Coolant Temp Problems

[Hugh]

NP. I edited that post to make it clearer.

 

[Hugh]

Repair successful. It was the gauge sender part# 37750-PH2-014

Ambient temp 75 degrees, driving at 30MPH dash gauge starting coming off the bottom at 4/10th of a mile. At 1 mile the needle was just below the half way mark. Exactly the same as my car which we all know is perfect. :biggrin:

 

[FF Drifter]

Wanted to add to this thread. My 1995 had very similar symptoms: the temp gauge on the dash first became erratic and would not go up to normal operating temp (on my car, between the halfway mark and the tick mark just below it). Going downhill, the needle would fall even lower, sometimes all the way down to the very bottom tick mark. Because the issue started soon after I had the coolant expansion tank replaced, I thought it was related to the coolant not being bled properly. I did that and it seemed to fix things... but not really.

I suspected the thermostat as recommended elsewhere on Prime and by my friends, and sure enough, yeah the NSX usually took several minutes to warm up. I changed the thermostat but still had the same gauge issue.

Did the same diagnosis Hugh did 10 years ago (pg 23-136 in the manual) and my readings on the ECT sending unit was 54 ohms at ambient temp and 149 ohms at operating temp. The manual says these temps should be 32 - 49 on the low end and up to 142 ohms at operating temp once the radiator fans come on... so I was off, but not as far off as Hugh was finding on the 1996 earlier in this thread. The manual says to replace the unit if the readings are "substantially different" but I wasn't that far off. Regardless, I went over to the local Honda dealer and picked up 37750-PH2-014 ~ replaced that and now I'm good to go.

The sending unit is a real pain in the ass to get to and removing it will spray some coolant in that area. I highly recommend doing this only after the car has sat overnight/is completely cold.

 

[Rotus]

I have a 1992 NSX that I purchased about 7 years ago. I noticed that the temp gauge stabilized at 1 tick below midpoint and never went higher, even on the hottest days.

Since that time I have replaced the sender 5 times without satisfactory results. One sender sent the gauge to full hot, scaring the daylights out of me until I realized that there were no other indications of overheating. A laser-aimed temperature gun gave a reading of about 185 degrees. (At least that told me that the gauge will go full range.) Another sender stabilized at 2 ticks below full hot, rising to 1 tic below hot until the fan came on.

I finally gave up on parts house senders and ordered the proper Honda part. Now I've gone full circle -- the gauge stops at 1 tick below midpoint and will not move further, even to the point of the fan coming on.

I can't believe quality control on all those senders is that bad. I must be missing something. Suggestions appreciated!

 

[Larry Bastanza]

I have a 1992 NSX that I purchased about 7 years ago. I noticed that the temp gauge stabilized at 1 tick below midpoint and never went higher, even on the hottest days.

That is exactly what it is suppose to do, that means everything is working properly.

Regards,
LarryB

 

[Rotus]

Thanks for the response. You are more optimistic than I am. Based on earlier discussion in this thread and my own experience, it's apparent that the NSX temperature gauge is not computer controlled. Given that, I have never seen a cooling system that can maintain temperature that perfectly regardless of changing conditions.

With a genuine Honda sender in place, I'll be watching it as carefully as I can with hot summer temps on the way.

Thanks again.

 

[nikey22]

He's not attempting to be optimistic, he's just plain right.

please don't buy a 7th sender unit.......you're fine.

 

[Larry Bastanza]

I have never seen a cooling system that can maintain temperature that perfectly regardless of changing conditions.

After 40+ years of Honda car ownership, not only the NSX, EVERY car I have owned works exactly the same

 

[Valhalla]

OK a resurrection from the dead. Spent 4 hours working on this project today. Background story: Gauge reading properly then all of a sudden, nothing. It might move a smidgen but not accurate and no progress. I have to admit after reading Jink's engine issues I got a bit nervous. So I read this thread and ordered a new OEM sensor. Couple of days later it arrives. i go to take out the old and it slips out of my fingers (tight quarters down there) and while coolant is flowing out I'm trying to install the new sensor when it slides out of my fingers too. Where do they end up? Glad you asked, in the galley between the two heads! I am able to retrieve one of them (what a fiasco) and install it back in. I burp all four coolant bibs and still no needle movement. I decide to try and retrieve the other sensor, just in case that one is the new one.

