Low power in low RPMs after reconnecting O2 sensor
 

Hey, folks -

I have yet another opportunity to sort out a CIS problem.

Back in late October, I installed a wideband O2 sensor in my '83. The new PLX has a narrowband output that connects to the ECU to provide the lambda signal. The O2 sensor in this car had been disconnected for many years -- maybe a decade.

I did a couple of test drives after installing it and everything seemed to check out ok. I didn't drive the car at all in November.

Last week the sun came out and I figured it was a good day for a drive so I drove to the store about 3 miles from the house (1-way). It was cold -- I'd say about 40F. On the way to the store the car started to wheeze and pop a bit in low RPMs and the car lost power in the low RPM range. Pushing on the accelerator didn't do much and then it was like it would suddenly "catch" and the RPMs would shoot up to where they should be. It did it a couple of times on the way and then when I was in the parking lot at the store it seemed like the RPMs kept wanting to drop. Did my shopping and about 30 min later when I came out the car started up fine and I drove home with no popping/wheezing but it felt like lower power than normal in low RPMs.

This weekend I drove around the neighborhood and the coughing/wheezing was back so I thought this is a not a fluke. My first thought was to replace the O2 relay so I went to the local NAPA store, got a new Bosch replacement (Part number is 0332019150) and installed it today. No change in symptoms.

It seems like I'm looking at a couple of possibilities. It is either (a) the new wideband O2 sensor has created a problem where there wasn't one before or (b) by reconnecting the O2 sensor I have found a new problem. Maybe this is why the O2 sensor was disconnected to begin with? I don't think the ECU uses the O2 sensor on a cold engine though so I'm inclined to believe that the new O2 sensor is not really the problem.

I'm looking for ideas on what to look at next? I could disconnect the narrowband output and see if makes a difference. If it does I will be very confused.

From a quick scan of your link, the wide band thing does nothing with respect to the Lambda box except provide the analog signal. Disconnect the entire unit and see if the symptoms repeat.

All you are doing in effect is unplugging the single wire O2 sensor with respect to engine operation.

If things smooth out, then you know that the problem is isolated to the new box.

You do mention you had to crimp some wires and use little connector pins. Could they be cross talking among themselves?

O2 sensor "closes the loop" once it gets warm enough to send a signal. Should not take long given it's exhaust heat doing the warming.

Did you "mess" with the 12 pin connector on the engine firewall? The frequency valve signal travels through there. Loose? Pin push back?

Thanks, Bob. Comments in-line below.

Planning on doing that today.

I got to thinking about this last night. I was thinking about the differences between the old and new O2 sensors and a couple of things occurred to me.

In my case the original O2 sensor was disconnected. With NO signal coming from the O2 sensor, I believe the proper behavior is that the ECU reads this as an open circuit and goes into its "open loop" operation mode. No surprises there.

Here's my question... Does the wideband O2 create a new problem? Here's what I'm thinking...

In normal operation with the factory O2 sensor, the voltage signal is actually created by the excited exhaust gas. In order to do this, the metal electrodes in the sensor only create a voltage "when the ambient air has a higher oxygen content than the exhaust and the ceramic material is hotter than 575F." (Bosch Fuel Injection book). Obviously it will only send the signal once it's sufficiently warmed up.

The new wideband O2 sensor adds a control module. So I'm wondering if it starts sending a signal immediately at startup or if it takes some time. Is it possible that the wideband unit is sending a signal too early and causing the ECU to move out of open-loop and into closed-loop too soon?

I don't know enough about the ECU logic to know how it would behave.

I was pretty careful with the crimps and they fit into a plastic molex connector so I don't think they are cross-talking. I used a posi-lock connector for the O2 sensor wire to the ECU. I suppose that could be introducing some noise. I will check the voltage on the green/white wire on the test port. I could also disconnect the connection from the wideband module and compare the readings to see if there is a difference.

If I messed with the 12-pin connecter it was by accident. Worth double-checking for sure.

I am wondering if I might have a bad thermoswitch.

