Multi Temp Sensor ('88 5ktq)

["Mike Hopton" <mike@genesis-microchip.on.ca>]

     Sorry if someone has posted this already. I'm guessing Scott Mockry, 
     Igor or others have already BTDT, but what the heck I'm a 5ktq virgin 
     ,so to speak, so here goes my experience to date with the multitemp 
     sensor....
     
     The temp guage on my recently aquired 5ktq ('88) was working 
     intermittently and so I replaced the multi temp sensor with an old 
     spare and cleaned the contacts. This fixed the problem for about 48 
     hours before it returned to it's normal "lazy" operation. Both 
     sensors were made by VDO and have the 4 spade terminals, 
     interestingly both have an 11/87 date code.
     
     So with nothing to lose I performed a bit of failure analysis. It's 
     really easy to disaasemble the sensor (I think easier than the WOT 
     switch) but I guess this is because I didn't receive any flesh wounds 
     on the sensor. The best tool for the job is a dremel with the cutting 
     disc attachement. Carefully cut away the metal about 2mm from the 
     connector end of the sensor,cut along the circumference (a vice for 
     holding the sensor helps). Poor ascii drawing below:-
     
        --------  4 Terminal Connector end
        I      I
        I      I
        I------I    
       (        ) 
       I        I <---- Cut just below where 
       I        I       the metal has been 
       I        I       bent over to form a seal 
       I        I
     --------------
     I  I      I  I
     I  I      I  I
     I  I      I  I
     I  I      I  I
     --------------
        I      I 
        I      I
        I      I Threads
        I      I
        --------
          I  I
          I  I  Probe
          I  I 
          ----
     
     The inside of the sensor slides out after the metal has been cut away. 
     Inside there is an electronic module connected to the 4 terminal 
     connector and a spring that connects the bottom of this module to the 
     bottom of the probe. There is also a rubber O-ring used to seal the 
     unit from the elements. IMO there are two problems with the design.
     
     1) The rubber o-ring does not provide a heremetic seal. This becomes 
     worse over time as the o-ring is temp cycled.
     
     2) There are two small triangles of metal attached to either side of 
     the sensor that make contact with the inside wall of the sensor. These 
     provide the return signal for the temperature measurement. It's quite 
     likely that the contact provided by these two small triangles breaks 
     down over time.
     
     The fix:-
     
     Reflow all the solder joints inside the sensor and clean all the 
     contacts, specifically the two triangular contacts on the side of the 
     board,and the contacts at the bottom of the board that mate with the 
     spring. I didn't use the o-ring when I reassembled the sensor, it had 
     become quite badly distorted and just used some epoxy which performs a 
     better job of keeping out moisture. Anyway for 30 minutes and the 
     satisfaction of fixing some thing I can recommend "having a go" at the 
     temp sensor.
      
     Incidentally there are two small boards inside the sensor. One a small 
     substrate with some surface mount components (and an LM2902) and one small 
     PCB with some through hole resistors and diodes. The boards are mounted one 
     on top of the other with the connector end embedded in some type of epoxy 
     goop. Quite interesting from an electronics packaging perspective and 
     definitely a package that is prone to a short MTBF.
     
     Regards, Mike