Purge solenoid INSANITY!

[GeneralDisorder]

Ok - I'm throughly fed up with this purge solenoid code.

 

Background: EA82 SPFI. Runs decent, no codes in U-Check. Idle is around 800 - 1000, and I can't get it any lower for some reason.

 

D-Check pulls a code 35 within just a few minutes of driving.

 

I've tried:

 

New solenoid.

 

47 Ohm 10 watt resistor at the connector for the solenoid under the hood

 

47 Ohm 10 watt resistor under the dash on it's own power supply, routed directly to pin 5 at the ECU.

 

68 Ohm 10 watt resistor under the dash.

 

I went back to a fully functional solenoid just a bit ago, and reinstalled the canister and hooked it all back in as stock. No love.

 

I've got the code in every single situation I've tried so far with 2 different ECU's.

 

Ignition on, not running I have 12.3 volts at the solenoid, running I have 13.4 at the solenoid.

 

When I had the resistors in place, both under the hood and under the dash the resistor would heat up so I know current is going through it.

 

I've tried clearing the codes, and when I put it in read mode I get the flashing CEL that tells me there aren't any stored. And it still comes back under D-Check.

 

What gives? I'm getting just a little pissed off.

 

GD

[somick]

I guess you have eliminated everything but wiring.

Try to run a new wire directly from the ECU. The old one in the harness may be broken and when you start driving it just looses connection.

 

Sam

[GeneralDisorder]

I guess you have eliminated everything but wiring.

Try to run a new wire directly from the ECU. The old one in the harness may be broken and when you start driving it just looses connection.

 

Sam

 

I did that - that's why I put the resistor under the dash - to eliminate the harness, and minimize the wire length.

 

GD

[keltik]

Is the purge solenoid operation dependant on anything else? i.e engine temp or TPS? Maybe its one of these other variables messing with the ECU.

[hooziewhatsit]

sounds just like the EGR solenoid on my car :-/ (88GL, SPFI)

 

I've even opened up my ECU looking for anything obviously wrong; didn't see anything.

 

I've also checked continuity in the wires from ECU to solenoid; checked out fine.

 

I've given up, and now check for new codes from time to time

 

-Dave

[GeneralDisorder]

I've been wondering that myself, but without knowing the interation in the ECU software, I'm forced to rely on the ECU's diagnostic. I have a feeling it's related to not being able to lower the idle speed even with the IAC screw all the way in, and that could be the coolant thermo-sensor. But that's just a feeling. I've tried a number of thermo-sensors without any real change.

 

I've actually replaced everything in the SPFI system at one point or another. I'm just starting to get it to behave, and this purge code comes up that I can't get rid of.

 

I'm about to just build a mega-squirt and screw the subaru SPFI. I can do it for about $400.

 

GD

[GeneralDisorder]

I've tried 2 ECU's, so I don't think the ECU is at fault.

 

I *thought* the donor car I got this engine/manifold from was good, but it turned out that someone just did a really good job with some black paint over the CEL.

 

GD

[daeron]

gd, you realize you just pass 6,666 posts?

 

I have been thinking about starting a pool on how long it was going to take you to go MS. i just got that vibe from you a few months ago when you were going on about something with your SPFI, i think this solenoid issue.. i thought, "how long until he throws it away, and megasquirts it?"

 

thats my two bits. my impression of who you are, im surprised you havent done it already.

 

I can't wait to get into it myself on the Zcar, but that miles down the road...the Zcar, she must run again and have at least all one color primer before anything along THOSE lines starts happening. so much time & money, and so little to do.....

[GeneralDisorder]

Yes - at least with MS I could tune it how *I* want, not what subaru thinks the consumer would want. I really want the stupid thing to idle back *immediately* after I let off the throttle. That, more than anything bugs the hell out of me.

 

I guess comming from carb land the SPFI is maddening to me because I have so little control over the system. I am reduced to a monkey throwing parts at a computer I can't configure, and don't even know the algorithms used in it's software. With MS, I'm sure I'll have a whole host of other problems to contend with, but at least it will have a minimal set of sensors, and everything else is software. Literally it's got a MAP (no MAF yay! ), coolant temp and intake air temp (same GM sensor for both - $8 each), TPS, and 02. It doesn't need anything else. And all the components are off-the shelf. I could even use a different throttle body for more low-end if I wanted. And of course no emissions systems of any kind.

 

I AM going to MS, it's just a matter of money right now. Being unemployed sucks.