If you did not touch the twelve pin, I would let it be. Main thing is continuity between the freq valve and lambda box. A guy in Dallas (older thread) had some corrosion at that pin and in really hot weather the circuit would break and would act like you pulled the O2 relay. Cooled down and all was well. Drove him nuts.

Thermotime switch would spray gas through the CSV if (switch) closed. Normally only when starter is engaged, right? Then should receive no more voltage.

Let us know what you find.

What's the "white stuff" at the base of the yellow fuse where you are pulling power? It looks different than the other two.

You may be pulling power from the hot side, though.

http://forums.pelicanparts.com/uploa...1387387819.jpg

Started the car with the dwell meter hooked up. Actron dwell meter. Green wire from meter to green/white wire on test port and black to ground. Cold engine. Ambient temp in the garage was 54F. Started running very lean. 20% duty cycle. Idles rough.

Stopped the car. Disconnected the narrowband output from the AFR system to the ECU. Grounded the green shield wire that goes to the ECU. Started the car again. No change. Disconnected the ground with the car running. No change.

I am pulling power from the hot side. I just looked at the yellow fuse and I don't see it. It might've been glare from the flash on my smartphone.

Just for shiz'n'giggles I swapped the new O2 relay with the old one. No change. What are the odds the new relay is bad?

Does it stink?

I always struggle with this but I think that with a 20% duty cycle the signal from the O2 sensor is saying it is too rich. The frequency valve is being instructed to lean the mixture out to compensate.

I think I remember that psalt says to set it around 35 duty cycle which is a little rich. The Lambda system will compensate and bring it to stioch in closed loop (dithering) but when you got to WOT you have a bit of extra fuel as the Lambda system is disabled.

You can confirm the freq valve is working with the car off. You need to jump a wire among three terminals on the the O2 relay and plug it in.. Let me see if I can find the related old thread.

See post #32

http://forums.pelicanparts.com/porsche-911-technical-forum/559200-911-sc-running-flat-no-idle-2.html

Also, is the O2 fuse ok in the front trunk?

Stinks to high heaven.

According to the Bosch book, a zero voltage output from the O2 sensor signals a lean mixture, which drives the the ECU to compensate by enriching the mixture. The duty cycle SHOULD go to 95%.

To test for the opposite, the book says to use a 2v battery. Connect the battery + to the green shielded lambda wire and - to ground. This simulates positive voltage (rich condition) and the ECU should send a duty cycle of 10%.

My problem seems to be that I'm now sending NO lambda signal to the ECU and I have the OPPOSITE expected condition (low duty cycle reading). I'm scratching my head now.

Either I have a two bad O2 relays, a bad ECU or something entirely different that I haven't figured out yet.

I'll give that a go.

If you mean the dome light circuit, yes it's working.

Confirmed frequency valve buzzes when pins 87, 87b and 30 are jumpered on the O2 relay.

Good - way better than pulling the 12 pin connector.

Low voltage from the O2 sensor is sniffing lean. I think the regular sensor goes up to a max of 0.9 volts when rich.

I am going to find the psalt thread regarding rich vs. lean. I read Probst's book and still had it backwards in my mind a few years back.

Here it is: http://forums.pelicanparts.com/porsche-911-technical-forum/774966-fuel-pressure-fv-anomalies.html

Here is a vid I made of pulling the O2 relay. There are little backfires at the end of the vid. Do you have any of these?

https://www.youtube.com/watch?v=5tdpCYyvqSc

Bob,

That was how it was running the other day. I thought I would try and make my own video and when I started it again the damn thing doesn't register anything on the dwell meter. I've checked and double checked. Something is not right here.

Dammit.

A new symptom. Let the car cool down for a few minutes, double checked all my connections AGAIN. Dwell meter plugged into test port. DVM plugged into narrowband output of the AFR module to see what it's reading. New O2 relay installed. Green shielded O2 sensor wire disconnected.

Started the car and it idled smooth. Same lean reading on the dwell meter. AFR gauge is putting out .645V and slightly fluctuating.

