No charge, no speedo

The battery tests fine with an AVR, that's all I have to test with, but it
should be sufficient.

Yes, revving the engine to 2000rpm with headlights, a/c, fan, and rear
defrost, voltage went off the scale, and the fuse popped.
It was a little slower, about 5 seconds.

 

Yeah, anything that's a short to ground should cause the voltage reading
to DROP, not jump up off the scale... that's pretty much guaranteed a
bad regulator, or something connected to the regulator (does it tie into
the car's computer anywhere??)

What year car is this again?

 

There is a green connector with multiple wires. It does appear to tie in to
the vehicle computer.

It's a 1998.

Spdloader

 

On your first point, it may not be possible to see the difference between
the
'short' (if there is one) coming on and the fuse popping. Both will see
the
voltage fall away.

On the latter point, I believe there is a compensatory signal applied from
the ECU when a heavy load item is switched on. It should be possible to
hear the engine load change when the headlights are switched off again
but revs shouldn't change.

It sure is a puzzle.

 

"> On the latter point, I believe there is a compensatory signal applied
from

Yes, at idle, there was a noticeable change in engine load while switching
on the electrics, and alternator output changed to meet the requirement, but
once the throttle was "tweaked", the fuse popped, and the engine note
changed back.

The new alternator won't be in until tomorrow (Thurs) afternoon, so, another
day to over analyze this, I"m afraid.

Spdloader

 

I don't suppose you've tried a substitute battery on the system?

 

No. I don't have one at my disposal, but disconnecting the battery
completely had no effect on the problem, so I ruled it out.

Spdloader

 

No, but either way, you still wouldn't see the voltage spike UP

I found an online manual for '95-'97 Civics... the wiring diagram there
shows a connection from the ECU to one terminal of the regulator...
without an actual schematic of what's happening inside the regulator
module, it's impossible to say for sure... the ECU lead ties with one
side of the regulator to the "low" side of the field coil, so it appears
it's entirely possible that a bad signal from the ECU, or even a short
in the ECU connection (either to power or ground, the latter is more
likely) could cause the regulator output to spike. That would at least
explain why the problem persists across three different alternators.

I love a challenge

 

We may be on to something with the ECU link... wonder how hard it would
be to grab another one from a wrecker and plug it in, see if that fixes
things? Hmmm... or... I wonder if a diagnostic analyzer hooked to the
ECU would show anything anomalous?

 

It sounds like you have the output and regulator wires mixed up. The
alternator's regulator should draw more power at idle and never blow a
fuse.

 

Um, nope, sure don't. Didn't read the whole thread, did ya?

Spdloader

 

Alternator taken off and tested, charged 20volts, 20amps on the bench, and
voltage regulator bad.

New alternator installed, worked perfectly for 5 mins, then fuse blew, and
I'm right back to square one.

Gonna go have a Mt. Dew now.

Spdloader

 

So something else is killing the voltage regulator(s - 4 of them now!),
which then of course isn't regulating the output properly. Going by the
schematic I found for the '97 Civic
(http://media.honda.co.uk/car/owner/media/manuals/CivicManual/pdf/23-a96.pdf),
that would most likely lead back to the ECU connection. The regulator
actually controls the voltage across the field coil of the alternator,
which of course affects the voltage produced by the rotor coils. From
the diagram, it looks like the ECU also "injects" a voltage at the low
side of the coil, which would allow it to alter the delta voltage in
addition to the regulator (note: I have a good background in general
electronics, but am not trained specifically in modern automotive
charging systems, so some of this is just educated guessing).

A short in that ECU lead SHOULD'T affect the regulator directly, as
there's a diode there that would block any reverse current flow if the
ECU lead were to short to ground, but maybe the '98's alternator is
different - without that diode, a short could potentially fry the
regulator, allowing full voltage to flow through the field coil and
cause the rotor coil output to spike. Or it might not even be affecting
the regulator - just shorting that lead would have the same effect (then
again, if it's testing overvoltage on the bench, the regulator probably
is fried - damn).

So there you have it. Short of looking at a proper schematic of the
'98, my best guess is still a problem in the ECU. The cooked regulators
are a symptom, not the root of the problem.


I'd love to see a schematic of how the regulator itself is set up
inside, it would help the diagnosis.

 

I wonder if there is something else in the circuit, i.e. sensors such as O2,
or something that could be shorted somehow.

I think I'll let it rest over the weekend and do some fishing. I think I can
do that right.

Spdloader

 

I don't think there'd be anything else specifically on that wire... the
sensors would have their own connections to the ECU.

Heheh... like I said, try checking with a wrecker, see if you can get
another ECU for cheap, and swap it in.

 

Here's an interesting snippet I found... dunno how much it relates to
your problem, but it's a bit more information on Honda's charging system...

"..... Returning to Hondas - a no-charging system may not be the
alternator. Honda has been using an ELD (electrical load device) relay
for the last 10 years or so. Honda uses PCM strategy to control the
voltage regulator. We have learned to let the engine run a few minutes
before checking alternator output (the PCM lets the initial start up to
stabilize before turning the alternator on, reducing the load on the
engine). Honda uses the ELD relay to control alternator output -- no
load, no work at idle.
...... The typical customer complaint will be they are noticing the dash,
or headlights, dim with turn signal use or brake application. Check the
ELD before condemning an alternator.
...... And, as a reminder, check the dash warning light -- it needs to be
working before the alternator will work. Heck, as long as I am there --
Honda needs a functioning dash light, but it is grounded by the
alternator -- so... for a no-charge, check the bulb. If it doesn't
light, go out to the alternator and ground the appropriate wire -- if
the bulb lights now, proceed with the rest of the charging system check.
...... In a nutshell, there it is -- a Honda may not charge because of a:
bad alternator, bad dash bulb, bad ELD, or a bad PCM (in that order)."

http://www.asaaz.org/canyonauto/techtips.htm

 

Okay, forget what I wrote before about how the alternator works... I had
it kinda back-assward.

I found a great site detailing how alternators work, and how their
regulators work, and how it all works together. I'll hafta go back over
all this and re-think it a bit, but in the end, it still sounds like
it's probably something related to that ECU lead - either a short, or a
problem with the ECU itself - that's cooking the regulators.

Anyway, for anyone else who's interested:
http://www.1stconnect.com/anozira/SiteTops/energy/Alternator/alternator.htm

 

Thanks for all your work and help Matt, I'll get to a salvage yard right
after some fishing.
Spdloader

 

Oh yeah, fishing definitely comes first!

 

It does after the week I've had!

Later,

Spdloader