power of any correlation (again, absent lotsa individual FSD's getting checked out for their true status post failure claim) between CTT and fail time to a level that would not be worth the pursuit. If the failure modes were more localized to the transistor-solder/heat trouble, your approach would be right on.
I gotta say, whatever is in the water in Finland, I wanna get me some ;) Do they bottle that stuff?

[tom.mcinerney]

FinSub-
First, thanks for your initial post/suggestion, I would aggressively support your plan if i didn't think there was a problem with a secondary assumption; also if resolution were not near at hand.
>Let

[gmctd]

While heat plays a role in the failures, it is loss of electrical contact due to chromium oxides on the stainless steel transistor driver case, stainless kep nut and washer, and stainless screws used as clamps and electrical circuit path that is the addressable problem.

This path connects the grounded Fuel Solenoid coil, via the printed circuit board, thru the screws to the transistor case, which is Collector, then thru Emitter to +12v. Very critical that there be minimal voltage drops in this path.

Simply re-breaking the oxides, by loosening the nuts and re-tightening them an eighth turn, re-establishes electrical continuity.
An eighth-turn works out to between 10-15"lbs torque, sufficient to maintain (estimated, based on number of loose nuts found so far) factory levels of tension.

Barring other failure, this procedure could be performed any required number of times to restore electrical continuity.

The white heat-conductive pads under the transistors are of heat-activated adhesive material, such that thermal conductivity could be maintained in event of loss of tension at the screw\nut\case interface.

And, if anyone has information on a Delphi Packard male weather-pak connector which mates to the FSD female connector in the IP harness, please post it here. TIA

[ 02-11-2004, 03:29 AM: Message edited by: gmctd ]

[FinSub97]

Two things:
1) I was delighted to read your replies. Professionalism in them speaks volumes.
2) I was afraid of that kind of comments in terms of scheming up additional metrics to increase our common knowledge of this POS.

Let me conclude. The initial assumption is not valid at least for following reasons:
a) Quality engineers at Stanadyne are not doing their work. It is outrageously infuriating to hear that the 1604-euros-reason (=2030-usd-reason) could be some loose nuts inside the box, out of reach for my fingers.
b) Many other engine related problems are possibly misdiagnosed as FSD failures. Absent post mortem autopsy on these