O2 sensor test and inverse duty cycle - PeachParts Mercedes-Benz Forum

LeaUK · #1 04-08-2005, 02:28 AM

Hi Guys

Thanks for the comments.

Just to be clear, my engine is a LH not KE and therefore has much more control all round.

When previously posted, my referral to the 'diag signal' is the signal that is present in the X11. The R129 (from around '92 upto around '95) was fitted with both x11/4 (38pin diag conn) and the traditional x11 (9 pin) connector for fault diagnostics.

I'm specifically interested by the o2 emissions as these are good clues to whether the Cat is hot enough - thanks for letting me know.

Lea

LeaUK · #2 04-10-2005, 08:09 AM

I got chance to make a few measurements this weekend and my observations were as follows:

1. My Cat just needed a real good heat up. 20 minutes in second gear up and down our local 60MPH motorway resolved that and produced these results:

At idle:

CO 0.062% (0.3% limit)
CO2 15.13%
HC 14ppm
O2 0.06%
Lambda 1.000
NO 8ppm

2500-300RPM (as required by UK Law)

CO 0.141% (0.3% limit)
CO2 15.10%
HC 23ppm
O2 0.03%
Lambda 0.996
NO 11ppm

For comparison, what's the 'normal' level of CO would one expect from a 12 year old LH V8 engine with Cat?

2. The 9 pin X11 diag connector (pin 3 and 6 for my car) outputs a nice signal (which as Duke2.6 previously mentioned) representing the amount of adjustment the engine is having to make to ensure that the Lambda sensor reads the 14.7:1 ie a Lambda of 1.

So this signal is ideal to determine if the car has any minor air leaks. Mine was perfect, at a fluctuating duty of 48-52%.

To prove that unmetered air would be compensated for I pulled one of the vac hoses of the intake manifold - and watched the duty change down to 10%, or of course up to 90% depending how you've connected your DVM/Scope and also if your reading negative or positive going duty cycles.

For me it was irrelevant, as to see the change right in front of my eyes was great.

BUT - from pulling a hose (only a 2mm diameter hose) it took nearly 10 seconds to complete adjustment and settle, or at least display the adjustment. So why so slow? Closed loop systems are usually capable of reacting much quicker.

Is this dependent on O2 sensor reaction time? Thinking about it I was at idle, but still 10 seconds!?

Any thoughts welcomed.

Lea

JimF · #3 04-10-2005, 12:02 PM

Quote:

Originally Posted by LeaUK

To prove that unmetered air would be compensated for I pulled one of the vac hoses of the intake manifold - and watched the duty change down to 10%, or of course up to 90% depending how you've connected your DVM/Scope and also if your reading negative or positive going duty cycles.

BUT - from pulling a hose (only a 2mm diameter hose) it took nearly 10 seconds to complete adjustment and settle, or at least display the adjustment. So why so slow? Closed loop systems are usually capable of reacting much quicker.

Is this dependent on O2 sensor reaction time? Thinking about it I was at idle, but still 10 seconds!?

Lea

'blue-baby' passed with flying colors!

Just came across this thread re comments about X11 pins. As far as I know, using my car as a reference, pin 2 is 'ground' (chassis) and pin 3 is the output signal. Pin 6 is the battery voltage, so reading of DC should be made with the stated convention.

There's a picture in MENU#4 on my page showing 'directions'; lean or rich and if you click on the "M0039" link, you see a pin out of X11 connector. In '95 the O2 signals were moved to the 38 pin connector (pin 14 O2, pin 2 ground, pin 3 battery).

Bosch rates their O2 sensors for about 60k miles, so if it's not been replaced you will be surprised at the better performance. I'd imagine that the response time could be used as an indicator of how well the O2 sensor is performing. 10 secs tells me it's may well need to be replaced.

I believe that it (O2 sensor) should 'switch' in a second or so from rich to lean given the right stimulus. In servo design, the 'electronics' behind the sensor is always made to be slower than the sensor otherwise the system would "HUNT" (oscillate). I would think that it should correct in 2-3 seconds at most.

LeaUK · #4 04-10-2005, 12:50 PM

Hi Jim

I did take a quick review of your page before testing and again some good info - ta.

Unfortunately I used the pins rather blindly as these were the two my documentation points too, I guessed one would be 0 or 12V but as I don't understand if I'm measuring negative or positive duty cycle, I decided it didn't really matter at first. I've read MB invert their DC reading and maybe this is why they connect between the Lambda signal and 12V as opposed to 0V? Of course, had I found that she was biased above or below 50% I would need to find out more .

