professorman

My wife has a 99 Accord LX ULEV Automatic Accord. This morning, my wife's temperature gauge gave an instantaneous spike from normal (1/3 temperature reading) up to H (hot) and back down to normal, in less than a second. It just spikes up, and falls back down really fast. She continued watching it, and the temperature gauge did this 2 times again within 5 minutes. I did not do it again after that. This occurred while she was driving on the interstate to work, after about 15 minutes at a constant 60MPH speed. Her total commute is 30 minutes.

A few months ago, she told me that her temperature gauge was wobbly, were it would wobble up and down slightly for a few seconds. This occurred on 3 occasions. That wobbly has not occurred recently.

The car exhibits no signs of physically overheating, and these instantaneous readings occurs so fast, it is hard to say that there is any problem with the coolant temperature.

What do you guys think is going on? How can I troubleshoot this? I am thinking that it is the gauge. What tests should I perform?

goldsy

The 99 Accord has 3 coolant sensors. One that sends info to the ECU, one that sends a signal to the cooling fans and one that sends info to the gauge on the dash. All 3 are known to break from time to time. My guess would be it's the sensor sending the readings to the gauge. I'm not overly familiar with the Accords and I'm not actually sure of the location of each. Neither one are a very expensive part though.

holmesnmanny

My guess is it's the temperature sending unit. It could be the cluster going bad but it's more likely it's the sending unit.

It's going to be located where part no. 14 is in this illustration. However, your part no. may be different depending on your car.

http://www.hondapartscheap.com/parts...er-pump-sensor

Also, just fyi, there are four temperature sensors/switches in the 99 accord as is the case in most accords.

DCFIVER

Sounds like air or possibly exhaust gas....

professorman

When the sending unit is going bad, can it cause the temperature to be 'higher' rather than 'lower' than normal? I would like that when it is failing, it might cause a low reading. I have no idea of how they work though.

tech8

This is how the temperature gauge works for the 4 cylinder:

Engine Coolant Temperature Gauge (’98-’99)
The engine coolant temperature gauge has two intersecting coils wound around a permanent magnet rotor. Voltage applied to the coils, through fuse 9 (in driver’s under-dash fuse/relay box), generates a magnetic field. The magnetic field, controlled by the coolant temperature sending unit, causes the rotor to rotate and the gauge needle to move. As the resistance in the sending unit varies, current through the gauge coils changes, moving the gauge needle toward the coil with the strongest magnetic field.

The coolant temperature sending unit’s resistance varies from about 137 ohms at low engine coolant temperature to between 30-46 ohms at high temperature (radiator fan running).

As long as the car really is not overheating and there are no large air pockets (so make sure there are no large air pockets) in the cooling system; basically the problem can be due to:
• Temperature gauge sending unit.
• The wire that goes from the sending unit to the gauge. If there is an intermittent short in this wire, such as being caused by wire insulation melted or nicked and the bare wire touching chassis/body ground, it will cause the gauge to move up.
• The gauge may be faulty. The gauge is separately replaceable from the rest of the instrument panel.

If you have a multimeter or ohmmeter, you can do a resistance test of the sending unit.

Unplug the Yel/Grn wire from the sending unit. Measure the resistance of the sending unit between the positive terminal of the sending unit (tip of the sending unit) and engine ground. When it is cold should be about 137 ohms. If the engine is at operating temperature, it should be about 30-46 ohms.



For info. only: (98 and 99 Accord 4 cylinders) Beginning with the 98 Accord 4 cylinder, there is no 2nd fan switch.

The 98-99 had the following:
• Temperature gauge sending unit
• One radiator fan switch
• ECT sensor for the ECM/PCM fuel/ignition data.

00-02 Accord 4 cylinders no longer had a temperature gauge sending unit, as the engine coolant temperature signal came from the PCM/ECM to the temperature gauge.

holmesnmanny

That's weird that the 98-99 doesn't have the second fan switch as it still shows up in the illustration.

thanks for the info

professorman

I can NOT located the temperature sensor part no. 14 37750-PH2-014 THERMO UNIT (DENSO)

I looked on the diagram, and I can not figure out where to find this, and can not see it on the car. My best guess right now is the remove the air intake and see if I can find it burried somewhere. Does anyone know exactly where to look for it?



Thanks! I will perform this test.

tech8

The sending unit is under the distributor area. That is what I was referring to for the resistance test.

professorman

Thanks for the picture. My engine was hot when I was looking and couldnt get to poke around adequately. My distributor is damp with oil. I suspect that I have a distributor shaft O-ring oil leak. In that location, it seem like the oil leak would cause a big problem and possibly mess up the temperature signal. I will do the resistance tests and see.

professorman

Uuuuugh..... So, I changed the temperature sending unit with a genuine Honda one, and I still have the same temperature spikes to Hot.

