My temp gauge was behaving in a similar weird fashion.  I got a new
sender..still weird.  there is a test protocol in Bently for the
instrument...Ground it goes to full hot, maybe?

 Anyway, eventually,  I discovered that the ground wire for my sender
(inline four, but same stuff) was almost broken and only getting a partial
ground, sometimes...So the gauge read pretty low most often, but
occasionally it would go full hot...Be aware that the gauge takes some time
to move across the range...Mine would 'lose the connection' to it's ground,
but the gauge would take about 8-10 mins to go up to 'Hot!'...that fooled me
into thinking I had a cooling issue..when in fact, it was just a rotten
brittle ground wire (very small gauge wire on my vehicle) that was
intermittent.  Worth a look.
  Don Hanson

On Tue, Nov 30, 2010 at 4:16 PM, David Beierl <dbeierl@attglobal.net> wrote:

> At 06:23 PM 11/30/2010, mark drillock wrote:
>
>> It is easy enough to test the gauge circuit with $1 pack of resistors
>> from RadioShack. Then you will know if the cluster  or some part of it
>> is the problem. I will help any owner do this testing on their own van
>> if they will get the resistors and send me a photo of them holding them
>> in front of their van.
>>
>> 100 ohm, 1/2 watt
>> part number 271-1108 for a 5 pack
>>
>
> Nominal is 35 ohms should blink the light and bring the needle to the
> extreme left of the overheat zone, +/- one needle width.
>
> 265 ohms should (eventually, maybe with a tap or two) bring the
> needle to the bottom of the "transition zone" again +/- one needle width.
>
> If you have the gauge out you'll see three little lines that you can
> sight along the needle to; nominal, +limit, -limit for each of the
> two calibration points.
>
> Yrs,
> D
>