Measure Twice, Cut Once
Hello DN members,
I have a case study to share. This one had me confused for quite a while, but in the end I learned a valuable lesson, and now you can learn it too!
When checking CKP to CMP correlation with a scope, I normally choose any ol' reference point in the CMP pattern and use that to compare to a known good capture. As long as I am using the same reference point in both captures, then a comparison can be made…right? Well, let's think about this. Does it matter if we choose a rising edge or a falling edge of the CMP pattern as a reference point?
Let's get to the details of this case study. The above listed car came in with a P0016 Cam/Crank correlation fault. One of the techs here diagnosed it as needing a timing chain. Seems reasonable. These engines are notorious for stretched chains, broken guides, and failed tensioners (among other things). The repair job ended up being handed to me, and sure enough the chain was stretched, one of the guides was broken, and it still had the original tensioner design, which is prone to failure.
Unfortunately, after the repair the same fault code came back. At first, I thought I must have got the timing wrong, but upon double checking my work, it was all good. Timing spot on.
So, next I check the CMP/CKP pattern with a scope, and it looks like it is off slightly compared to a known good. To be precise, it was off by about two notches of the CKP pattern. The CKP pattern is 60-2, so each notch (or peak) equals 6°, and therefore, being off by two notches is 12°. In addition to that, the scan data showed it to be about 12° off. Looks like I confirmed that there is a problem, but what is it exactly?
This is where I go down the wrong path. It's a long story, but basically we have a spare engine laying around, and it sure is tempting to swap in some known good parts when you have all the spare parts you could want just laying there. Alas, after swapping various parts around, the scan tool still showed it to be off by 12° and the P0016 was still setting.
To make a long story short. The boss said to try swapping the CMP sensor. To my amazement, it worked! Now if this were a Chrysler or a Nissan, then that's probably the first part that would have been replaced, but on German cars, the only time I have ever seen a CMP sensor cause a correlation fault was on an old car that had an adjustable sensor. So, I had to capture a before and after on the scope to see what the heck was going on.
First, the Before/Bad CMP pattern.
Next up is the After/Good CMP pattern.
Remember the question at the beginning of this case study? Does it matter if we choose a rising edge or a falling edge of the CMP pattern as a reference point? How about measuring the width of a CMP pulse? Obviously we should check the width of the pulse, but at first glance the pattern looked the same as the known good (2 small humps, followed by 2 wide humps), and I was concentrating on the correlation, so I didn't get the rulers out to measure the width of the pulses, I was measuring the correlation, and I was using a rising edge of the CMP pattern to do so.
As you can see in the images, the rising edge actually didn't change much at all with the new CMP sensor, but the falling edge moved by 2 notches (peaks), which, as was explained earlier, is equal to 12° of crank angle. With the new CMP sensor installed, the scan tool now shows 0° for the CKP/CMP adaptation, and no fault codes. I'll definitely be checking the rising edge, falling edge, and the width from now on. I think most ECM's use the falling edge for measurement. Maybe all of them do? If anyone knows, leave a reply please :-)
An interesting side note, is that the car was here a few months ago and it ran fine and had no fault codes, so this strange failure of the CMP sensor just happened one day and it stayed that way (with that pattern) ever since, which is kind of an unusual failure to me. Also, the failed sensor was an aftermarket “no name” part. Big surprise, huh?
A great observation! I can't wait to get my 8 channel Escope you would never know what caused that without a scope only that you swapped it and you fixed the car thanks for sharing.
Excellent, i can certainly understand it took time to “see the full picture ” XD I will be using this strategy tomorrow on VW I currently have with a correlation code.
Thanks for sharing that one was not easy. Good lesson that could happen to any body else at least to me , but that's to this post I will check from now on risin and falling edged.
Thanks for sharing Steven! I most likely wouldn't have picked that out either…. I don't completely understand how a different Cam sensor changes the pattern? It obviously does but with the same reluctor / pickup why does it change??
Most likely it is a design flaw of the internal electronics of the sensor. Take a look at the resolution to this post, which could be related.
I would have to agree with you. Had this exact situation on a Chrysler with an aftermarket cam sensor.
The “polarity” (winding direction inside the sensor) error will cause a 1 pulse shift one way or the other. Go the SEMA/AAPEX and visit the International Marketplace downstairs at the Sands Expo and you will see what is seemingly the exact same sensor from dozens of different copycat suppliers. If you cut them open, you will see WILDLY divergent design choices which change the phase, shape and…
Thank you for your great detective work, we all learn from the vehicles that we work on to see what a scope what's going on and to share that information with others thank you
Yes, great job. These jobs are very self rewarding when we figure these out!
Good lesson. Personally, I may end up getting bit someday because I err on the opposite side of your coin. I want to see an exact pattern match whenever I look at these wave forms. If I see even the slightest deviation(like in your capture here) Im calling it bad and am looking for a problem. Now I say it may bite me because its possible that some deviation might be normal on some current…
Great share. I once had an oil pressure sending unit fail, upon replacing it with a Napa part ($70) The oil pressure was good, the new sensor read accurately, but it kept coding. Freeze frame showed the code setting at Key on, Engine off, 0 RPM. (A contradiction to the code set criteria) I thought to myself, how strange that was, there is never oil pressure with the engine off (0 rpms). Why…
Possible amperage draw difference,,, milliamp clamp would be another test… Absolutely outstanding case study, I had no idea this was a possibility!! Thank you.
Bernie Thompson and I have had a long discussion about this exact subject in the past. He once enlightened me to it but I failed to retain what we had discussed. Hopefully he can shed some light?
Every time I watch one of Bernie's videos I realize I'm not being thorough enough when analyzing waveforms. It's easy to miss subtle differences if not paying close attention and measuring.
Steve, Thanks for the great case study. That's something that could easily be missed if not really scrutinizing the waveforms. Bob