Interesting sensor anomaly on R400D

[aerobod - near CYYC]

In tracking down a slight but persistent rough running with my 2012 R400D, I found what seems to be a mistake in the Caterham wiring loom.

The camshaft position sensor uses pin 30 for signal and pin 23 for return in the loom as delivered on my car. Looking at the pinout for the 992/9A4 ECU, MBE show the camshaft sensor return as pin 31, pin 23 is shown as a 0V reference. I can see that using the 0V reference as a return may work, but it may also cause the ECU to ignore the sensor input if it doesn't see the signal between pins 30 and 31. If this is the case, the engine may be running batch fuel injection instead of timed injection relative to the camshaft position.

I have also checked all sensor wiring, grounds and voltage inputs and have added some extra and more direct ground bonding from the engine to the chassis to the battery isolator to the ground on the firewall next to the fuse box to my wideband oxygen sensor ground. This has reduced the noise/variability in the alternator supply to about 200mV at 14.2V (from about 400mV before) and the voltage on the ground between the ECU and engine block (where the worst difference was seen) from 15mV to 8mV.

I still need to replace my failed oxygen sensor, though (wideband Bosch LSU 4.9), until I can see if I've smoothed out the engine across it's full operating range.

[slap_ed]

How did you get on trying to eliminate your rough running?

Did you re-wire the camshaft position sensor pins?

How do you test for noise/variability in the alternator supply?

[aerobod - near CYYC]

I did rewire the camshaft sensor as per the MBE pinout. The variability in the alternator output is implied by the Easimap battery voltage readout, which resolves down to about 20 milliseconds, although the min/max hold feature on my decent Fluke multimeter correlates with the Easimap voltage ripple. I think it is now about as stable as can be expected with the extra earthing I have done between the chassis and main electrical components. In terms of ground variability, from information I have gleaned on the Internet, 100mV of ground voltage between components is about the acceptable maximum, I'm now down below 10mV in the worst case and 0.1 to 0.2 Ohms ground resistance between the ECU, engine ground, chassis, firewall and battery.

Everything is very good now, after a couple of misleading pointers. I had replaced the TPS due to what seemed to be noise at the lowest throttle openings, but in hindsight it seems to be a benign effect from what I believe to be an inductive source that causes a small ripple in the TPS voltage, but no effect on the car from a smoothness or fuelling perspective. The brand new TPS ended up failing after less than a month, but I didn't realize it was causing problems until after I had loaded a new map and went out for a drive, the car was suffering from severe over fuelling followed by surging, leading me to believe that I had totally messed up the map.

A few kilometres from home the car would hardly run, so I pulled a quick u-turn at some traffic lights and headed for home. Unfortunately this u-turn was noticed by the police, leading to me being pulled over. The cop realized that I was having issues with the car when he walked up to me and I hadn't switched it off. After asking me what engine I had in there (my reply was 4-cylinder Ford Duratec with normally 210bhp, but only about 10bhp at the moment), he said that he could see that it wouldn't be easy to put the car on a tow truck, understood why I pulled the illegal u-turn (was good enough to not even mention a ticket or warning) and wished me luck in getting it back home.

Back to the tuning woes... I luckily checked the Easimap dashboard while still trying to keep the car running when I got home, with closed throttle I was not at throttle site 0, but typically site 5 or 6, the new TPS had gone bad, not dropping below 2.2V or so at closed throttle. After swapping back to the original TPS, I went out for a couple of hours of data logging, paying attention to part throttle running in general. I dumped a couple of hundred thousand data points at 50ms intervals into a spreadsheet and analyzed the lambda settings from my freshly installed and calibrated wideband oxygen sensor. Based on this data logging I did a bit of fuel reduction from idle to 3500RPM at throttle sites 0-5, fuel addition from 5000-6500RPM at mid throttle sites and set closed loop lambda at 0.95 from 1000-2000RPM when throttle sites are 0-8 and lambda at 1.0 from 2250-5000RPM when throttle sites are 0-8 (beyond 5000RPM and throttle site 8 lambda is open loop).

The car now runs brilliantly, idles evenly, pulls linearly at all speeds and trickles away from rest smoothly with virtually no throttle needed. No noticeable fuel smells, flat spots, surges or any other annoying characteristics I have found. I may tweak the cold start fuelling for temperatures below 10C, as I need a slight touch of throttle for it to start immediately when it is between -5C and 10C. I may also put it on a rolling road next spring to confirm the timing is optimally tweaked, but there is no sign of pinking or lack of power across the rev range.

I am now very happy with the way the car is running, very good compared with any vehicle I have ever owned and on par with the linear and un-burstable feeling the S54 engine in my old BMW Z4M had.

