I would like some opinions as to diagnosis of a car problem, specifically
in the cooling system. When cold, the engine runs normally and boosts well.
However, once it's warmed up, certain things occur which seem strange:

* The engine temperature cycles up and down. It gradually warms up to about
  210 degrgees F, stays at that temperature for a while, then drops rapidly
  to 170 degrees. Then it gradually warms back up and repeats the process.

  (Presumably (and hopefully) it is merely the sensor that is thermal-
   cycling the full 40 degrees, rather than the entire engine block.)

* After the temperature has been at the low point for a bit, and while its
  on its way up, again the engine runs normally.

* While the temperature is dropping, however, the engine bogs down and
  produces only about half the acceleration that it should. After a few
  seconds of this it will often kick back to full power in one or two
  second spurts, then only be bogged down for a second or two at a time.

What is most likely wrong?

184 responses total.

Sounds like the fan is turning on when the engine gets "too hot" and stays
on until it gets "cold" again.  It also sounds like the fan is drawing a lot
from the battery/engine.  Could be a problem with its bearings.

It sounds like you may have a problem with your thermostat, possibly combined
with a leak in your cooling system that is introducing coolant into the
combustion cycle.  Do you notice big white clouds coming from your exhaust
when the engine bogs down?  That's a sign of coolant getting into the
combustion chamber.  The car's thermostat sits between the engine block and
the radiator, and only opens when it reaches a certain temperature.  If the
thermostat is sticky, or the wrong kind was installed at some point, it could
exhibit the type of behavior you're seeing.

It's not too difficult to replace, so that'd be one of the first things I'd
try, but the engine sputtering when the thermostat is open (apparently) is
not a very good sign.  It cost me $1,200 to fix that problem on my Sable when
a narrow piece of head gasket gave out, admitting coolant to the number one
cylinder.

I just had the same symptom, of the temperature gague cycling between all
the way hot and 3/4 hot.  The problem was apparently that my radiator
leaked, so there were air bubbles in the coolant system which caused the
temperature to go up and down.  (Sorry, my car knowledge is very limited, so
I am just repeating what I was told.)  The solution was a new radiator.

Head gasket leaks can also cause air bubbles in the coolant, as the combustion
gases are forced back through the leak into the cooling channels.

Mike Peltier probably nailed this one.  Thermostat.  But if when cold
you pop the radiator cap and view something that looks like chocolate
milk instead of clear bright flourescent green then you have a problem
where oil is mixing under temp/pressure with the coolant and is an
indicator of a major $$ problem.  Another thing to do is to check the
PCV valve.  If this is an 'automatic transmission' instead of manual 
then the check the state of yer tranny fluid.  Take out the dipstick and
touch the point to a piece of newspaper.  If it isn't a nice pink color
and has a 'burned' smell then at least replace the fluid.  (Likely as
not last winter you got stuck and not knowing what you were doing you
are looking at major tranny repairs.)

The PCV valve is the "positive crankcase ventilation" valve, that takes gasses
from the crankcase and mixes them with the engine's intake in order to burn
them off.  It's an emissions control device.  If it's stuck open, you're
getting a constant stream of oily crankcase gasses in the intake, when
normally it only feeds when the engine is running in a state that it can
handle the gasses without a significant loss of performance.

Re #6:
    Your description of a stuck-open PCV doesn't match my experience, as the
engine runs fine when cold.

Re #2:
    I had thought of the thermostat, though it can't be sticking continually
closed as I do get ample heat out of the vents. There was some leakage at
the high end of the cycle apparently out of a radiator/coolant bottle cap
with a broken gasket. I've replaced with a new cap.

    Radiator fluid and transmission fluid do require topping off from time
to time. But I haven't seen any chocolate milk in the system, and the oil
level remains adequate during the course of an oil change interval. The
transmission fluid looks like normal transmission fluid, and I don't recall
getting stuck. There is some white stuff coming out of the exhaust, but I
see as much coming out of most of the cars around me.

    Now to a somewhat harder part.

    A couple days ago I was looking at thermostats, and according to the
manuals I have a choice of at least three different temperature settings.
Part numbers for 160 degrees F, 180 degrees, and 195 degrees are all listed
for this engine, with the 195 degree part number in bold print indicating
that this is supposed to be the original factory installed 'stat. (There may
have been a listing for a 210 degree 'stat but I'm not certain of this.)