I also note that in documentation (and on your site) the 'purge' valve should be disconnected before testing - do you know of it's function? and what effect on the test this has. I never removed mine, if my car has one of course!

With regards to the O2 sensor I've placed the order already but unfortunately my local dealer has no stock, so next week should see further testing and comparison

10 seconds seems slow but I certainly agree it's response has to be much slower than a sensors output with the characteristics of a typical O2 sensor. The control system has further interesting factors, such as, although the engine prefers to run at 14.7:1 infact cats actually need O2 to operate correctly so I guess that's one reason why we see high swings on the O2 sensor o/p - and of course this sensor also has a very high transfer ratio around the 14.7:1 mark too.

Can't wait to get a little more time to carry out further testing now...

Cheers

LeaUK · #5 04-10-2005, 01:12 PM

Hi Duke

Thanks also for your comments.

Maybe a change of 40% was a little harsh but hey I was simply looking to generate a reaction and yep - I got one. Shame we have no data on the response time as it may yield further clues.

I need to recheck the HC as 15ppm just seems rather low in my opinion but I have little practical experience and have to rely on those that do and often what the measuring instrument tells me. MB specs are 300ppm max so if these weren't erroneous readings then she's burning fuel fine

UK CO limits are <=2.5% for cars used before '93 (excluding classic and other strange classes) after it's 0.3% and I've heard that's it's soon (if it hasn't already) to be <=0.2%.

Lea

JimF · #6 04-11-2005, 12:28 AM

Quote:

Originally Posted by LeaUK

I need to recheck the HC as 15ppm just seems rather low in my opinion but I have little practical experience and have to rely on those that do and often what the measuring instrument tells me. MB specs are 300ppm max so if these weren't erroneous readings then she's burning fuel fine

Lea

test, my baby's HC was . . . are you sitting down . . 2ppm at idle and 4ppm at 2500. So 15ppm is not too low. Means the cat is doing it's thing! And you car's engine ain't too bad either. Enjoy it!

LeaUK · #7 04-11-2005, 02:36 AM

See this is where you just cant beat practical experience. Thanks for the HC confirmation, so cars can have extremely low HC emissions. Mind you I thought my 15ppm was good...

Out of interest, what's your CO Jim?

Just checked our UK limits again and it seems that it's 200ppm not 300ppm, but hey, no problems here.

I'm still a little concerned with the relatively high CO readings even after a good cat conditioning - although I do have to admit to idling for around 5 to 8 minutes before the test was performed as I had to wait for a second machine calibration. Not ideal I know, but I had no choice. Next time I'll ensure the measuring instrument is ready to go as I arrive

Duke2.6/Jim - what's your feeling on my CO out?

Thanks
Lea

Duke2.6 · #8 04-10-2005, 01:13 PM

Quote:

Originally Posted by LeaUK

Hi Jim

I did take a quick review of your page before testing and again some good info - ta.

Unfortunately I used the pins rather blindly as these were the two my documentation points too, I guessed one would be 0 or 12V but as I don't understand if I'm measuring negative or positive duty cycle, I decided it didn't really matter at first. I've read MB invert their DC reading and maybe this is why they connect between the Lambda signal and 12V as opposed to 0V? Of course, had I found that she was biased above or below 50% I would need to find out more .

I also note that in documentation (and on your site) the 'purge' valve should be disconnected before testing - do you know of it's function? and what effect on the test this has. I never removed mine, if my car has one of course!

With regards to the O2 sensor I've placed the order already but unfortunately my local dealer has no stock, so next week should see further testing and comparison

10 seconds seems slow but I certainly agree it's response has to be much slower than a sensors output with the characteristics of a typical O2 sensor. The control system has further interesting factors, such as, although the engine prefers to run at 14.7:1 infact cats actually need O2 to operate correctly so I guess that's one reason why we see high swings on the O2 sensor o/p - and of course this sensor also has a very high transfer ratio around the 14.7:1 mark too.

Can't wait to get a little more time to carry out further testing now...

Cheers

The purge valve controls air flow to purge the evap cannister of stored HC. It only opens above a certain coolant temperature (typically 50C) and is controlled by a "ported" vacuum signal line that does not open the valve until the throttle is opened. If the purge valve is not disabled the duty cycle above idle speed will be affected to one degree or another depending on the amount of stored HC in the cannister.

Even if the A/F ratio was kept exactly at stoichiometric, there would still be a fraction of a percent O2 in the exhaust since the entire mixture is never completely burned. This is due to local deviations in A/F within the mixture and quenching at the combustion chamber boundaries. Thus, the engine out emissions contain a small amount of HC, O2, CO and NOx. A three way catalyst actually works best if there are slight rich and lean periods as they store O2.