I'm about two hours into removing the intake manifold when I am able to shine a flashlight into the galley from the rear and spot something shiny. I toss my lighted probe down there and as careful as a surgeon I extract something. It is a temp sensor! But it's the old one, so no difference by installing the new one. I again burp all four coolant bibs with the ass end of the car in the air, and still no difference. Radiator is getting coolant, radiator fan comes on, heater works in the cabin. Just at a loss as to my temp gauge. What might I be overlooking? Thanks in advance

 

[Old Guy]

Did you do the test in the service manual to confirm that the gauge itself is operational?

 

[Valhalla]

Ok, so with a little more light reading I find there are three sensors on the coolant system and the one I changed out might be for the radiator fan?! Back to the manual. So is this the sensor I'm looking for? Thanks Old Guy.

 

[Kaz-kzukNA1]

As you suspected, the one in your photo is for the rad fan control so won't do anything for the dash gauge.

You need to replace this one although you better check the gauge first.
Please be careful not to create GND/low resistance circuit for prolonged period as it could damage the gauge.
Best if you can connect about 50ohm resistor between the sensor connector and nearby GND.
Your gauge should be at just about middle/horizontal line of the temp gauge.

You must apply liquid gasket/Honda bond at the thread at the time of installation and in order to create dry mating thread, you will be forced to drain enough coolant again.

For your reference, these are the 3 sensors related to the coolant temperature;

Temp sender unit: for the temp gauge on the dash
Rad fan control sensor: obviously, for the rad fan
TW sensor: for EFI control


Kaz

 

[Old Guy]

I think its pretty hard to mix up the sensors. According to the service manual, the gauge sensor is a single terminal sensor (ground return through the engine block). Both the ECU and fan controller coolant temperature sensors are two terminal sensors (return path is not through the engine block / chassis).

That photo almost looks like two wires going into the plug; but, I can't be sure. Also, the wire color looks odd. I thought the single wire to the gauge temp sensor was yel/grn (or maybe grn/yel?). The wire color may have changed with the production year, so check the service manual to confirm the correct wire color for your year. However, the definitive test as to sensor type would be; 1 wire - gauge sensor, 2 wires - not gauge sensor.

Full disclosure - my temp gauge works just fine so I have never actually gone looking for the sensor.


Update: I see Kaz posted while I was responding so I think you have the definitive answer.

 

[Valhalla]

I really appreciate both or your replies. Much love to you both. A couple of the most respected people on this board.

Once I realized there were three sensors, I started to second guess myself that I had replaced the right sensor. Looks like I did. No change to the gauge function. Barely moves off the very bottom and just stays there.

I did check the gauge function after following the diagnostic check on SM 23-136 and the gauge does move all the way up so the gauge is working properly. The engine is cold and I put a voltmeter on it and it every time I check, I get a different reading anywhere between 104 and 190. Coolant temperature is 65F. Is there something I am doing wrong? Just to make sure I'm not crazy (debatable of course) I tried the old sensor and I get the same readings. They are all over the board between 125 and 197. Is this normal?

 

[Old Guy]

.

I did check the gauge function after following the diagnostic check on SM 23-136 and the gauge does move all the way up so the gauge is working properly. The engine is cold and I put a voltmeter on it and it every time I check, I get a different reading anywhere between 104 and 190. Coolant temperature is 65F. Is there something I am doing wrong? Just to make sure I'm not crazy (debatable of course) I tried the old sensor and I get the same readings. They are all over the board between 125 and 197. Is this normal?

When you say ' put a voltmeter on it' I am guessing that in fact you are using a multimeter and that you had it set to measure resistance and those are the values in ohms? If not, that is what you want to be doing. Also, just to make sure, you are measuring the sensor resistance with only the multimeter connected to the sensor?

According to the service manual, the sensor has a highly non linear change in resistance with change in temperature and the lower the temperature the more sensitive the sensor. I expect that at room temperature, just picking up the sensor and holding it in your hot little fingers could cause a significant change in measured resistance. The service manual specifies a resistance of 142 ohms at 56 C. My back of the napkin kind of calculation suggests that closer to room temperature, say non air conditioned 30 C the resistance should probably be around 250 ohms or more. Your measurement of a resistance of between 104 and 190 ohms with a coolant temperature of 65 F seems really odd.