As I'm looking over everything I hear the RPMs change a bit. Look down at the dwell meter and the needle is back at zero.

I'm thinking that the ECU was switching from cold start to partially warm mode.

I also think the dwell SHOULD'VE gone to 50% duty cycle. This is not right. Do I have a bad ECU????

I r stumped right now.

If your car sounds like my video, slow throttle response, etc, the connection to the freq valve is being disrupted.

Check the 12 pin to the right of your left rear shock. Number one connection. Black and red, both sides. May be corroded.

However, you never had this problem before so double check your wiring.

Check the connection on the freq valve too.

Your engine is rich. Not lean. I got a 12 pack that says so. :-)

However, not the issue right now.

Another thought. Hook up that relay jumper you tried earlier and wiggle stuff.

I can offer up a few things about your problem. I have just started on a quest for data to post on the Primer regarding Lambda. Currently I have practically no thorough info on the site about troubleshooting Lambda, but have just finished spending a good bit of time and effort gathering info and am working on posting what I have learned.

One thing I found that I can share with you now. I had bought an Actron meter to use in the dwell mode to measure duty cycle. I gave up on it after getting readings I knew to be wrong, and I mean way wrong. I verified this by attaching three things into a test setup I made for the workbench in my garage: a real duty cycle meter, an oscilloscope, and an inline DC ammeter. They confirmed that the Actron dwell meter could not be relied on to provide accurate information. An example is reading 20% duty cycle (yes, after doing the conversion) when everything else said 93%.

This was my experience, others may have a different one. The d.c. meter I used, which I found to be dead on in every case, was one I've had in my tool box for years, but always used for other measurements, like voltage, resistance, rpm, and temperature. A SunPro CP 7680. Your AFR meter reading indicates (if it represents what voltage an O2 sensor might be putting out) a slightly rich mixture.

I just hate to see someone else struggle with something I know was a problem for me. YMMV.

Cripes.

That is only a statistical mass of one. However, that's all I have as a reference point.

So I look up your SunPro CD 7680 unit. Actron owns it now.........

I won't sleep tonight.

Jim,

Wow that is interesting about the Actron unit. Thanks for sharing that. Now I'm worried about trusting that thing to set the AFR correctly. Bob and I talked by phone (thanks again Bob!) tonight and we made some progress. Here is what happened:

- Bob suggested I disconnect the 12-pin connector and clean the pins. It didn't look bad but I hit it with a wire brush anyway.

- I left the narrowband O2 output to the ECU (green shielded wire) disconnected but I hooked up the DVM.

- New O2 sensor relay was plugged in.

- The Actron dwell meter was connected to the green/white wire on the test port (and to ground).

I started the car and the dwell read 16 on a 4-cyl scale. The DVM read .63 V so it's running rich. I was misreading the Bosch book. With a rich signal, the ECU sends less of a duty cycle. The less time the valve is open, the less fuel injected. To prove this, I grounded the green shielded wire and the dwell dropped to zero. I also heard a slight RPM change so that leads me to believe the ECU is working as it should.

My next step was going to be to adjust the AFR a bit using the 3mm hex screw after I got off the phone with Bob. I turned my garage upside down looking for my long allen wrench set and couldn't find it. I finally decided to have a beer and watch Christmas Vacation with the kids. I'll run over to HF and pick up another long 3mm allen wrench tomorrow.

The plan is to back the AFR down until the DVM reads about .45 V and see what happens to the dwell reading. It will also be interesting to see how it runs after another cold start tomorrow. I'm also going to fish around and make sure nothing is loose or disconnected. If everything checks out I'm going to re-connect the narrowband output from the AFR unit to the ECU and see what happens.

I'm still a little troubled that the symptoms and the evidence don't jive. The thing that got me going down this path was more of a lean running/bad O2 relay problem.

The only other thing I can think of is that maybe I've got some crud in the fuel tank. I ran a can of BG44k through the tank a few months ago.