 

GD

[bgd73]

I've been wondering that myself, but without knowing the interation in the ECU software, I'm forced to rely on the ECU's diagnostic. I have a feeling it's related to not being able to lower the idle speed even with the IAC screw all the way in, and that could be the coolant thermo-sensor and fan. But that's just a feeling. I've tried a number of thermo-sensors without any real change.

 

I've actually replaced everything in the SPFI system at one point or another. I'm just starting to get it to behave, and this purge code comes up that I can't get rid of.

 

I'm about to just build a mega-squirt and screw the subaru SPFI. I can do it for about $400.

 

GD

 

My 1993 did same things, same codes. A bizarre event happened every time I sprayed the radiator at the local car wash... The CEL light would go out with a click. The radiator only has the thermosensor and fan... I never did decipher it, my car is wrecked now. But to pursue this annoyance, that may be a route that is affecting the egr solenoid. I had 2 functioning "noids" and same code kept kicking in with an audible click. My fuel mileage was great, started great in cold, and idled down decently. even after dismantling the car and seeing every inch of wiring. the only thing close to loose wires was the electric fan on the radiator. Maybe a shared ground within ecu is finding a way to other daintier things like the solenoid valves. I did find some crazy clumped up shared grounds all over the place- one cluster was 5 wires all clamped together and wrapped in oem electrical tape by oem manufacture.

To experiment- I was going to put ground from that area in a different spot,from fan side or the the thermosensor, or both, but then I scrapped the noid valves and said to heck with them, along with egr valve. and left the cel light on forever.

[DasWaff]

Ground fault is definitely a high percentage bet. A decade ago, my old SPFI GL had CEL roulette, threw a wide range if incorrect codes for a couple of weeks, but especially TPS and solenoid. I had a ground wire broken inside the insulation (mine was the one at the back of the intake manifold) and after I finally found and fixed it... no more incorrect codes. DasWaff

[GeneralDisorder]

I fixed the idle this morning. For some reason the idle air valve (not the solenoid) has a weak spring or something. I had changed it to one I cleaned (left it in carb cleaner overnight), and the idle would not adjust lower than about 1000 give or take. I swapped back to the old (and still dirty) unit and now it works fine. Idle is about 700, and everything seems to check out ok. I'm going to call it good enough and focus on aquiring the knowledge and parts for Mega-Squirt.

 

To clarify, I only get the purge code in D-Check (Dealer Check) mode - that's when the test connectors are connected, and you run the test procedure (ignition on, full-throttle then back to half for 2 seconds, start engine and drive at greater than 5 MPH for at least one minutes, etc). It has never come on in normal U-Check without running the tests. After fixing the idle, I still get the code, but since it only comes on in D-Check I'm going to ignore it like a good consumer than doesn't see his CEL , enjoy the good running SPFI that I have now, and in the mean time plot it's downfall to Mega-Squirt.

 

The thing I really don't like about the SPFI is that without the purge solenoid and it's carbon canistor, the thing just won't idle right. It's basically expecting a small vacuum leak through the (correct diameter) port of the carbon canistor. But here's the real crappy part - it opens and closes the purge solenoid every 1.5 seconds. There is no easy way to approximate this action and "fool" the ECU, so you either put up with a rough idle, or you leave the system in place and have to look at it, and deal with it's hoses, and solenoid. I chose the latter for now. I don't wish to try and fool the ECU when my time is better spent working towards MS.

 

I think I may add a few more grounds to it just for kicks. I did the power/ground test at the ECU from the manual, and it all checks out, but that doesn't mean something isn't corroded or only partially broken. It's strange that I only get that one code, and not the code-salad that you would think I would get with a bad ground..... I'm beginning to hate this SPFI stuff almost as much as the Hitachi carbs.....*almost*. There will always be a special kind of hell for the Hitachi engineers it seems - their fuel injection sucks almost as bad as their carbs.

 

In truth the same problems with the old Hitachi carbs are still a problem with the SPFI. The added complexity of the "emmissions" equipment makes the fuel and ignition algorithms in the ECU software overly complex. It muddy's up the water such that it's hard to pinpoint the exact component that's failing or out of range. In the world of "reliable" fuel systems, less really is more for this 80's stuff. The ECU's in the newer cars have become complex enough (faster processors, etc) that they are able to do much more diagnostic work and provide more accurate feedback to the mechanic. That, and the advent of water-proof under-hood connectors has helped a lot to prevent often difficult-to-find corrostion related problems, and ground issues of the past.

