Coolant temp gauge replacement (late petrol)

[937carrera]

...and yet he has had the same results with two gauges and two senders.

He's already ordered another sender, so that'll be here soon, plus he has one that is a known good one in his other van, which is why I suggested he hooks his independent wiring up to that. If it works correctly then it also disproves his current theory of other as yet unknown components in the circuit.

This is a real curious one. I predict whoops of joy and much self kicking when it's finally solved.

[itchyfeet]

The clue seems to be that at some point previously a third party temp gauge was installed and the question I ask is why? If it were me, I would be happy with the dash and focus on the temp sender part of the circuit and associated wiring. Might be worth just re-looking at the Fuel injected 2.1 cooling system schematic for anything that has been overlooked.

I solved the why was the old gauge not working puzzle here

http://forum.club8090.co.uk/viewtopic.p ... 5#p8225074" onclick="window.open(this.href);return false;

the heater wire was touching the PCBA, as soon as I moved it it worked with the resistors as it should, this origional gauge is the one I took appart and now have removed PCBA.

I also found the regulator had a bad connection to the centre leg so I haven't wasted all the 10+ hours I have achieved something

I had a look at the DJ wiring schematics look the same to me.

[itchyfeet]

...but I think Pauls position will be that he is testing the gauge / sender independent of the van wiring with a 10V regulator in place providing the correct voltage. The needle swing / LED only happens with senders (2 of them), not fixed value resistors.

I would get a replacement sender, re do the test. One of the senders was tested on the bench and, at 80 degrees, had the gauge virtually full scale. Either the gauge is duff, the temp was higher than measured or the sender is the incorrect type/faulty.

Using appropriate fixed resistors is how VW check calibration and is valid as a method. The sender currently fitted could be unstable or insulation breaking down as temp and current increases. A dash direct to the sender with independent 10v supply and wiring, and isolated from the rest of the system wiring suggests sender, pcb foil or gauge at fault, I cannon think of any other possbility.

What I really can't get my head around is:
I have tried small resistors and power ressitors at 100 ohms and both gauges work OK ( just to left of LED)
both senders read high with both gauges either warming up in vehicle or in a cup of water at 85C (typical running temperature) tried all combinations in and out of vehicle.

Broken senders is the logical answer so why does the sender read much higher resistance than 100 ohms with a multimeter and yet behave like a lower resistance?
Muiltimeter isn't testing with a similar current ( sender pulling up to 100mA) is the logical answer.

So I bought another sender, as said maybe my 'new sender' is also broken some how by being dropped or thermal shock from being dunked in boiling water.

[AngeloEvs]

Odd indeed. Did not notice that you had a new sensor and testing that.

Time for some pure physics methinks but you need a second meter, one measuring current, the other the voltage across the sender. R=V/I which gives the true DC resistance of the sender as temperature increases or at a fixed temperature, i.e, 100 degrees boiling point.

[itchyfeet]

I even started testing with my 'scope, a current probe and voltage probe at work but just didn't have enough time to.get meaning ful results.

[itchyfeet]

I also tried 100R across the VDO gauge and got alot lower reading than I do with the sensor

[937carrera]

Broken senders is the logical answer so why does the sender read much higher resistance than 100 ohms with a multimeter and yet behave like a lower resistance?
Muiltimeter isn't testing with a similar current ( sender pulling up to 100mA) is the logical answer.

So I bought another sender, as said maybe my 'new sender' is also broken some how by being dropped or thermal shock from being dunked in boiling water.

What about the link from Samba, did you read that through ?

As I understand it current going through the thermistor affected its resistance, leading to high readings. He changed the current input to 100th of normal, measured the resistance and then scaled the output back up again and all was good. My own opinion is that he shouldn't have had to do that...... but it to have seemed worked. Is there a clue there ?

[itchyfeet]

No I didn't read it properly

I must admit it looked like thermal.runaway to me I didn't mention it becase I was unsure and didn't want to throw more into the mix
If I remember correctly when I started measuring current with the sensor it increased after I connected it even though it was already in hot water for a while so the current flowing was affecting the resistance.

[AngeloEvs]

Im on hols at the moment, I have temp gauges and should have spare sender. When I getback I will do some measurements on bench and also of the sender in my van just to see what readings i get.

[itchyfeet]

Odd indeed. Did not notice that you had a new sensor and testing that.

Time for some pure physics methinks but you need a second meter, one measuring current, the other the voltage across the sender. R=V/I which gives the true DC resistance of the sender as temperature increases or at a fixed temperature, i.e, 100 degrees boiling point.

Did this tonight on the 'new sensor' that I fitted at the weekend ( the one that's been dunked in boiling water) it's the early type which is fitted to the 88 tin top with 6.35mm spades and not the later type fitted to 89 Pop top with 2.8mm spades.

Tomorrow If it arrives I can repeat on the brand new sensor.

It's all academic because if there is enormous current ripple these readings are not going to show that resistance is calculated from V and I

dash in van 10V to regulator Level sensor relay disconnected and the red wires with 6.35mm spades are connected where I cut the sensor wire in black box and then to engine earth.

