the first thing I would have done is replace the gauge.I thought you did that. what does the gauge read when cold, does it go back down all the way .

 

I had the exact problem. Tried two different sending units and they were all showing 30 degrees higher that the actual temp. Finally got an oem unit and all is good.

 

Replacing the gauge should never be "the first thing you do".

 

Yep. "the first thing you do" is learn what the real problem is.

 

for me time is money, and replacing the gauge would be simple and fast along with the sender.I know you guys are gunna say why spend the money on gauge, well to me why spend the the time.

 

Thanks for the additional thoughts. Re replacing the gauge, the main reason I haven't is because I used a potentiometer set to approx 125 Ohms (about equivalent to Ohms at normal operating temp) to the gauge and the gauge reads correctly (about 160-180 degrees). And it stays at the same 'temperature' regardless of the RPMs, the voltage, speed, etc. That seems to imply that the sending unit is heating up beyond the 'real engine temperature'. BUT, when I connect an ohmmeter to the line from the sending unit (disconnecting the gauge) it stays at 125-130 Ohms regardless of the RPMs, the voltage, speed, etc. Reconnecting the gauge, I see the gauge immediately go to the correct (normal) temperature, and then within a few seconds rise to above 3/4 of the heat range. I plan to look into the current issue at the sending unit as mentioned by Jim (or replace it again with another).

Off the wall possibility - the bottom of the sending unit is immersed in antifreeze in order to operate. All works correctly until normal operating temperature is reached (~180 degrees). Then the thermostat opens and releases antifreeze to the radiator. Could this be evacuating the antifreeze in the same chamber as the sending unit causing it to maintain the Ohms sent (but changing the current) increasing the temperature gauge reading?

 

One that likely requires a completely separate circuit for the gauge in which current flowing through it is controlled proportionally by the current flowing through the sender, which is a varister, or variable resistor. I might have had the aptitude to design such a circuit using descrete components from Digi-Key, but not anymore....when I ran into this same problem with my FE Ford I bought every part number NAPA had and the third one out of 5 was right on the money.

Dan

 

I've been thinking about how to solve this thermal runaway issue. One way might be to drill into the sending unit and pump some heat-sink compound in such that the internal thermistor is completely immersed. But that's too easy and too simple. It wouldn't be any fun.

So in an effort to complicate solving this problem and make solving it fun, I propose:

Power the sending unit with a constant current source. Choose the current level such that the developed Voltage across the sender when cold is approximately 5 Volts and close to zero Volts at, say, 220*F.

Measure the developed sender Voltage with an Arduino controller, do some calculations on the measured value and output a Voltage between 0 Volts and 5 Volts corresponding to a cold reading and 220*F respectively.

Apply the output Voltage to a Voltage controlled current source which is connected to and drives the gauge.

Finally, tweak the Arduino code until there is accurate tracking of the temperature gauge and actual engine temperature.

Now THAT sounds like a LOT more fun than simply making the sending unit work correctly, doesn't it?

Jim

 

I had a similar issue with my 65 when I first got it running. I installed two aftermarket gauges and both had bad readings. Finally, I found a 66 NOS gauge and it worked as it should.
Many times, tough, it is the sending unit on the manifold. Again, finding an original is always the best bet. JMO

 

Indeed. Cadillac "Comfort Control" introduced in 1964 used a string of in series varistors and a 2 transistor amplifier to control current (constant source as you mention) to a resistance wire which would expand and contract in accord with said current, and mechanically attached to a vacuum valve that metered vacuum to a mechanical servo. So it can be done along the lines you suggest...but am I correct in my assumption that the sender self-heating matter afflicts only aftermarket senders and not NOS/OEM units?

Dan

 

To the best of my knowledge, Dan, that's true. Every instance of self heating of which I'm aware has involved aftermarket senders.

Further thoughts: Sender current supplied from a constant current source makes self heating impossible. Think about it this way: Heat production is a function of I^2*R. If R decreases due to an increase in temperature but I remains constant, heat production goes down, not up.

Jim

 

Thanks to all respondents - I think my best bet is to try to find a 'good' sending unit. Discovered that GM discontinued the part so I'm stuck with non-OEM sensors. I appreciate the "FUN" ideas offered to limit the current to the sender, but realistically I'm not up to them. I'll let you know when I 'get lucky'. Happy New Year to all.

OK - Finally got lucky with a good sending unit that doesn't 'self heat'. I guess quality control on these is not a high priority! Again, thanks to all for the advice.

 

IIRC more than one member here has bought a handful of Wells TU-5 senders and played around with them until they found one that matched their readings with an I/R temp gun. A PITA but it worked out.