If you have the old sensor out of the car and your multimeter has alligator type clips, connect the multimeter up to the sensor and let it sit so that it comes to what ever the ambient temperature is. That resistance reading should stabilize after a couple of minutes. If the ambient is around 30C, that resistance reading should be around 225 - 250 ohms or more. If it is a lot less, that suggests that the sensors may have a lot of - lets call it uncertainty rather than error - at the lower temperatures. Honda may not care about temperature gauge 'uncertainty' at low temperatures which may explain why the sensor has a much different response rate in the critical temperature range between 85 C and 100C. The acid test for the sensor would probably be to connect it to your multimeter and measure sensor resistance with the sensor suspended in a pot of boiling water. As long as you are not in Denver CO, that water will be at 100 C and you should measure right around 32 ohms (according to the manual). If you have a thermometer, you could suspend it in the pot and let the water cool to 85 C and the resistance should increase to around 49 ohms. If the hot water test gives you those results on the sensor, the sensor is probably as good as it is going to get.

Honda does not provide a calibration test for the dash gauge. However, if you want to test the operation of the gauge, I suggest the following. Go to Digikey or some similar electronics vendor and purchase a 50 ohm and 30 ohm resistor. Digikey lists about 100 million resistors, you want a through hole style with a 1% (or better) precision with 1 - 3 watt rating. Something like this:

https://www.digikey.ca/product-detail/en/riedon/UB3C-50RF1/696-1043-ND/2176599

If you are anal, you can buy a 32 ohm and 49 ohm resistor; but, those tend to be oddities and you will need to order a minimum of 1000 to get them. For testing, I think the 50 ohm and 30 ohm should be just fine. Connect the 50 ohm resistor between the gauge sensor wire and a good ground. Switch the ignition to on and the temperature gauge should be reading the equivalent of around 84 C which is probably slightly below normal. Turn everything off and replace with the 30 ohm resistor. Switch the ignition on and the temperature gauge should be giving you a temperature reading equivalent to around 101 C which I am going to guess should be approaching the panic mode. For interest, you could purchase an additional 40 ohm resistor and I am thinking that this should probably put you in the above comfort but not panic mode on the dash gauge.

If the dash gauge tests out OK with those resistors and the sensor passes the boiling water test, that is probably as good as it gets with the OEM parts. Something to note is that in the normal operating temperature range, the sensor and the gauge have a sensitivity of somewhere around 1 C / 1 ohm. The result is that if you have a bad electrical connection at the sender (or anywhere), it can cause a material error in what the gauge displays in its 'normal' range. Make sure the connections are clean and you may want to try applying something like Ox Gard to the terminal to improve the connection

 

[Valhalla]

Thanks Old Guy for the response and direction. Yes it was a multi-meter and I swear it would read a different value every time I touched it. I had it set at the 200 ohm register. Once I bounced the multi-meter up to 2k ohms the reading stabilized at 83 for both sensors. So I just left the new one in and hooked the wire to it. Just installed a new steering rack and needed to get the car to the alignment shop and lo and behold, the gauge read correctly. Just a notch below half, right were it should be. I'm thinking that the new sensor and the "exercising" of the gauge with the test might have woken the system up. Thanks for all the help, much appreciated.

 

[Old Guy]

If its working, don't mess with success.

Inexpensive multimeters can give flakey results. I like Flukes. More $; but, reliable operation.

 

[Jinks]

[MENTION=25419]Valhalla[/MENTION], I am glad to hear my tragedy had meaning for other owners. Also, Kaz is the nsx encyclopedia!!!
I will religiously keep my eyes on the coolant system now on my next motor!

 

[Valhalla]

There are a number of the older members (crap, I think I might be qualified now) that are always there to help us find our way if we are not to self righteous, arrogant or stupid. This site has always been one of the first I go to for information and sharing. I may not be proud of some of my past failures, but posting them here will provide insight to others, experience to all and a documented trail of what not to do. Those that do not follow history are doomed to repeat it.

