Friday 19 November 2010

Long overdue update

On the bright side, the seven continues to provide a rich learning experience, so to speak. The less cheery way of looking at it is, it still doesn't work, and in fact has even regressed in some ways.

After the failed MOT, much time was spent looking over the brakes and trying to figure out why they were seizing. The most significant factor turned out to be insufficient slack in the pedal: since the pedal (and hence the master cylinder) couldn't return fully, once pressurised, the system could never release fully, leaving the brakes dragging somewhat. As they dragged, temperatures rose (considerably), raising the pressure... which is why the car didn't make it to MOT without a rest; nothing to do with the handbrake, which was adjusted just fine.

A secondary factor, which Will and I spent quite a bit of time looking at, was the fact that even with this sorted, I still got about 3 ft-lbs of drag from each rear wheel. (For perspective, the diff on its own, being an LSD, takes about 7 ft-lbs to turn, which is just about turnable by hand using the discs, or not too hard to turn using the bigger radius of the wheels - so this roughly doubles the resistance at the rear). We tried winding back the pistons (no improvement); checking the callipers could slide freely on the pins (they could); checking the discs were central in the callipers (a washer was added on one side as a spacer to achieve this); with no improvement. Eventually I got the bench grinder out and ground the pads back by 1mm or so, trying not to inhale too much, and now they are fine.

Unfortunately a slight mishap occured while up on axle stands fixing the brakes: after everything was sorted, I started lowering the front of the car... and the rear fell off its axle stands. Happily, Emily's bike and Chris' lawnmower took most of the hit and absorbed a lot of the impact, but one rear wing of the car was damaged. It's an easy replace, but very, very annoying.



Brakes sorted and wing gaffer-taped up, I tightened up a couple of loose-ish unions that had leaked under the high pressure, topped up the fluid at the last minute, and had another go with Will following, and Becky looking nervous in the passenger seat. No issues getting there this time, but on arrival Will told me he was seeing big clouds of black smoke every time I put my foot down...

As half-expected, the engine indeed turned out to be running very rich. So rich that I hope the cat hasn't been destroyed - I don't think it's been run enough to seriously damage it, but we'll see; it was certainly another MOT fail (along with some a couple of loose nuts which will be trivial to sort out).

I've spent this week trying to figure out why the car is running so rich. I thought I had found the answer when I looked through the throttle body and saw this:

Oil in the inlet manifold.

But unfortunately, inspection of a friend's healthy 328 revealed the same thing, so I don't think this can be the issue. I've ordered a new CCV valve (the bit that's supposed to separate out the oil vapours before feeding back into the manifold) anyway, which I will fit anyway, as this is probably on its last legs at 100k miles (as indeed the state of the manifold suggests).

I had a bit of a poke around the fuel rail and discovered that all the injectors click at a sensible rate, so it's unlikely that one is stuck (and all the spark plugs are equally sooty anyway, so I don't think it will be a particular cylinder that has issues). Unfortunately I don't have anything suitable to measure fuel pressure, which, if too high, would explain things, but I will look into this.

Finally, last night, I had a bit of a breakthrough. After much research into lambda sensors, I removed mine, and tested them by heating with a blowtorch (which gets them to operating temperature - about 500º - and deprives them of oxygen since it's being consumed by the flame), and measured resistance as I did so (multimeter needs to be set to read at 200kΩ). Initially it should read infinity (lots of oxygen = lean conditions), then, when lacking oxygen (rich conditions), a finite value. One of mine just reads infinity, so it's always telling the ECU that there's not enough fuel...

Fingers crossed a replacement (£80!) will sort the engine out. I'm tempted to not bother and just wire up the working one so that the ECU reads it twice, but I suppose I might as well get it right. Also there's the possibility that the ECU fuels each group of three cylinders differently, as each one feeds its own lambda sensor.

This weekend's plan, then, is to replace the broken lambda, do a lot of tests to see if the engine is functioning properly, tighten up the loose nuts, and then the car should be in a fit state to pass MOT. Assuming, of course, that the cat is still OK - it will be expensive if it isn't.

Monday 8 November 2010

Videos from Saturday


Setting out at 7:30am



Emily drives back into the garage.

Saturday 6 November 2010

Deferred success

The good news is that the engine now appears fully sorted. The two problems that I'd observed - low compression and a faulty MAF can be blamed on a faulty compression tester and a bug in the laptop software. Phew. A lot of time wasted trying to track down non-existent problems due to faulty diagnosis tools, but at least I'm happy with the state of the engine now. Better wasted time than a dead engine, although it means no M3 engine for now...

Pete lent me a different compression tester, which produced numbers in the range 160-175 (cold, dry), which is slightly above what Haynes says I should expect, and all within 10% of each other - so this is completely OK and suggests that there is no compression issue. Assuming, of course, that Pete's tester is not faulty - but given that he's used it before and mine has only been used on the seven, and also that I wouldn't expect it to run at all, let alone with no obvious symptoms, with a PSI of 35 (which is what mine measured), I'll assume mine is broken.

