Date:         Fri, 6 Oct 1995 11:28:58 -0500 (CDT)
Sender:       Vanagon Mailing List <vanagon@vanagon.com>
From:         Gerald Skerbitz <gsker@lenti>
Subject:      Don Lancaster's Intermittents

On Fri, 6 Oct 1995, Steve Hoge wrote:
> 
> There's also the rumored 3rd fix for this problem (aka The Don Lancaster
> Solution), which is resoldering some steel-leaded (!) resistor or another in
> the ECU.  Hey Gerry - is that Electronics Now article describing the fix (Feb
> '95?) ever going to make it into the archive?  I *have* heard from Vanagon

Actually, what Don sent me was from January 1995. And, although the 
article is long, it's not really about the practicalities of fixing car 
computers, but rather a lot about philosophy and theory.  Here's the 
pertinent part of the article.

                         Servicing Intermittents
   My 1987 Synchro 4WD van started showing an intermittent and a 
total loss of power. Naturally, I did not suspect for an instant that 
all those 138,000 off road desert miles I put on it had anything to do 
with the problem.
  Cleaning up the airbox, checking connectors, and swapping the fuel filter
did not help.  Hauling it off on a 350 mile trip to my nearest factory
authorized service center cost me an outrageous bill. 
  For zero improvement
  In all fairness, it _is_ hard to fix an intermittent when it doesn't show
up on demand.  But as soon as I started treating this as an electronic
service problem, rather than an auto problem, the cause became obvious. 
  I next sent out for a shop manual, something I should have done years ago. 
The _Robert Bentley_ manuals are really outstanding. 
  Meanwhile, I decided I had to try and catch this intermittent in 
action.  While driving down the road.
  So, I hooked up an oscilloscope.  My first guess was the Hall Sensor, 
so I monitored the green wire from the sensor by using a temporary test 
pin I have shown in figure three.  The sensor output was continuous, even 
during a dropout.

Figure 3 -- A QUILTING PIN makes a safe test point along stranded wires.
 
Quilting Pin                           wire
          ------>   O    ------------------------------ 
                    |   /
 ----------------------/
                  /--------\
                 /  _____   \
                /  /|   /   | <--- Polystyrene Shipping Bead
               /--- |  /   /
                  ----/   / 
                 /  _____/
                /  /|_______
               |___________/
                  
  Finally, a stroke of blind luck.  I hit the computer with my fist and the
engine died!  Something I should have thought about long before.  Something
which that mechanic _certainly_ should have tried. 
  Cleaning the connector didn't help, so I resoldered the computer.  The 
culprit was this bad solder joint on a steel lead power resistor.  Aging 
and corrosion caused the failure.  
  In hindsight, the tachometer would drop to _zero_ during failure, with 
the engine obviously still stumbling over.  Because the tach connects 
directly to the coil primary, the problem almost _had_ to be the engine 
computer or wiring.
  Interestingly, there is this newer wiring harness/filter that's 
supposed to eliminate the very same _symptom_ that is apparently caused 
by the steel lead on that big computer resistor not being solder 
wettable.  I suspect they never found the real problem at all.
  And still haven't got a clue.
  All of which did get me thinking about servicing intermittent problems 
in general.  So, figure four gives you a set of my rules that should get 
you started.  The key points are to _always _ have documentation on hand; 
to be _certain_ to get the problem to actually show up; to 
divide-and-conquer by finding out where the problem is _not_; to attack 
probably causes first; and to think logically, paying attention to _all_ 
of the symptoms.

  Figure 4.
    1.  Make certain the problem is real.
    2.  Have _all_ service manuals on hand.  Read them!
    3.  Make the problem show up.  Collect data _during_ the fault.
    4.  Divide and conquer.  Find out where the problem is _not_.
    5.  Assign probable causes. Look, smell, touch and listen.
    6.  Associate mechanical problems with connectors;  time ones with
        solder joints; temperature ones with memory chips.
    7.  Cause the patient no harm.  _Never_ hot plug!
    8.  Isolate temperature problems with fans, cold, or hot boxes.
    9   Pay attention to _all_symptoms. Be willing to change tactics.
    10. Hit it with your fist or otherwise apply _controlled_ shock.
    11. Attack the disease, not the symptom.  Things "burn out" for a 
        reason.  Find and fix that underlying reason.
    12. Reduce the system to the simplest possible.
    13. If all else fails, try to make the problem _worse_.
    14. When possible, compare against an identical working unit.
    15. Back off and give it a rest;  let your subconscious do the work.
    
  Yeah, there are fairly low cost data loggers out there.  But nobody has 
yet come up with a universal intermittent "flight recorder" that a car 
mechanic, a cardiologist, or an air conditioning repairman would fight 
over.  There's a bunch of opportunities here.
  Many thanks to real-world auto expert Bob McKnight for his superb help 
on this matter
  

All credit for this is due Don Lancaster.
Synergetics
Box 809
3860 West First Street
Thatcher AZ  85552
Last known email: Synergetics@genie.geis.com

 --
Gerry
Gerald Skerbitz <gsker@med.umn.edu> U of MN Med School Admin 6-5379
Home St. Paul,Ramsey County,Minnesota, USA