 

GD

[TomRhere]

originally posted by GD

 

In truth the same problems with the old Hitachi carbs are still a problem with the SPFI. The added complexity of the "emmissions" equipment makes the fuel and ignition algorithms in the ECU software overly complex. It muddy's up the water such that it's hard to pinpoint the exact component that's failing or out of range. In the world of "reliable" fuel systems, less really is more for this 80's stuff. The ECU's in the newer cars have become complex enough (faster processors, etc) that they are able to do much more diagnostic work and provide more accurate feedback to the mechanic. That, and the advent of water-proof under-hood connectors has helped a lot to prevent often difficult-to-find corrostion related problems, and ground issues of the past.

 

 

AMEN Brother, AMEN!!!!!

[RONAN]

I can sympathize. I bought my 93 Loyale wagon 7yrs ago and dealt with the same issue. I noticed after I bought it that the woman had customized a piece of black electrical tape over the cel indicator on the dash. So I called her up about it and she profusely apologized about it saying that she forgot to mention it. (Yea Right) Anyway, she said that she had spent a couple of hundred to have it troubleshoot by a Subaru tech and they replaced the solenoid and it did`nt do any good. They threw the towel in and admitted that it is beyond their capability to fix and that it would be very costly to correct. (New one on me) They told her if she can stand the light to be on, that it would`nt affect the performance of the car. So she covered the light with the tape.

 

So I downloaded the code and sure enough, it was the carbon canister purge solenoid. So I got another from the u-pull it and changed it out again, to no avail. So, I`ve been driving it with the light on ever since. Every once in a while, it will go out for a few miles and then come back on again, but were talking about every 4 or so months.

[GeneralDisorder]

So now I'm wondering....

 

Since the ECU obviously can tell the difference when the solenoid is open and when it's closed (MAF signal? Something must change anyway), I wonder if a clogged carbon canistor can cause an otherwise working solenoid to appear to stay in the closed position, and so the ECU see's no change when it opens?

 

GD

[keltik]

Could there could be a small change in the O2 sensor when the valve opens due to the fuel vapors in the intake charge, i didnt think it would be enough for the ECU to reliably measure. Anyways its worth a shot - nobodies got any other ideas yet.

[DaveT]

Since the ECU obviously can tell the difference when the solenoid is open and when it's closed (MAF signal? Something must change anyway),

GD

 

Actually, all the ECU can check is that the solenoid draws current when it is supposed to be energized.

 

Maybe run a seperate ground wire from the solenoid if the existing one is suspect?

[GeneralDisorder]

Actually, all the ECU can check is that the solenoid draws current when it is supposed to be energized.

 

Maybe run a seperate ground wire from the solenoid if the existing one is suspect?

 

Well - I had thought the same - but if that's the case, then why does the code only show up in D-Check? If that were the case then the ECU would pick it up in U-Check and light my CEL.

 

The solenoid's ground is provided by the ECU on pin 5. That's how it energizes it. The solenoid get's ignition switched 12v, and the ECU uses the ground to open it. I wired it to unswitched 12v with a 20 amp fuse, and directly to pin 5 on the ECU - total wire length was about 12 inches. When the ECU is off, no ground exists so it's safe to use unswitched voltage to the resistor. The resistor does get warm when you turn the ignition on, and it takes a while for the code to come up - about a mile I would say.

 

Nothing I have tried has stopped the code, so it's either a ground to the ECU or it's senseing the purge isn't openeing through other means. I am still skeptical about the "other means" possibility, but unlike MS, I can't just rife through the code and check

 

02 Sensor is new (BRAND new in fact - as in three days ago). Seems to function just fine.

 

GD

[GeneralDisorder]

I just had another thought.... what if the added ground of the test connector and the added CPU workload of the diagnostic functions were enough to overpower my ground connections causeing there to not be enough current flow to the purge solenoid?!

 

Wow - that would be the strangest combination ever. Or have I jumped into the deep end trying to save BDG?

 

GD

[keltik]

My vote: Send the Hitachi gear to hell and go for MegaSquirt.

If the same dudes who designed the carbs programmed the ECU - it could take years for you to solve this one.

[NorthWet]

I think that you guys are too hard on Hitachi, one of the foremost engineering and production companies in the world; I would take them over Bosch any day. AND...

 

...I also think that you are placing too much complexity in the ECU. It was designed during, and using components from, a relatively primative era in electronics, processors and support parts. I doubt that there is any provision in these early ECUs for anything more than simple on/off monitoring (using a simple voltage comparator); almost certainly not an ADC circuit needed to properly monitor whether it had activated normally.

 

I am mildly surprised that the 10W resistor is warming up. Power dissipation should be around 3 watts. My surprise isn't so much that 3 watts would raise the temp noticably, but that the little plastic package of the original would be subjected to that much warming. Is it possible that we are misjudging what is needed for the solenoid's fake-out? Is it possible that the ECU is basing the solenoid's functioning off of an inductive EMF "buck"? Admittedly, this would not explain why the code is happening with assumed-working solenoids.