 

[Valhalla]

And just when I thought it was safe.....

Drove the car to work this morning and the gauge/sensor is acting up again. It reads just above the bottom line with no movement. Once it cools down I'll try to pull the wire connector off the sensor pin and see if there is a bad connection. Of all the things that can go wrong with cars, electrical issues are the ones that cause me the most frustration. Anyone else?

 

[Old Guy]

In the lotto of automotive failures, between the annoyance of an intermittent electrical problem versus the definitiveness of something like a ball joint failure while driving down the road, I prefer to 'win' the intermittence of an electrical problem versus the definitiveness of the biggies :smile:.

If your other gauges were doing flakey things at the same time, I would be inclined to say have a look at the connections at the instrument cluster, particularly the 12 v supply; but, since you haven't reported that, the instrument cluster connections are probably OK. For me, the likely candidate would be the sensor connector under the rubber boot and the first 15 cm of wire after the connector. The wiring gets a fair amount of heat at that location and the insulation may be starting to fail. I would look really carefully at the wire where it enters the boot on the connector to see if you have broken strands at that location.

If inspection of the wiring yields nothing, I suggest doing the gauge test with the 50 ohm resistor. Connect the 50 ohm resistor between the sender connector and ground and then turn the ignition to on. If things are working, the gauge should come up to slightly below the mid point. Try tugging on the wire to see if the gauge needle moves. If it does move, then that suggest that you have a bad connection someplace and its down to tracing out the problem.

I am going to hazard a guess that the wire from the sensor to the cluster is not one continuous wire. I expect that it probably goes through a multi pin connector between the engine wire harness and a body wire harness and perhaps another connector between the body wire harness and the harness going up to the instrument cluster. The engine to body harness connector would be another good place to look for damage.

 

[Valhalla]

Old Guy, I wish I could agree. One nice thing about working with electrical is, it works or it doesn't. You know right away. The intermittent stuff will drive you just left of bat shit crazy. I'm just about there.

Tried the test on the gauge again and it ran all the way up when grounded so the wires there seem to be intact. Maybe it's the contact point(s) inside the rubber boot that aren't tight or just the way I'm holding my mouth, I'll try and adjust the boot and electrical connector and see if I can make a more "tight" fit. Short of replacing that little boot, I'm not really excited about running another wire to the gauges.

FYI, I did have the entire cluster rebuild by Brian. Been flawless since. Great service. #fingerscrossedagain

 

[Old Guy]

I was rereading Kaz's post, and noticed his comment about Honda Bond on the threads. Did you do that? Its late and my Service Manual is out in the garage, so I am not going to check to determine whether that is the SM recommended procedure. I have seen Hondabond and it is not electrically conducting. The gauge sensor uses ground return through the engine block so a poor electrical connection at the threads because of the Hondabond could affect gauge operation. However, its not a problem that I would expect to be intermittent. I would think the connection would always be good or always bad. However, you could test by using your multimeter to measure the resistance between the gauge body and chassis ground. If its a couple ohms or less, then the Hondabond is probably not an issue.

I really expect that the problem is somewhere close to the boot on the sensor connector. Maybe try applying some Ox gard on the sensor terminal and up in the boot to see if that improves the connection.

 

[Valhalla]

I did not use Hondabond. I placed the sensor in dry and there are no leaks and I would expect excellent grounding to the motor because of this. I'll pull as much of the wire and boot up as possible to see wht the internal boot looks like. The circuit certainly looks intact by how quickly the gauge rolls up as soon as the key is turned in the grounded state of the test procedure. Yes it does get quite warm in that location and possibly the connector has warped with the heat. I'll let you know what I find out tonight.

So tonight I pulled out as much of the wiring as possible to access the internals of the boot. As expected a lot of the boot rubber was dried and cracked. I cut about an 1/8th of an inch off the bottom and slid the rest of the boot up. Then I could see the copper clamp inside the hard plastic sheathing had warped over time and the contact patch had expanded to the point it no longer made contact with the sensor. I carefully pried and bent the clip back into a meaningful position and while reattaching to the sensor made sure it had a firm grasp. Finally.....success.