[aerobod - near CYYC]

Here is the lambda output for data logging at lower revs to mainly verify closed loop operation, although there is one full throttle value at 0.88 lambda that had at least twenty data points to give a stable reading:

G

The other important quadrant is the high rev/ high load area, the lambda readings are looking quite good here, too, aiming in the 0.85 range under full load:

Here is the current base fuel map:

http://photoshare.shaw.ca/image/7/b/a/238530/image.jpg?rev=0

The current base ignition map:

http://photoshare.shaw.ca/image/7/b/a/238530/image-5.jpg?rev=0

And this is the lambda target map, with a zero target setting the open loop area:

http://photoshare.shaw.ca/image/7/b/a/238530/image-6.jpg?rev=0

[Smiffy69]

Thanks for the update and an interesting read.  Slight thread hijack but how do you get these plots in easymap?  If I try and download my ECU it gets to 50% then gives up.  I had assumed that this was as a result of them being "read only"

Am I missing something and this is "user error" ?

or do you know the password ? 

[aerobod - near CYYC]

The maps themselves definitely require an unlocked ECU. I tried all combinations of 4 and 5 digit codes using an automated script against the original 992 Caterham ECU in my R400D, no code would unlock it, I believe it is directly flashed as opposed to being flashed by Easimap. I replaced the 992 ECU with the MBE 9A4 ECU, which is fully plug compatible, these are the maps I'm showing in this thread.

The spreadsheet info relating to data logged can be collected from either the 992 or 9A4 ECUs in the same way, using the export function in Easimap (I normally aim for about 50,000 output lines from a data logging session by adjusting the time interval, typically to about 50 milliseconds). The export CSV data is then put into a spreadsheet to build a pivot table of, for example, throttle site vs engine speed for the appropriate logged value such as lambda, injector duty cycle, ignition advance, etc.

[Smiffy69]

Thanks Aerobod - As mine is the CC ECU its locked so no mapping exports for me then.

[slap_ed]

Aerobod

By correcting the crankshaft sensor pins alone, did this help smooth out the rough running at all?

[aerobod - near CYYC]

I think good grounding helped a little, but the pin swap would use the cleaner logic 0V reference as opposed to the chassis ground, so if the chassis ground is also quite clean it shouldn't make a difference. What I have noticed is a lack of short ECU resets compared with before re-pinning and adding better grounding. These resets typically seem to be about 20 milliseconds and increment the ECU reset counter, they appear as an occasional spike in ECU parameters, which seemed to cause a slight stumble, although they would only occur a couple of times per hour.

Just a general attention to the best possible grounding, best use of the available pins, using some extra cable ties to make sure cables are well routed and secured have I think left me with better consistency between the sensors and ECU. The car is now much more predictable in the way the engine performs than when it left the factory just under 20,000km ago.

[Shortshift]

Well, this probably gets some sort of prize for dredging-up an old thread (7.5 years and counting) but I've found something that I think is relevant here and worth sharing, relating to 992/9A4 ECU pin-outs and wiring.

James (Aerobod) made the recommendation back in 2015 that the cam return should go to pin # 31 ('cam sensor return') and not to pin # 23 ('0V') as our cars all seem to be wired ex-factory from Caterham Cars.  I followed his advice and rewired the pin-out accordingly - oh, how I love working on that 36-pin connector...

However, I've just come across some further information from MBE which shows that the wiring of the cam sensor return to 0V (and not to ground) is very deliberate. 

The pdf is attached MBE9A4 4Cyl COP-1342.pdf and I'll load a lower-resolution image in this post as well.  Food for thought?

 

I'll also make a separate post in TechTalk to make the diagram more readily available.

James

[aerobod - near CYYC]

Hi James, Cam Return on Pin 31 is to the MBE specification as opposed to the way Caterham does it to ECU ground. Not sure why Caterham doesn't follow the MBE spec, perhaps due to some legacy wiring from when they used a different ECU that didn't have a capability for a separate cam return signal.

I assume your diagram came from an MBE installation on a Swiss/German/French vehicle due to the spelling of cylinder as zylinder, battery as batterie and relay as relais, as opposed to MBE in Cirencester. SBD is the main MBE distributor and source of documentation, this is their pin-out: https://sbdmotorsport.co.uk/wp-content/uploads/pdf/MBE9A4-PinoutIssue_F.pdf

Using Pin 31 for the cam return signal will eliminate any noise that indirect return via the ECU ground could cause, but using the ECU ground for return will be fine if there is no induced noise on it.

[Jonathan Kay]

That SBD Motorsport pdf:

MBE9A4-PinoutIssue_F.pdf