What should my next diagnostic step be? I am thinking now that its either the wire leading to the gauge, or the gauge itself. Is there a way that I can test the temperature gauge?

tim73

If you have access to a decent OBD2 reader, which shows live data, plug it in, and set it to live data. Then watch the ECT. This comes off a different sensor, if it goes to hot and back to normal in under a second, I would bet 98% it is an indication issue, not a true life. I would say either the gauge...the wiring or the sending unit has a glitch.

DCFIVER

Already mentioned the most likely reason,stop throwing parts at it. You either have a leak in your cooling system,or a blown head gasket....

professorman

Maybe You are reading a different tread than this one, but no where is mention of the items you listed as possible problems.

There is zero loss of coolant. There are no symptoms of physical overheating of the car. The only symptom is a split second rise of the temperature gauge and immediately fall back to normal. Happens in less than a second. A physical problem twin the engine would not happen, then solve it self that fast. This is something going bad with a reading. I want to fix the problem so that I can trust the temperature gauge.

I did the resistance test on the previous sensor, and it was out of the specified range slightly. When the temperature gauge works, it works dead on accurate, except for the few times it spikes and back down within a second. In a 30 min drive, it might do it 0-3 times. When I am running the car at idle, I never see the problem. My wife told me about it, and I saw it twice in the past few months, but she sees it a lot more on her daily commute to work. It only happens for less than a second. The chance of me catching it messing up while the car is sitting still at idle is virtually impossible. I have yet to see it do the problem during my diagnostics.

tech8

Is it spiking to Hot right away when you turn on the ignition to ON, even when the engine is cold?

If so:

1. Unplug the Yel/Grn wire from the sending unit. Measure the resistance of the sending unit between the positive terminal of the sending unit (tip of the sending unit) and engine ground.

2. Plug back in the Yel/Grn wire to the sending unit. Go inside the car and measure the resistance at the instrument gauge's Connector C (16P connector) Yel/Grn wire.

Compare the resistance readings at the sending unit and inside the car. Are they relatively similar?

DCFIVER

See post #4. Sorry if you are unable to connect the dots......
Well I obviously have no idea what Im talking about then.....continue in your own way




(Oh you should note however,that a "diagnosis" consists of much more than simply following a trouble tree. You should check all that is relevant to the sensor instead of dismissing them...)

professorman

The temperature does not spike to Hot right away. The temperature starts out cold, then when the engine warms up, it moves to the normal position like normal. As I say, I have never seen the spike under normal conditions, except the 2 times when I am riding in my wife's car, and she points it out to me, "see, I told you that it spikes". This happens randomly. I drove the car for 1 hr on Saturday, and I did not see any spike, however, I asked my wife today if it spiked when she was driving, and she says yes. She watches the temperature gauge a lot more than I do though, because she knows to look out for it. It possibly could have occurred while I was driving, and I just did not see it, because it spikes so fast and go back down to the normal position.

I did the resistance test for the sending unit before replacing it, and even though the gauge reads dead on accurate while performing the test, the resistance was consistently slightly higher than what the manual says. My logic tells me that if the gauge is pointing to the exact correct position when the engine is cold, as well as pointing to the exact correct position when the engine is hot enough to cause the fans to come on, then the gauge, at this instance in time is giving the correct reading, and thus the sensor seems to be working properly. I decided to eliminate the sensor being the problem. It is the cheapest of the 3 possible problems. The problem can either be the sensor, the wire, or the gauge.

Over a duration of 30 minutes, (that is 1800 seconds), the gauge gives a maximum of 3 spikes, with each spike lasting 1-3 seconds. If we take worst case scenario, that means that the gauge spends 9 seconds total in the Hot position. That means that gauge has failed for 9 seconds out of 1800 seconds, result in the gauge reading correctly 99.5% of the time. The probability of measuring and getting a faulty reading is very low, because the gauge spends 99.5% of its time giving the correct reading.

That sounds like a good test to do. I will do that test when I get to working on the car next.

I wish I had a spare cluster to swap out and observe it the problem is with the gauge.

professorman

I have finally fixed this problem. I realized that the temperature gauge reacted to me 'knocking' the cluster with my hand, so I realize that the problem lies in the gauge cluster. I took apart the gauge cluster to see if I could find anything, and I realized that the pins that goes from the circuit board to the temperature gauge were wobbly. I compared this to the pins that goes to the other gauge, and all the others were firm. I used a soldering iron to melt the solder joints, and upon cooling, they resealed themselves. I added a little extra solder to each. I did this for all the wobbly gauge pins. I re-installed the cluster and now EVERYTHING WORKS GREAT!!! I am so happy to have finally fixed this problem