    So which temperature should I choose?

    Recall that the cycling is between 170 degrees and 210 degrees. The
"preferred" stat would be near the top of the cycle. But I've also read of
problems with too low a temperature like the engine computer not getting
good information and trying to compensate with extra-rich fuel mixes.

re #3:
Mark, you were probably sold a new radiator that was not needed.  If you ever
have radiotor problems, I strongly recommend checking out A&B Radiator on
Jackson Road.  They repaired a very damaged radiator of mine for $20 when a
replacement (part only) would have been $300.  And the repair was flawless.

It would help to know what kinda car it is and how many miles it has.  Hell,
it could be a bad sensor.  My sensors were fooled by a bad timing chain, adn
refused to fire the plugs at the right time when I tried to accelerate.

Who would have thought a bad timing chain would fool the sensors that regulate
the spark?

Some thoughts:

Re #0: Is this one of those cars where the engine is higher than the 
radiator, and the cooling system has to have the air bled out of it?  A 
bubble trapped in the cooling system could cause your temperature 
cycling, especially if it formed around the gauge sensor.

Re #5: Also check the engine oil dipstick.  If it has a lot of white 
and frothy stuff on it then coolant is getting into the oil and you 
probably have a bad head gasket.

Re #7: On most cars, a closed thermostat will let coolant run through 
the heater core, so you'll still get heat.  It just won't run through 
the radiator, so a thermostat that's permanently stuck closed usually 
causes overheating.  A thermostat that's stuck open will give lousy 
heat because the engine will never warm up properly with full flow 
through the radiator all the time.

A good thermostat should *not* allow the engine temp to cycle through a 
40 degree range, I don't think.  And 210 degrees is a bit high for a 
thermostat setting.

I'd try replacing the thermostat, then topping off and bleeding the 
cooling system.  If that doesn't help, you may want to pay attention to 
when the radiator fan is cycling on and off (if it's got an electric 
one.)

Bruce, what is it you think the timing chain DOES?

There is probably no more obvious a result of a bad timing chain than bad 
spark timing...

Re #9:
   1993 Pontiac _Grand Am_ 2.3L 4cyl.

Re #10:
   The top of the radiator proper is below the level of the 'stat. The highest
point in the system, however, is the coolant reservoir, which takes a pressure
-bearing cap unlike most coolant reservoirs. The "Full Cold" line is near the
top of the bottle.

    As for the cabin heating system, I've seen both technical diagrams and
the hoses protruding from the 'stat housing. It seems impossible to me for
anything on the engine side of a closed 'stat to get to the hose feeding the
heater core. There might be another route, but I doubt it.

    Replacing the 'stat is what I'm planning, next time I have a couple of
free nights in a row. I'm still at a loss, however, as to which temperature
to select. I'd rather do this job only once.

I was being facetious other...

I dunno.  I gave up on modern cars when they gave up on points and gaps.

nothing like rebuilding a the carb on a 3 on the tree
'81 f-100...it was empowering.

i invented swearwords that day but didn't need to go to a
mechanic. 

Re #12: I'm not sure the coolant recovery bottle is sufficient to bleed the
system, since until the pressure gets high enough to open the radiator cap's
relief valve it's not connected to the rest of the system.  Are there any
bleed screws for the cooling system?  If you have to bleed the system the
service manual will explain the process somewhere.

Which thermostat you choose probably won't matter that much, really.  I'd go
with the recommended (original) value, which I think you said is the 195
degree one.  Installing a thermostat that's too cold will sometimes cause
lousy running and bad fuel economy by preventing the engine from getting
warm enough for the ECU to go into "closed loop mode".

Does anyone know how accurate the speeds reported by GPS units are?  I'm
wondering if one would be accurate enough to use to check the calibration of
my car's speedometer, assuming I'm driving at a steady speed.  (I understand
there'll be some lag between the GPS reading any my actual speed.)

It is more accurate than your speedometer. As you observe, however, you
must drive at a steady speed as the cheap handheld units take a few
seconds to calculate your speed from the doppler shifts. 