The MAF issue seems to be simply that carsoft wasn't displaying sensible values for airflow on the laptop. We replaced it with a pot, and twiddling the pot made the engine stall, so the ECU is obviously reading it and using the data from it. But still the laptop displays non-changing values as I twiddle the pot, which fairly clearly points to a software bug. Unplugging both MAF and throttle position sensor causes a stall, again, as expected; unplugging just the MAF causes a little twitch in the revs, again, normal. So all signs suggest a laptop software bug.

On Thursday, Will came over and we spent the day going over the car before MOT. Will spotted a potential issue with how the handbrake attached, and very kindly went and made up some pins to attach it with on his lathe (although I secretly suspect he was just missing his lathe after a year apart from it). He also managed to resolve the dragging rear brakes... or so we thought. We also tightened up the suspension bolts, cleaned up some wiring under the bonnet, and probably did some other minor things that I've forgotten.

Today was MOT day. I'd stupidly booked it for 8am so had to get up at silly o'clock to drive there. Insurance sorted (turns out sevens are really cheap to insure, which is cool), we set out, me wrapped up in snowboarding gear. It was wet and cold, so I didn't try giving it much in the way of throttle, so sadly can't report on how nippy the car felt.

Unfortunately, halfway there, the car was getting less and less responsive, and I had to pull over and stop. It turned out that the rear brakes were well on their way to melting - I'm not sure if this was because the handbrake had been on very slightly, or Will didn't get the rear brake adjustments right. I suspect the former, because both were equally affected... wonder if I should have fitted a handbrake switch. After ten minutes, we decided to try carrying on, which probably would have worked, if I hadn't run out of petrol.

I sent Emily off to buy some, and eventually we made it to the MOT centre.

Unfortunately, the short of it is they abandoned the test due to a loose steering rack, and leaking brake fluid. I'm sure the former is just a case of being loosened by the vibrations of being driven for the first time, so an easy fix (tighten the bolts). The latter might well be the fluid having boiled, since I had zero drips in the garage (although of course this was the first time I'd used the brakes significantly), but I will need to investigate this and make sure all the connections are tight.

I trailered the seven home (thanks Simon for the use of the trailer)... repairs start tomorrow; the MOT is now next Saturday. Technically it will be a first time pass as they didn't perform the MOT today, just took one look at the fluid leak and didn't bother, so that's what I will be telling people.

Videos of the first drive to follow...

Wednesday 3 November 2010

MOT with a broken engine

The past week or two has been mostly spent trying to understand the various engine issues.

Firstly, after noticing the quantity of smoke produced, and the appearance of some white-yellow mayonnaise-ish looking gunk under the oil filler, I tried a quick compression test, with alarming results.

Expected range for this engine is 142-156 psi (when hot). I couldn't easily get the engine hot without driving it, so I did it cold / warm, and ended up with values between 35 (cold, dry) - 90 (warm, wet [i.e. with a drop of oil in the cylinder]). This was pretty inconclusive, but:


  • cylinder 2 looks a bit lower than the others, indicating head gasket issues or liner wear
  • all these numbers are very low
Other issues: the car is, or has been running, very rich, hence the smoke.

However, after discussions with Will and others, I've come to the conclusion that there are no symptoms directly attributable to the low compression, so I am inclined to ignore this until I find otherwise, as the fixes are either expensive (head gasket) or very expensive (liner wear = new engine).

We plugged the car into the laptop, and using that and a multimeter, decided that the MAF (airflow sensor) was probably not working, which would explain the richness. Will tried hooking up a pot in place of the MAF and twiddling it by hand to achieve a good idle, and we found that the car would stall when it was twiddled too far, proving that the ECU is indeed using a reading from the MAF. So it was duly replaced with a second-hand MAF off Ebay (new ones are £200, hence the preference for a £20 second-hand one).

However, results with the new MAF are not very clear. The car runs much the same, and again the laptop does not show a change in airflow when I blip the throttle. However:
  • wiring between ECU and MAF has been tested and found good
  • The signal coming off the MAF is in the right range and responds to throttle blips
Tonight I tried hooking up a pot in place of the MAF again, and seeing what the laptop had to say. It turns out that the laptop still didn't display useful data, even though the car would once again stall if I twiddled it too far - the obvious conclusion being that the software isn't displaying a correct value.

So, the way forward seems to be to assume that the MAF is working OK, that the over-fueling is a thing of the past, and see what happens at MOT on Saturday. The emissions tests will hopefully tell me some useful things. It might even pass.

Longer term, I doubt that the engine is in good shape given the compression numbers, but I guess I can always look for an M3 engine when my bank account is a little healthier - I suspect the engine will probably run for now.