Flaky temp gauge - sender?

[JHZR2]

Hi,

over the last few months, my 91 318i has started to get an erratic temp gauge. The car has two sensors in the head, one for the gauge and one for the ecu. The temp gauge gets jumpy after creeping about halfway to normal, and it never goes past 3/4 of the way to where it would normally sit, which is at 9/20ths.

In the e30, it is well known to have a jumpy temp gauge. The common reason is a bad ground, due to a mounting nut getting loose. However, it is not the case as I've cleaned the connection and tightened the nut.

So the issue is either the gauge itself, or else some place along the line back to the sender. I once had a jumpy fuel gauge in another car that was a bad sender, but a float on a fuel gauge sender is a different kind of sensor...

Any thoughts?

Thanks!

 

[fsskier]

The approximate ohm values for your sensor are usually published and available - or hook up an ohm meter and just watch it as it warms up, if its jumpy, its failed.

On the other hand the only temp sensor I ever had fail was so cheap and easy to change...... I would probably just do that first.

 

[JHZR2]

The problem is that the sensor is under the intake manifold, so some metal parts have to be unbolted... The harnesses have 20+ pins each, so it is toughto sample.

Thus the question... More theory and what people have seen, before I go testing stuff out in the cold...

 

[fsskier]

Ouch! The last one I changed on a 4 cyl ford engine, one wire with a slip on connection, and the sender I believe just screwed out with a hex driver!

 

[L_Sludger]

I'm almost willing to bet money that it is the gauge itself.

 

[JHZR2]

Well as mentioned, I removed it, cleaned the connections, and re-tightened it on. Think it is the gauge mechanism itself, or is it still just not connected exactly right?

 

[mechanicx]

It'd be good to hook a scanner up and to see what the coolant temp is reading from the other sensor. Is it posible the t-stat is bad? As some mentioned if you could got the ohms scale for the cooalnt sensor and back probed the wiring with a needle and multimeter it should point out if the sensor is bad.

 

[SteveSRT8]

sure there's not another ground somewhere you missed?

Usually more than one. Just fixed a friends Dodge doing this, and it was one tine little chassis ground. Then go to the sensor next.

But the thermostat is always a suspect, just doesn't fail often in most cars unless it's been monkeyed with.

 

[Gary Allan]

Have you grounded the connection to assure that you can get FSD on the gauge?

 

[cchase]

I just went through this on my truck. Turned out to be that the gauges just get tired. Unless the gauges have stepper motors, it's unlikely that a "jumpy" needle is a worn gauge.

 

[JHZR2]

Originally Posted By: Gary Allan
Have you grounded the connection to assure that you can get FSD on the gauge?


sorry, fsd=? my brain isn't on yet. When connected one way, I could get the gauge to peg, fwiw.

 

[Gary Allan]

Full Scale Deflection - on some analog meters you'll see a current number either before or after the notation ..for example 1ma FSD ...or 1.5ma F.S. . The move to digital on most things may make that obsolete to some degree and I think the notation has been abandoned even in analog for the most part.

 

[mechtech2]

For a bunch of good reasons, check and clean your ground wires and multi connectors on the block and firewall - anywhere -nose around real good. A bit of corrosion causes anomalies for all sorts of stuff. Take it off and clean the substrate shiny - no paint or cooties.

Same for power cables.

Old style torpedo fuses also develop corrosion at the ends - check this out.

Now dive in to the sensor , it's connections, and it's wiring.

If this is simply from a low coolant level, shame on you!

 

[JHZR2]

Not coolant - checked that first. Plus, the sender is in the head, next to the one that feeds the ecu - and no issues there. I was thinking of adding a new ground directly, though certainly I'll clean other grounds that I find.

Thanks!

 

[JHZR2]

Messed around with my gauge more. I added a ground and torqued it to the nut on the back of the gauge that tightens to the ground on the pcb that it attaches to.

Less jumpy now it seems, but it still jumps. It definitely does it with no rhyme or reason too... super hard bump or pothole and it will stay stationary, sitting at a red light and it might jump...

Ill try to get into the sensor end and clean it real good. Havent been able to because of time and weather. Hpefully that fixes it...

 

[mechtech2]

Here's the list:
Coolant temp is stable and not jumping up and down.
The Sensor itself.
Connectors and wiring.
Dash gauge.
Gauge voltage regulator

 

[JHZR2]

If it is the gauge voltage regulator, does this effectively mean the gauge is shot?

I am assuming that the coolant temp cannot be changing as fast and erratically as the gauge is jumping, and since the sensor is in the head, I also assume there is no air in there.

I can test the sensor resistance, and ensure the connection is clean... Then change the gauge. Seems that is my set of options.

Thanks!

 

[mechtech2]

A bad regulator would affect all the gauges.
It is often a 5V regulator.
And it is rather rare for one to go out.
[This is not the voltage regulator in the alternator.]

You gotta dive in and methodically check things out.
And it could also be a combination problem - those are tricky.

 

[JHZR2]

Originally Posted By: mechtech2

If this is simply from a low coolant level, shame on you!


Well, it was not coolant level, that was fine, none is leaked or consumed... But it was an air bubble of some sort. From the last time I did coolant. Didnt start right after that, but the bubble must have migrated and gotten stuck somehow near the sender...

In the past I had re-bled the system, massaged the hoses, getting all air out. Well, I played with that again.

The M42 engine has two sensors, both more or less under the intake manifold. The outer one is for the ECU (more important, placed so it is easier to replace), the inner one for the gauge. I pulled the lead to the one for the ECU, and got funny operation and an OBDI CEL. Put it back on and it was fine. Played with the hoses, re-bled the system, and now the gauge works like normal, no jumpiness.

These gauges are known to get bad grounds to the PCB in the cluster, causing jumpiness. I added a ground which will be good for the long-term. Not a bad thing...

But it was a slight bit of air in there. Hopefully no damage done because of this and the resultant hot zone that may have been present... I wonder if the ECU sender was getting funny readings too. If so, it was beneficial... the last few tanks I actually have gotten about .5MPG higher than usual, with no change in driving profile!

Thanks for all the help!