I apologize in advance if I may have created any panic on the Acton dwell meter. My experience was that of one person. The Actron meter I referred to was a model CP7605, a relative cheapy, Sears price about $40. It was described as a meter for measuring dwell on point-type ignition systems, and so I had assumed it would fill the bill for checking Lambda ECUs in accord with Porsche directions.

When I tried to double check it against the actual duty cycle setting of the SunPro meter, as soon as I connected the Actron CP 7605, I watched the scope waveform fall from ~12 volts to roughly half that. I figured that the 7605 was loading the signal from the ECU, so I discontinued the use of it. I just today checked the internal DC resistance of the Acton and found it to be ~ 1MΩ. This seems to be a high enough resistance not to have caused a load indication like I was seeing. So all I can do is pass on is what I saw.

The SunPro meter readings jibed with the scope readings and also the relative indications from the DC amperage draw as the duty cycle changed from rich stop to 50% to lean stop.

tirwin:

From your post where you say -
"I started the car and the dwell read 16 on a 4-cyl scale. The DVM read .63 V so it's running rich. I was misreading the Bosch book. With a rich signal, the ECU sends less of a duty cycle. The less time the valve is open, the less fuel injected. To prove this, I grounded the green shielded wire and the dwell dropped to zero. I also heard a slight RPM change so that leads me to believe the ECU is working as it should."

Zero volts (O2 sensor grounded) tells the ECU that the O2 sensor is connected, and the mixture is reaaly lean; go to the rich stop at 93% (or so), so the dwell reading (indicated duty cycle) should go high (to max).

Remember, the dwell meter reads ignition points dwell. Dwell is the % of time the points are closed. However, this is also the % of time the coil current is flowing. This is equivalent to the ECU set up, and why a dwell meter should theoretically read the duty cycle of the Frequency Valve. The set up is electrically equivalent.

Ignition coil = Frequency Valve (both connected to +12 volts)
Ign. points = ECU (both connecting each of the above to ground)

However the electrical impedance of these two systems is very much different. When ignition points are closed, resistance is LOW. When the ECU "closes the solid state points" that pulse the Frequency Valve, its characteristics are not the same. I don't think all unsophisticated dwell meters, meant for checking ignition points, are up to the task of checking ECU duty cycle.

If anyone wants to chime in here or maybe has a different take on this, please speak up.

Only thing I will add is based on what I have read and that is the dwell meter is only a substitute, a proxy, for a duty cycle meter. So, in my mind, the dwell meter does not work the same way as the correct tool. After that statement, I am maxed out. :-)

Tim,

The metal fuel lines are a PITA regarding the wrench. I finally made a frankenstein 3mm tool that I can fit down in the hole and then with a larger allen wrench make the tweaks.

http://forums.pelicanparts.com/uploa...1387470757.jpg

Jim,

I'm with you on the theory part of how the lambda system functions. I guess what I'm trying to figure out is how that information is displayed on the dwell meter. Maybe it is a perspective problem.

My interpretation of duty cycle is the % of time the FV is OPEN. More time open = more fuel delivery. More time closed = less fuel delivery. I would EXPECT that duty cycle would move toward 100% on the gauge with the O2 sensor wire grounded. If I connected a 2V battery to the sensor wire, it should move toward 0%.

I clearly saw the EXACT OPPOSITE of that yesterday and reproduced it. Grounding the sensor wire made the duty cycle go to 0%.

Given that we now confirm that the mixture is rich, I am wondering if the Actron unit is actually displaying the duty cycle backwards (at lest to our way of thinking about it). If you are correct that the meter is reading the amount of time CLOSED, then the correct way to read the duty cycle on the Actron is actually 1 - (decimal value of % reading). So what I was interpreting as a 20% duty cycle is REALLY an 80% duty cycle ( 1 - .20 = .80). That makes a lot more sense given what I am seeing with my own eyes. The other possibility is the Actron unit is a piece of junk. :D

I have a timing light with a digital dwell meter. I am going to hook it up later today and do all the tests again and compare that.