I note that your test results include O2 down to one hundreth of a percent (0.01 percent = 100 PPM). In the California test they only report O2 down to one-tenth percent (1000 PPM). Assuming they are rounding, a reading of 0.0 percent (which is what you want to see) means that there could still be up to 499 PPM O2 remaining, which could further support oxidation reaction if the catalyst was hotter or otherwise more efficient, but most engines will be under the test limits if O2 is less than 0.1 percent.

If the O2 content is reported at 0.1 percent, that means that there is still about 500 to 1499 PPM O2 remaining, which should support considerably more oxidation reaction. My rule of thumb is that the difference between 0.1 and 0.0 percent O2 is about 30 PPM HC.

Duke

Duke2.6 · #9 04-10-2005, 12:40 PM

Quote:

Originally Posted by LeaUK

To prove that unmetered air would be compensated for I pulled one of the vac hoses of the intake manifold - and watched the duty change down to 10%, or of course up to 90% depending how you've connected your DVM/Scope and also if your reading negative or positive going duty cycles.

For me it was irrelevant, as to see the change right in front of my eyes was great.

BUT - from pulling a hose (only a 2mm diameter hose) it took nearly 10 seconds to complete adjustment and settle, or at least display the adjustment. So why so slow? Closed loop systems are usually capable of reacting much quicker.

Is this dependent on O2 sensor reaction time? Thinking about it I was at idle, but still 10 seconds!?

Any thoughts welcomed.

Lea

I doubt if there's anything wrong with your system including the 02 sensor. Measured emissions are commendably low, but your standards seem unusually tight. For example, in California the CO limit for cars from the late eighties to early nineties is near 0.75 percent.

MB specified a replacement interval for O2 sensors because in their first decade of use degradation was not well characterized, and no one knew how long they would last, but they turned out to be more durable than originally thought. My recommendation is to not change them unless testing shows that they have definitely failed or their response time has become sluggish. The addition of on board diagnostics has eliminated the need for specified O2 sensor replacement since modern OBD systems monitor their performance.

I don't know how quickly the system should respond to a signficant lean condition, such a pulling a vacuum hose, since there is no MB published spec, but all feedback control systems are designed with "damping" that limits the rate of correction. The lambda system is actually designed to "hunt" back and forth centered on lambda = 1.0, but the nominal deviation its designed to correct is very small. Also, your "test" pushed the lambda system to near, if not beyond, the limit of its control authority, and the conditions are not something that the system would have to deal with in normal operation. Only an engineer with indepth knowledge of the lambda system design could answer questions about system response time to various deviations in lambda.

Given your test results, I would say leave it alone, but your experience is another illustration of the importance of "conditioning" i.e. making sure the catalyst is a hot as possible before the start of the test. The LH system may be inherently cleaner than the KE system. I've never seen a KE system that was this clean! Many KE systems in the US just scrape by the limits, especially HC.

Duke

JimF · #10 04-10-2005, 01:06 PM

Quote:

Originally Posted by Duke2.6

MB specified a replacement interval for O2 sensors because in the first decade of their use, degradation was not well characterized and no one knew how long they would last, but they turned out to be more duration than originally thought. My recommendation is to not change them unless testing shows that they have definitely failed or their response time has become sluggish. The addition of on board diagnostics has eliminated the need for O2 sensor replacement since modern OBD systems monitor their performance.

rates their O2 sensors for 60k miles; MB doesn't make them. Sure, to be positive that you need to replace it, it should be tested. If I remember correctly, Art Dalton described a test that he has done to see if its response time is ok.

Quote:

Originally Posted by Duke2.6

I don't know how quickly the system should respond to a signficant lean condition, such a pulling a vacuum hose, since there is no MB published spec, but all feedback control systems are designed with "damping" that limits the rate of correction. The lambda system is actually designed to "hunt" back and forth centered on lambda = 1.0, but the nominal deviation its designed to correct is very small. Only an engineer with indepth knowledge of the lambda system design could answer questions about system response time to various deviations in lambda.
Duke

The term "hunt" is probably a bad term because "hunt" means that the system will never lock. I'd offer "correct" as a more appropriate term since when it's 'locked', it corrects. When it's OPEN, it 'hunts'. As an engineer with expertise in servo systems, that would be a better use of the more correct terms.

As I said, the overall system should correct a few seconds slower than the O2 sensor. If I recall correctly, a good O2 sensor will correct in within a second from the extremes; if it takes 10 secs as an example, I would say that it needs to be replaced. So would Bosch.

On my car, that test 'corrects' within a few seconds. I have a new O2 sensor with 5k miles on it.