 

Sorry for rambling... sleep deprived. :-\

[youareabus]

I'm certainly no expert at this problem like most of you are. That being said I did just fix mine by using 3 100 ohm resistors tied together in parallel to get 33.33... ohm. The resisters I used at first ohmed out right but did not do the trick. The ones I tied up are .25 watt resistors from Radio Shack making a total of .75 watt capability. That worked, no CEL now. RS sells the resistors in a 5 pack (first link) and in a set (second link). I had the set and got the 6 I needed for 2 solonoids but if you are only having a problem with one you can buy the 5 pack. Heres the links:

 

http://www.radioshack.com/product/index.jsp?productId=2062339&cp=2032058.2032230.2032267&allCount=100&fbn=Type%2F1%26%23047%3B4+Watt+Resistors&f=PAD%2FProduct+Type%2F1%26%23047%3B4+Watt+Resistors&fbc=1&parentPage=family

 

http://www.radioshack.com/product/index.jsp?productId=2062304&cp=2032058.2032230.2032267&parentPage=family

 

 

 

I also replaced the alternator at the same time I put in the new resistors. I am going to check mine by pulling out the resistors and hooking the solonoids back up to verify if the CEL returns. I have an idea that with the alt going bad it's possible that I was getting voltage spikes or AC ripple at the ECU and maybe that was the "real" problem. I will post with the results.

[pisces_0]

I'm certainly no expert at this problem like most of you are. That being said I did just fix mine by using 3 100 ohm resistors tied together in parallel to get 33.33... ohm. The resisters I used at first ohmed out right but did not do the trick. The ones I tied up are .25 watt resistors from Radio Shack making a total of .75 watt capability. That worked, no CEL now. RS sells the resistors in a 5 pack (first link) and in a set (second link). I had the set and got the 6 I needed for 2 solonoids but if you are only having a problem with one you can buy the 5 pack. Heres the links:

 

http://www.radioshack.com/product/index.jsp?productId=2062339&cp=2032058.2032230.2032267&allCount=100&fbn=Type%2F1%26%23047%3B4+Watt+Resistors&f=PAD%2FProduct+Type%2F1%26%23047%3B4+Watt+Resistors&fbc=1&parentPage=family

 

http://www.radioshack.com/product/index.jsp?productId=2062304&cp=2032058.2032230.2032267&parentPage=family

Just wanted to bring this one back up as I've fought an intermittent EGR solenoid check-engine light for quite some time now. I've got the EGR valve blocked off and all the associated vacuum lines to it and the solenoid disconnected. The solenoid is still present and plugged in, but sure as shootin' after ~1-2 miles of driving the check engine light pops up with code 34.

 

In my continued efforts to simplify the engine bay I'd rather just remove the solenoid, solenoid connector and replace the whole works with some resistors. youareabus, how have these resistors worked out for you so far? I'd think that the wattage rating on them would be a bit low as it seems there's a good bit of power flowing through the solenoids. :-\ If it works I'm all for it though.

[azsubaru]

I drove mine with a 68 ohm 10 watt resistor for almost two years; the CEL never came on once during that time (at least with an EGR code).

 

I recently put a solenoid back in because emissions testing is due again. 5 Watts should be ok, but less than that, you are going to overload the resistor's rating, and it may burn out.

 

What I did was cut the plug off the bad solenoid and soldered the resistor across the wires, then plugged it back in. So putting a solenoid back in was a piece of cake.

[DaveT]

It's basically expecting a small vacuum leak through the (correct diameter) port of the carbon canistor. But here's the real crappy part - it opens and closes the purge solenoid every 1.5 seconds.

 

Hmm, Every once in a while I've noticed the idle bouncing up & down at a rate close to that. Very rarely. Next time I'll have to time it. Think about what else could combine to cause the weird idle.

 

Yes, the ECU computer is simple. The check for the solenoid is some kind of check for current when it's energized. If you disconnect it, it will trigger the CEL, but only when the ECU tries to turn it on. (Actually, I saw this while troubleshooting the EGR solenoid, (I put a monitor LED on it) but it should be the same.)

 

I know from experience, that the O2 sensor can go bad in a way that the ECU does not see as an error, and the front catalytic converter ends up bad.

 

This is a guess - maybe the dealer code is "figured" in software by some combo of signals.

 

I have thought about reverse engineering an ECU.

 

In a few years, I could do a megasquirt, since the cars will be old enough to skip emmissions.