Wow. If the Actron unit is actually BACKWARDS of what we expect, then that is a really important tidbit of information!!! :eek:

Good tip. I just got back from HF and then I saw this. I'll try this new one and see if I have any luck.

A couple things I wonder about...

The voltage produced by the O2 sensor is pretty low. Any chance that all of the additional wiring and the additional control module changes the value seen by the control unit in the car? For instance, if you have .5V out of the sensor, what do you have when the signal finally gets to the car's control unit? Same? Less?

Have you verified the proper operation of the throttle microswitch? The 35C temperature switch?

Given the temperature in your garage, I'd suggest you have a copy of the '81 Service info booklet handy, so you know what the duty cycle should be. It's complicated.

If it were my car, I'd swap in a stock O2 sensor and get a better meter to measure the duty cycle.

Good luck,
JR

Looking forward to finding if there is impact from tweaking CO.

javadog,

The voltage produced by a narrowband O2 sensor is only supposed to be 0-1V. So that's completely in the normal range. I haven't checked it at the ECU but that's probably a good idea to add to the list.

The O2 sensor was disconnected for a long time and it's currently disconnected now. I'm not ready to throw in the towel on the wideband sensor yet. :)

I have been talking of 45 degrees as 50% duty. That is on a 4 cyl setting. You are using an 8cyl scale so 50% duty is 22.5 degrees dwell. You are not far away. Slightly rich.

Yep. My concern was that it doesn't take much change in the resistance of the circuit to alter a small voltage, right? I'd be real surprised if all the crap downstream of the wide-band O2 sensor doesn't screw up the voltage. That's why I recommended putting a stock (meaning original connector) O2 sensor back in while you sort out your other issues.


I thought the duty cycle defaulted to 65% on WOT.

Warm mode? Not sure that's an accurate way to think of it. The ECU wants to know whether the engine temp is less than 15C, between 15C and 35C, or higher than 35C. It also cares whether the O2 sensor is cold or warm. The last input is whether the throttle angle is above or below 35 degrees. Enrichment is added when the throttle opens past 1 and/or 15 degrees, if the engine temp is less than 35C. It's a somewhat convoluted table of inputs and outputs.

The only time the duty cycle should be at 50% is when the engine is above 15C and the O2 sensor is still cold.

JR

Bob said:
I think I remember that psalt says to set it around 35 duty cycle which is a little rich. The Lambda system will compensate and bring it to stioch in closed loop (dithering) but when you got to WOT you have a bit of extra fuel as the Lambda system is disabled.

javadog asks:
I thought the duty cycle defaulted to 65% on WOT.

Response:

The 35 I mentioned is not the duty cycle, I am sorry. It is the dwell degrees on a 4 cylinder setting of a dwell meter - 90 degree scale.

This dwell meter reading indicates the car is fighting a slightly rich mixture at idle (which is the intent making the adjustment psalt mentions). The Lambda box is telling the freq valve to reduce fuel. The Lambda system corrects for the rich mixture.

If you floor it, Lambda is disabled. and since the 3mm fuel mixture screw is adjusted a bit rich, there is a little extra fuel introduced under WOT at the 65% duty cycle.

I am backing up a few posts. You guys are getting ahead of me a bit.

Some basics, as sometimes all this is confusing to think about.

The higher the duty cycle, the more time the Frequency Valve is open. The more time it is open, the lower the pressure in the lower chambers of the Fuel Distributor. The lower the pressure, the richer the mixture.

Where you say:"I clearly saw the EXACT OPPOSITE of that yesterday and reproduced it. Grounding the sensor wire made the duty cycle go to 0%."

But you are believing the dwell meter. I propose that if you are having the same issues with the Actron meter that I had, you can't believe it. If you can't believe it, you don't know what the duty cycle actually is.

A dwell meter, by definition, is set up to read ignition points closure, yet it is also reading the time current is flowing in the coil. This would equate to the time current is flowing in the Frequency Valve, as the ECU is duplicating the action of the points. A different type of dwell meter might read ECU duty cycle correctly.