P1100643 by Paul_Barr, on Flickr

523 Ohms 18mA 9.41V Switch on ( remember measures about 1K Ohm with a meter)
128 Ohms 57mA 7.28V 3 mins 1/4 gauge
86 Ohms 74mA 6.33V 5 mins approx 100 Ohm resistor point just before LED
59 Ohms 86mA 5.10V 6 mins
54 Ohms 96mA 5.17V 7 mins 3/4 gauge
49 Ohms 100mA 4.90V 8 mins
45 Ohms 104mA 4.69V 9 mins Light on
39 Ohms 111mA 4.30V 12 mins full gauge
38 Ohms 112mA 4.27V 15 mins full gauge ( thermostat not open yet)

the VDO gauge connected at 15 mins, it sits just below 80C normally and it was close to this ( wrong sensor so not accurate)
68 Ohms 50mA 3.4V


looking at the thread the timings and the gauge level is the same as the original sensor ( the one with the 2.8mm spades) when tested in the van and later from a remote battery
http://forum.club8090.co.uk/viewtopic.p ... 5#p8225577" onclick="window.open(this.href);return false;


Two duff gauges with the same fault is another equally logical possibility as two duff senders but I'm struggling to see the failure mode that would cause the difference between a sender and a fixed resistor.

[AngeloEvs]

Those readings and resulatant guage needle position look correct, if it were anyone else I would say overheating/faulty thermosat.

See how you get on with the new sender...........

[itchyfeet]

if it were anyone else I would say overheating/faulty thermosat.

I doubt it it happens so quickly and my test with a sender in hot water of a known temperature send the gauge high just the same

This picture from page 5 is in 79C water ( that's a thermocouple reader)

P1100615 by Paul_Barr, on Flickr

[itchyfeet]

if it were anyone else I would say overheating/faulty thermosat.

You are right this is a logical conclusion so I need to rule it out and not go by my instinct, it's been running for a few thousand miles in hot weather also and never any signs of overheating but silly not to check.

Knocked this up earlier, its aluminium so should transfer heat well and the thermocouple probe is deep inside with heat sink grease, screw to pull it out and secure the probe
Will probably have some lag but should be reasonably accurate when up to temperature.
You have to wonder if a long probe and a reader on the dash is an option

I'm going to tape a probe to the stainless large water pipe as close to the thermostat housing as possible to see if there is a corrilation, it might help me and others take measurements later and apply an offset.

20180808_121754 by Paul_Barr, on Flickr

[itchyfeet]

Fitted the thermocouple in the sensor position and taped one on the stainless pipe.

I added the results to the previous results taken after the same time so they should be reasonably aligned.
Sensor position 1st metal pipe 2nd
The engine was idled in the same way as before and the evening temps are similar.



22C 22C 523 Ohms 18mA 9.41V Switch on ( remember measures about 1K Ohm with a meter)
50C 49C 128 Ohms 57mA 7.28V 3 mins 1/4 gauge
62C 60C 86 Ohms 74mA 6.33V 5 mins approx 100 Ohm resistor point just before LED
68C 66C 59 Ohms 86mA 5.10V 6 mins
72C 69C 54 Ohms 96mA 5.17V 7 mins 3/4 gauge
78C 74C 49 Ohms 100mA 4.90V 8 mins
82C 78C 45 Ohms 104mA 4.69V 9 mins Light on
94C 89C 39 Ohms 111mA 4.30V 12 mins full gauge
93C 85C 38 Ohms 112mA 4.27V 15 mins full gauge ( thermostat not open yet - well that's an assumption top pipe not hot)
92C 85C 20 mins top pipe hot but probably a bit more to go fan not cut in.

So it looks OK to me perhaps it's running 5 or so degrees hotter than it should or perhaps the coolant metal pipe is a better reflection of the temperature, or perhaps the tempertaure would come down some more with the engine revved and more flow.

Full gauge should be 115C I think because it's written on it.

Before fitting the new, new sensor that arrived today I may try swapping the thermostat and repeating , the one in there is approx 1 year old Brickwerks Whaler standard 87C one, I have another new Whaler 87C one or an 80 deg C one yet to decide.

Things that might affect results

The sensor is directly in the flow from the head and may be slightly higher in temperature
The thermostat has cold coolant on one side and a trickle hole so may be slightly cooled before getting up to full temperature
The thermocouple is not calibrated and +/-2 degrees stated as a tolerance for safety but probably better than that.
Being at idle the pump is not shifting as much coolant so the temperature may be slightly raised in the engine.
The Heater was on which may have slowed to opening of the thermostat.

P1100686 by Paul_Barr, on Flickr

Thats another 2 hours of my life today I won't get back

[itchyfeet]

Existing Stat- New Stat

22C 22C - 19C 19C Switch on
50C 49C - 47C 46C 3 mins
62C 60C - 60C 60C 5 mins
68C 66C - 65C 65C 6 mins
72C 69C - 71C 70C 7 mins
78C 74C - 76C 75C 8 mins
82C 78C - 80C 77C 9 mins
94C 89C - 92C 89C 12 mins
93C 85C - 97C 92C 15 mins
92C 85C - 99C 95C 20 mins

When it got to 100C I switched it off, the stat was barely opening the top pipe was barely warm.

Quite alarming really that there is so much variation in thermostats

there was a little air in the expansion tank so pressure will be lower and temp slightly higher but exactly the same amount of air today as yesterdays test.

it looks like it was manufactured 2.5 years ago even though I only bought it this year.


I wonder if they don't get used do the capsules stop working?

P1100688 by Paul_Barr, on Flickr