Debugging a Funny PGM-FI EACV Problem ('89 1.5)

The short version: What, on an '89 DPFI 1.5l engine, does the ECU use to determine target EACV opening and idle speed? What would cause the ECU to suddenly decide to command a full-open EACV without also indicating an
error?
The long version:
I've got an '89 Civic Wagon, 1.5l DPFI, 2wd, 5sp, and its idle speed bounces randomly up and down between normal and 1-3k RPM over normal.
The "check engine" light is off, and the ECU displays no trouble codes. The EACV has been cleaned and then replaced (twice!) and the ECU once with used and known-good parts, and the symptoms are identical in all cases.
We've spent some time driving around with a portable oscilloscope wired up to the EACV and various sensors. When the idle shoots up, it's because the ECU is actually commanding the EACV open. Readings from the MAP sensor, TPS, ELD, and O2 sensor look perfect compared to the graphs in the Helms manual while it's doing this. We've made briefer checks of the intake air and coolant temp sensors and they look OK; more importantly I've driven with them disconnected and the periodic idle jumps still happen.
Running the car with either the EACV disconnected or with its bypass air hose plugged results in the expected low idle and no other problems except for possibly a 15% fuel economy hit. (What with rapildly changing gas formulations, though, it's hard to be sure the fuel economy issues really stem from the engine control problems.)
My best guess is that there's some intermittently failing sensor that's important enough to the ECU at idle to make it want to whack the throttle open. The thing is, outputs from what seem like the obvious culprits (MAP sensor and TPS) check out fine at both the sensor and the ECU harness. Wiring and vacuum hoses have been thoroughly checked. The RPM/crank position sensor hasn't yet been checked during operation, but my understanding is that they usually either work perfectly or fail badly enough for the ECU to throw an error, but it's on the agenda for the next time we have a few free hours that aren't rainy.
Any educated guesses would be deeply appreciated. Anyone who can provide a fix gets both my undying gratitude and (if you want it) a free bonus used-known-good EACV!
thanks, andrew
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
andrew m. boardman wrote:

hmm, the #1 cause on this group is insufficient coolant level. check inside the radiator - if there's the smallest leak, air sucks back rather than coolant, so the expansion bottle level stays the same. insufficient coolant level in the block means the coolant foams and gives spurious temp readings to the ecu, so it hunts back and forth between "cold" and "normal" idle behavior.
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
jim beam wrote:

also, make sure your "hard" idle setting is correct - disconnect the eacv plug and set idle for 750 with the throttle body adjusting screw, just like it says in the honda manual. the eacv provides supplemental air in addition to the "hard" idle setting - it's not a substitute.
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

Yeah, I spent some time reading through the archives and was impressed by all of the ways that low coolant level can make the engine misbehave. Unfortunately, the coolant level is good, overflow isn't dropping after use, and it's been bled by the book.
It's also jumping around very sharply; the EACV signal will jump from 3.5v or so up to 9v, sit for a bit, then back down to 3.5v; I don't think a temperature sensor could react that fast. More to the point, though, it does this even with the coolant temp sensor disconnected. (Voltages quoted are the difference between the two EACV wires; it gets a constant +12 plus a control line from the ECU which varies between +12 and ground depending on how far the ECU thinks the EACV should be open.)

What might also be important here is that the engine is being revved up way way beyond what might be called for even with a very cold engine; seeing it pop up to 4k RPM is not unusual. I suppose there's some chance that flaky coolant temperature sensor wiring is making the engine look really really really cold, but not far enough out of spec to flag the sensor as failed. (It does flag it as failed when unplugged, though.)
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
andrew m. boardman wrote:

but it should! level in the expansion [overflow] bottle should go down as it cools. and you /did/ check inside the radiator, not just look at the bottle? need to be sure...

sure it would!

ok.
ok, but that's output - you need to find the input that's triggering this response from the ecu.
i read before that you disconnected the temp sender, but there are three on this vehicle: one for the radiator fan, one for the gauge on the instrument panel and one for the ecu - just check you have the right one. and i'd test the ecu output with a potentiometer wired in place of the sender. it's supposed to range between about 20k and 100 ohms, with lower resistance being the higher temp, so you can read the effect it has on ecu output accordingly. that area of the head tends to get badly fouled with oil leaks from the distributor, so check the quality of the connection as well.
if still no dice, i'd check the electric load sensor.

that just corresponds with open eacv.

right, but the system also tolerates a degree of noise, so a full disconnect is different to something fuzzy or high resistance.
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
| Any educated guesses would be deeply appreciated.
Air Boost, Fast Idle & EGR valves, disable 1 @ a time & see. www.autozone.com/images/cds/gif/large/0900823d800cf4a6.gif
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
snipped-for-privacy@bronze.lcs.mit.edu (andrew m. boardman) wrote in

You have a giant air leak.
I don't know if you have a Fast Idle Valve, but on multi-port cars this is the prime culprit.
--
TeGGeR

The Unofficial Honda/Acura FAQ
  Click to see the full signature.
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
"TeGGeR����������������������" wrote:

yes, but the ecu's commanding the eacv to do stuff. it's got to be one of the input sensors...
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload


It's a big air leak. That means one of the secondary air inputs into the intake is bad. Or there is an input that is not supposed to be there (something split, etc). The ECU is feeding gas to keep the mixture correct, then cutting fuel when the RPM limit is reached, then repeating itself...
Another diag tool I forgot to mention: Partially cover the throttle body with your hand. Cover more and more until the idle settles down. If this works, something is admitting way too much air.
With any port-injected car, the thing to do is to cover the lower port when the car is warm and idle is fluctuating. If vacum is present at that lower port and idle settles down when you cover up the port, then that's the problem.
--
TeGGeR

The Unofficial Honda/Acura FAQ
  Click to see the full signature.
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
Thanks again for the replies; sorry about the long delay in answering the latest. (Some unexpeted travel, then some unexpected getting-sick.) I'm back under the hood of this thing, though, so here are replies to suggestions that were offered while I was in la-la land:

It's within spec, though that spec is 625rpm (+/- 50) in my Honda manual with the EACV unplugged and no load, 750 with the EACV plugged in. (Well, sometimes it does 750 just fine, sometimes it does 1800+.)

Sorry, I meant it wasn't dropping between checks with the engine cool. The overflow bottle fills up an inch or two when the engine is hot, then goes back down when it cools. I did indeed check the radiator (full up) and the bleed point (bleeds, no bubbles) after driving. I've also just put another thousand miles on it, which I'd hope would chase any bubbles out.

I do believe so. "TW sensor" per the book, 20k ohms cold and a few hundred hot, just under the distributor, six blinks on the ECU when it's disconnected. (But even disconnected and blinking, the ECU pops the EACV mostly open from time to time.)

Nope, no Fast Idle Valve. (Also not multi-port; throttle-body injection with stacked main and and aux injectors.) I do have a dashpot diaphragm that cracks the throttle during cranking, but that's not misbehaving either.
New data after 1k+ miles: I'm really confident the coolant system is bubble-free. Mileage is definitely down the toilet, though, so it appears that the ECU is not only opening the EACV but increasing injector opening as well, which makes me think one of: - throttle angle sensor (ECU is seeing "full open" when it's not) - crank position sensor (ECU is seeing "high RPM" when it's not) ...even though neither of them is far enough off for the ECU to indicate an error.
Side question: the Honda manual says "replace throttle angle sensor" if it's bad. It doesn't have instructions for doing so, though, and a quick look at the sensor on the throttle body shows what looks like a pair of rivet heads. Is this actually replacable without swapping the entire throttle body?
I've got a friend coming by with a multi-channel digital recorder this weekend; we'll wire up *all* the ECU inputs plus the EACV control and see what corresponds to the EACV getting thrown open.
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
andrew m. boardman wrote:

wow, that's serious stuff! i love toys like that. keep us posted on how it goes.
regarding throttle angle sensor replacement, it's actually quite easy. if you're skilled, you can even remove it without removing the whole throttle body. use a chisel to nick the edge of the shear head bolts [the things that look like rivets], then use the chisel at an angle on one edge to tap them free. then they'll unscrew by hand. i replaced mine with m5[?] allen bolts. carefully mark the position of the old tas before removal, obviously. there's a more detailed write up here that google will reveal.
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
Oops. This is pretty old news at this point, but better late than never. (I found many a similar problems in the archives, but there was never any followup once it got fixed; perhaps this will help someone down the line...)
jim beam wrote:

Sadly, "how it went" involved my friend's nice laptop blowing up, so I don't have any lovely eight-color graphs of engine sensors over time to put on the web.
We did, however, find the problem: it was the throttle angle sensor after all. The reason we didn't catch it the first time is that the arm had shorted to the +5 side, but the resistance readings between the ground side and the arm (which is what we checked) were perfect. Electrically, this didn't raise any error with the ECU because everything that was being reported was within spec, but it was all compressed into the upper part of the range of valid TAS readings. Thus, the ECU thinks I've always got the throttle way open and opens up the EACV and dumps in lots of extra gas to suit.

Indeed! It was a bit of a lousy angle, but I got in there with a Dremel and it took only moments. I'm at a loss as to why Honda doesn't consider this a replacable part.
Sadly, the bottom end of the engine started falling apart after another few thousand miles, but I'm not completely unhappy with getting 380k out of it. (One owner since new, no real work beyond the occasional valve check and oil changes every 8k or so.) The same throttle body is on the new engine and is working perfectly. (I may get around to posting pictures of what the bottom end looked like once I have it apart; while I went for a new engine I've saved the old one.)
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
andrew m. boardman wrote:

great followup andrew - closing the loop on problem solution is a wonderful thing to anyone searching the archives. also, i note the fact that you're standard transmission, not auto. i suspect that shifting wears the tps a lot more on the sticks than the autos.
shame on the old motor - what replacement did you use? jdm?
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
replying to andrew m. boardman, ryhatch1990 wrote: Loosen the throttle plate nut that hold the accell cable. Back it off till its loose. Now tighten with your fingers as tight as you can. Spray it with some lubricate.
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

Related Threads

    Motorsforum.com is a website by car enthusiasts for car enthusiasts. It is not affiliated with any of the car or spare part manufacturers or car dealers discussed here. All logos and trade names are the property of their respective owners.