I tried it (using a hand-held tape recorder to note the readings, so I 
wan't trying to drive and write at the same time.)  I plotted the 
results on a graph and found that my speedometer's response is linear, 
but it has the wrong slope.  The faster I go the farther off it is, 
percentage wise.  I'd suspect this was a symptom of using the wrong 
tire size if I didn't know for a fact the car has the original wheels 
and proper sized tires on it.

Maybe Honda just doesn't want to get sued over speeding tickets.

Until that list sentence, it wasn't clear which way the error went.

Sorry.  The speedometer always reads high.  For example, at 45 mph, the 
speedometer reads 48.  At 70 mph it reads 75.

That makes sense. When I went from a Yugo to a Honda, my speeding ticket count
went WAY down (I think I've gotten two the whole time I've been driving a
Civic, which is something like five years, as opposed to the five or six I
got in the five years I was driving a Yugo). I'd attributed it to cop biases,
but it would also make sense that Hondas deliberately miscalibrate their
speedometers at higher speeds so that speed demons aren't doing as fast as
they think they are.

I suspect my Saturn may be doing something similar: at 70, people blow by me
like I was doing 60.

Re #22: Technically, mine is miscalibrated at *all* speeds...it's just 
that below 45 mph the error is small enough to be negligable.  People 
will certainly notice if you're going 65 in a 70 zone, though.

This would have other benefits for a car maker: Gullible people will 
think the car is faster than it is.  Also, if the odometer also 
advances faster, people who lease will return their cars more often or 
pay higher milage fees.

I would think that deliberately miscalibrating the odometer (in either
direction) would be a violation of federal law. Deliberately miscalibrating
the speedometer, in contrast, could be taken as a safety feature.

It's also possible that Honda just doesn't calibrate its speedometers well,
or that your specific speedometer is messed up. (Ditto for Saturn.) And
certainly with all meters there's an acceptable amount of play.

I think the law says that odometers have to be within +/- 10%, which is 
a pretty wide range.  I remember way back in _Unsafe at any Speed_, 
Ralph Nader noted that somehow they all manage to end up near the top 
end of that range, and almost none end up reading *less* than the 
actual milage.  Kinda makes you go "hmm..."

I know my Yugo was way off, maybe even over that 10% mark (it would register
1.1 miles every mile marker, when I was watching). But I took that as
incompetence more than anything else (Val has to have her speedometer/odometer
cluster replaced three times, and even then, it didn't work quite right, in
her Yugo).

If you want a real accurate speedometer on your ride, get yourself a Sigma
Sport BC 800 bicycle speedometer.  You can enter the wheel circumference to
within 1mm and it will read in either MPH or KPH.  It will also give you
odometer function, trip odometer, max speed, average speed and time.  You
can't use any bicycle speedometer as most of them stop working around 50MPH.
For some reason Sigma decided that the BC 800 would go up to 183 MPH!  ~$25.

And how does it sense the periodicity of rotation in order to calculate 
speed?

There's usually some kind of magnetic sensor which gives one pulse per 
rotation.

I suppose you could mount the magnets on one of the half-shafts 
somehow, and rig a bracket for the coil, if it works like I'm thinking 
it does.  Car tires flex a lot, so you'd want to find the circumferance 
by doing a 'roll-out' instead of by measuring the wheel.

   Drew, if you haven't done the work already, go with the 195-degree
thermostat.

I've got the 'stat - 195 degree as recommended. I made a try at installing,
stopped and put everything back together after having problems disconnecting
the heater-core hose. It's impossible to get decent leverage in there...
I might try again in a couple of weeks.

If the hose is a touch warm, it's a lot easier to work it off.  You can
get some large-mouth pliers, like channel locks, and wrap some newspaper
around the hose and gently squeeze right where the end of the sleeve
the hose is mounted on ends, and it'll usually ease its way off.

If all else fails, buy a new hose, then slice the old one lengthwise 
where it goes over the fitting and peel it off.  (Try not to gouge the 
fitting, though.)  I've seen rubber hoses practically glued to their 
fittings before.

I've gotten the hoses off and the housing removed.

Now is there any good quick way to clean off the remains of the gasket?

Razor blade.

Sandpaper.

In other words, no, there's no quick way.  You gotta scrape it off. ;>

- Backtalk version 1.3.30 - Copyright 1996-2006, Jan Wolter and Steve Weiss