The Actron meter may do a decent job with a points/coil ignition. I am suggesting that it does not do a good job with the ECU. I don't believe (although I have no way of knowing this) that the dwell meter the Porsche workshop manual used was an inexpensive two lead meter, but rather a three lead meter, considering the three pin connector the factory installed to connect up a dwell meter. It used +12 volts, ground, and the test lead to pin 17 on the ECU.

I think the best way (and maybe the only way) to resolve the uncertainty you are seeing is to beg, buy or borrow a duty cycle meter, or better yet find an oscilloscope to monitor the test connection. If you can look at a waveform trace, you will know what the duty cycle is.

Jim,

I just got home from a test drive. Progress!

I started out with the Actron dwell meter hooked up. I was getting a reading of 6 on an 8-cyl scale, so 12 on a 4-cyl scale.

I got out my Innova timing light with a digital dwell meter. I got a totally different reading. 45 on a 4-cyl scale. The O2 sensor is still disconnected so it is running on a fixed 50% duty cycle. That looks more accurate.

I then backed off the mixture a hair and went for a drive. Everything was good. The exhaust doesn't smell as bad.

So tomorrow I'll hook up the O2 sensor again and see what happens.

But I think this confirms your theory that the Actron unit isn't good for this purpose.

JR -

Do you have a reference as to what the '81 Service info booklet is that gives the duty cycles for different conditions? I have my own data that I've measured recently on the workbench, but I'd sure like to see what the factory has to say, as a way to double check the numbers I got.

This is from 1981, when Porsche introduced the O2 sensor control unit. I don't think they made any changes to it until 1983 when a new unit was introduced. The main difference in 1983 was that the enrichment was kept in place until 25 seconds after the O2 sensor was warm. The enrichment was for 2.5 seconds, which may be a little longer than the enrichment in 1981.

http://forums.pelicanparts.com/uploa...1387554697.jpg

http://forums.pelicanparts.com/uploa...1387554716.jpg

http://forums.pelicanparts.com/uploa...1387554731.jpg

Quick update.

Started the car this morning after last night's adjustments. Strong start and very smooth idle. RPMs initially went to 1600. Noticeably less exhaust smell. That is an improvement to me. I used to see more hunting. The hunting seems to only happen in cooler weather -- say below the 60's.

Let it idle for a few minutes. RPMs gradually started to drop as expected. When it dropped below 1000 and settled into 900 it started chugging a bit. Best way I can describe it is less like a Porsche and more like a muscle car sound.

About a minute later the RPMs dropped down to the 650 range and the chugging got a lot worse so I turned it off. I didn't have any more time to look into it.

Last night I did have to slightly increase the idle as I leaned the mixture a bit.

I feel like I've almost got it dialed in now. What should I do as the next step? I need to do another cold start and give it more time to get good and warmed up.

Another update. I had a few minutes at lunch so I thought I'd test the effect of hooking up the O2 sensor. I started the car and let it warm up. Then I adjusted the idle up to 900 RPM.

I rigged up a setup where I could monitor the narrowband output of the sensor with my DVM while it was connected to the ECU.

First I started the car with the O2 sensor disconnected. Sensor output was .6V.

Next I used the same Posi-lock connector to see if I could measure a difference in output just to rule out that possibility. (Still not connected to the green shielded wire.) I measured a difference of .01V.

Then I connected the sensor to the green shielded wire. The car started stumbling badly. RPMs dropped and started surging as the ECU tried to compensate. It almost died a couple of times. The sensor output fluctuated in response to the changing conditions. As soon as I disconnected it, it returned to normal.

I think this confirms that the ECU does not like the O2 sensor being connected.

We know what signal is being sent by the O2 sensor. What I do NOT yet know is what is being RECEIVED at the ECU as javadog pointed out.

Why is the ECU reacting so badly? I wonder if this is the real reason the O2 sensor was disconnected for so many years.

Maybe the moral of this story should be to ditch the O2 and put in a set of SSIs with M&K 2-in/2-out!!!