idle sensor

1993 Honda Accord EX idle sensor constantly surges. Recently replaced with
a used sensor, could this be the problem (needs brand new one) or could it
be an electric problem.

 

--------------------

If your rad has any air in it, (some of) the temp sensors will not be
emmersed when when the thermostat opens. Fill the reservoir at least to
the MAX line, and top the rad up to the top. Air will often burp out
when engine next cools and things will start working correctly. Don't
put tap water in your Honda (Hondacide). Honda premix coolant would be
best.

'Curly'

 

Classic symptom of a vacuum leak.

 

Yes, if the idle surges ABOVE normal speed and does not drop to normal or
below while surging.

A slight air leak will cause an elevated idle; a gross leak will cause
surging well above idle.

Low coolant in the head will fool the ECM into thinking the engine is cold,
since the temperature switch is exposed to air. The ECM then increases the
idle, then discovers things are still not right, so it cycles back and
forth unable to find the correct setting.

A thermostat stuck open can also cause a surging idle, for the same reason
as low coolant.

 

I suppose it depends on the car. My other-car Bosch L-Jetronic has a
fuel cutoff at high idle w/ throttle switch closed. (Safety feature.
Probably universal.) A gross air leak sends RPMs up, fuel is stopped,
RPMs drop, fuel is turned back on, RPMs rise, etc.

As for your low coolant theory, I feel any temp sensor in a running
engine is hot whether immersed or not, being efficiently heated by
radiative transfer inside a closed cavity. Like an oven.

I'll wager if you disconnect your coolant temp sensor, assuming an NTC
thermistor, so infinitely high resistance simulates low temperature,
your engine will not surge. Just my uninformed opinion.

 

Sounds exactly like what happens to both the cars I tried a gross air leak
on. I don't know enough about the Honda PGM-FI to know if it has a similar
cutoff, but I suspect it does, otherwise you'd run the risk of a throttle-
closed over-rev.

Could be. I'll admit to parroting what I've consistently read, which may
not necessarily be true. Parrots know not of empirics...

Worth a try, that is.

Mine's right below the distributor, so it's even easy to unplug.

I've got calls to make tomorrow, so I'll have her nice and hot. Be back to
you tomorrow with the exciting results!

 

no, it's definitely the real deal. i've worked on that issue a number
of times. the people that have the most problem with it are the ones
that think the sensor "should" transmit the right signal with the logic
shown above.

 

Well, I just tried disconnecting the TW sensor the way misterbeets said.

The idle did not surge.

I made sure I was unplugging the correct one too.

However, in addition to the Check Engine light coming on, the idle
increased from 750rpm to about 950. Then it began a slow, unsteady hunting
between 850 and 950. It would go down to about 850rpm, then unsteadily
increase to about 950, then down to 850. Total tach needle movement during
the hunting was less than 1/8".

It took about ten seconds or so for it to hunt up between 850 and 950, and
another ten to go back down to 850. If you did not have a tach and weren't
really paying attention to the engine, you'd easily miss the cycling. The
wild, violent swings I saw when unplugging the PCV valve were another story
entirely.

When I plugged the sensor back in, the car seemed to ignore it completely,
with the hunting idle and Check Engine light remaining on. Once I shut the
car off then restarted it, it was back to normal, at a steady 750rpm.

 

A commendable effort. On second thought, though, I'm not sure it proves
anything. If the ECU is turning on the Check Engine Light, it's
probably ignoring the sensor. The true test would be to progressively
reduce the coolant level and see whether surging commences. But, as I
mentioned, there's no reason to think it would.

First because--I still maintain--the sensor output is nearly the same
whether it's heated by liquid or hot gas, or by conduction via its
threaded mount in the cylinder head, or radiative transfer, or all
three. Certainly it's not cold enough to trigger fuel enrichment

And second because, even if this were not true, there is no feedback
mechanism I am aware of to bring the RPMs down if fuel enrichment were
to raise them.

 

....which I am not prepared to do on my car, having paid a substantial chunk
of change to replace a failed head gasket a few years ago.

It would be nice if someone with an end-of-life vehicle that was being
junked or rebuilt could perform this test to confirm your theory.

It sounds plausible to me. Confirmation by experimentation would be nice.

Based on what I've seen, this may be true. An air leak RAISES the idle in a
feedback EFI system. A carbureted car would see the idle DROP.

The two cars (one OBD-I and one OBD-II) I induced air leaks on were unable
to maintain correct idle speed with even small air leaks.

 

Could be. But if so, why did I experience the elevated idle and slow
hunting?

 

And second because, even if this were not true, there is no feedback

Stupid me. Of course there is: the RPM sensor which normally keeps the
idle stable. But why would a fuel enriched idle be unconttrollable by
this mechanism, and cause surging?

But I'm still pretty sure any temp sensor screwed into the cylinder
head but not in direct contact with coolant is simply not cold enough
to cause fuel enrichment.

 

look at the sensor dude, it's not reading from the thread conduction,
[steel is a lousy conductor anyway], it's reading from the central
insert. that needs to be bathed in fluid, and dense fluid too, not
aerated foam.

if you don't believe it makes a difference, drain half your coolant.
the motor will run fine, the gauge will not show overheating, but the
mixture will be /way/ rich, you'll get intermittent code 1's [obdc0],
significant loss of power and gross fuel consumption. and this effect
is not unique to hondas.