Gregg Gruen says...

> My 1994 Accord seems to do something strange at highway
> speeds. It seems that the torque converter is locking
> up and unlocking repeatedly. I try to maintain a very
> steady speed, but I'll feel it unlock and watch the tach
> jump up ~250 RPM, then watch it drop it back down again.
> This will happen on flat road, but very much more so on
> a slight (and I mean slight) incline. If I use cruise
> control, the problem it's so bad, but still present. It
> seems like it's more sensitive than it should be. It'll
> just as much as every 5-8 seconds, especially if the
> road isn't perfectly flat.

I too have a 94 Accord, and had a problem with the converter
working sometimes, but not others, but I didn't have the
short-term switching you seem to have.

In my case it turned out to be a temperature problem, not a
transmission problem at all. The computer will allow the TC
to engage only if the engine is FULLY warmed up. When I
changed out the thermostat, that fixed it.

In retrospect, though, I think there is a chance that the
cooling system just needed to be bled of air, so that the
coolant temperature sensor stayed fully immersed. I
started having problems after the coolant was drained and
replaced at 30k miles, and my guess is that the system just
wasn't bled fully. Anyway, I DID bleed it properly with the
new thermostat, and it's worked fine ever since.

This is going back several years, so I don't remember all
that well, but someone here told me how to do an important
test to see where the problem was. There are two solenoids
on the transmission which control the lockup converter, and
each of them is controlled by a pair of wires. I soldered
a straight pin to the end of each wire of a length of
regular four-conductor telephone line, and inserted the pins
down into the control line connectors so as to make contact,
and then brought the phone line into the passenger
compartment.

Then you can read the voltage across each control pair with
a volt meter, or hook up an LED across each pair in series
with a few hundred ohm resistor. When the computer is
trying to engage a solenoid, the voltage across that control
pair will go high (up to 12VDC), and your LED will light up,
but the voltage will be low otherwise.

As I remember, it is a three-stage process as your speed
increases. First one LED comes ON, then the second one
comes ON, but it flutters, as though it's only ON half way.
And then it goes solid ON when the TC is fully engaged.

The idea is to find out whether the solenoids are not
getting the control signals to turn on in the first place,
or whether they are getting the signals but are not engaging
properly. In my case, I found that the signals weren't
being sent in the first place, which pretty much locked down
the coolant diagnosis.

Your problem probably isn't the coolant, but you still might
want to make that test. There's no point farting around
with the transmission if the problem is with the control
signals not staying on.

Hope this helps.