- June 29 -
It's been a while, hasn't it...
In order to address the emissions issue, I have been busy increasing my debt. I have purchased an exhaust with a catalytic converter in it (so that I don't have to keep borrowing Ian's). I have also purchased a Power Commander, so that I can control the engine's fuelling. And I have borrowed, with a view to buying later on, an Innovate LC-1 wideband lamdba sensor controller, so that I can see the effect of altering the fuelling.
The initial goals are to use the latter 2 widgets to map the SVA emissions test sites (Idle and 2500rpm fast idle) to have a Lambda ratio of 1 or slightly less - SVA requires that it averages out at between 0.97 and 1.03. Hopefuly the catalytic converter will mop up enough HCs and CO to pass those requirements as well.
Later on I shall use logging software and ChrispyG's log analysissoftware to build up the complete map
So - to start, I used this evening to fit the PC and the LC-1. Both can be connected to a computer, so I did so. The PC software sees the PC fine, but the LC-1 logger software doesn't seem to see the LC-1. Strangely, one of the other Innovate tools, the LM Manager does appear to detect the LC-1. I shall consult the Cam7 oracle for ideas...
getting the new bits to work
- July 17 -
Well, despite the lack of website updates, I have been able to try a few more things to try and sort out this emissions stuff.
The LC-1 problem turned out to be that the logger software isn't supposed to see the LC-1...
So, William came round with his LM-1. We plugged it in, got the engine running, and William noticed that the lambda readings were very erratic - more than he would expect. Other observations were that the likelhood of a misfire increased as the engine warmed up, and that a misfire was often accompanied by the LM-1 resetting itself. We hummed and hawed for a bit, before William decided that my throttle bodies needed balancing. We listened to them through a rubber hose, and it was clear that they didn't sound balanced.
William was also worried about the LM-1 resetting itself - something it only normally does if it doesn't see enough volts. Perhaps the battery isn't very healthy. It's quite new, but perhaps was irreversibly damaged by being run completely flat on 'the bad day'
A few days later, I borrowed Dan Bromilow's 4 way carb balancer, connected it up to the IAPS take-offs on the throttle bodies, and sure enough, they weren't balanced. After several many lots of minutes I eventually got them balanced, but found it necessary to bump the hot idle speed up to 1400rpm for the engine to be able to run without stalling.
A few days after that, I decided to try and do the fuelling changes necessary to get a solid lambda of 1 at idle. However, when I started, I realised that the lamdba reading was still too erratic, and there were still misfires.
So, the time to investigate the electric supply had arrived. Following the Suzuki service manual, I determined that the alternator appeared to be OK, delivering the requisite number of Volts and having the right coil resistance at the specified engine speed with no load. I couldn't unfortunately test the regulator as the test involves switching the meter to some sort of Diode setting that I don't have.
One point of note is that the tests are carried out at 5000rpm, which is an indication that the alternator isn't running at full whack until there or thereabouts - so when (if...) I drive I'll have to keep the revs up to keep the battery charged. Officer.
I stopped the engine to reconnect the alternator cables - and couldn't get it started again. Lots of petrol, which occasionally appeared to manage to catch fire, but clearly there were no sparks to speak off, and the only thing I was succeeding in doing was pumping petrol out of the tank and out of the garage door via the exhaust...
No choice for it then but to try and get hold of a better battery. Fortunately, my housemate had a spare one. I charged it up, connected it up, and ... nothing. Not a sausage. (He's not called 'my useless housemate' for nothing.) It read 12V without any load, but didn't appear capable of delivering any current.
So I gave my existing battery a top up and reconnected it. The engine ran for 2 minutes, and then stalled and refused to start. I checked the voltage - 12.5V. I put it back on charge, and the Ammeter lept up to 5A and stayed there. I don't know much about electricity, but I don't think that's right...
So. What now? It appears that the engine needs a good solid voltage, so I need to source a good battery and see if that fixes it all: the stalling, the misfiring, the erratic lambda reading, the funny smell coming from the bins...
- July 19 -
William brought round a monster battery yesterday, which I left charging for 24hrs.
Into the garage, connect the battery, turn the engine over. A brief, lumpy, unhappy grumble. Try again, the same. Try revving the engine - stalls.
The sound is reminiscent of a time when I disconnected the IAPS whilst the engine was running (just to see what happened), so I try without that connected. Engine roars into life, idles for a 20s, and then stalls as the idle control reduces to minimum.
Swap the IAPS for the (identical) normal APS. Exactly the same - so it's not the sensor itself failing, unless both have failed simultaneously.
What does that leave? Dodgy loom, and dodgy ECU, I suppose. I guess my first port of call is to replace the wiring between the sensor and the ECU... Oh joy.
getting worse still
- July 24 -
Another disappointing day. Having had the Cam7 mob around to help diagnose the problem, I bought a new set of spark plugs yesterday, and fitted them today. Now the engine at least fires, but with the same symptoms as before - will only catch if the IAPS is disconnected, and then dies shortly afterwards.
As I don't want to kill another set of plugs, I stopped trying to figure out what might be wrong and came back in the house.
getting high tech
- August 1 -
Tim came round a few days ago, and we poked an oscilloscope at various potential offenders, such as the Crank Position Sensor and the injectors. All gave a clean bill of health. The conclusion of the evening was that everything was fine, except that fuel was no longer getting to the cylinders.
Subsequently, I tested the fuel pressure, and during this process, I found that the tank had run completely dry, which explains why the engine had stopped running altogether . A fresh batch of fuel, and the engine ran fine - in fact, I've never heard it run better.
trying to set the fuelling
- August 2 -
Today, full of optimism, I went out to the garage to set the fuelling for emissions.
The engine still ran fine whilst warming up, but after it was fully hot, it occasionally misfired. Twice, it stalled completely, coinciding with the fan kicking in.
Still, it ran well enough to have a good stable lambda reading, which I was able to tweak to centre just about 1.0. But whilst the fan is running, the reading drops to 0.96
Which is strange.
If I had to rationalise a change, it would be an an increase, as the fan sucked up electrons and left the injectors with less of them.
The image below shows a typical section of the log of the session. On the left side, it's running at idle, centred slightly below 1.0. Then the fan kicks in and the engine stalls. A spike as I restart, and then it settles to a lower value as the fan continues to run
And fast idle? A joke. Massive lean spikes.
trying to set the fuelling again
- August 6 -
Cam7 suggested that the richer running when the fan kicks in is the ECU (over)compensating for the drop in voltage. Since the voltage shouldn't really drop that much, they've ordered me to fit a new battery. They've also ordered me to make sure there's a good earth from the engine to the chassis. They've also ordered me to replace the regulator, but I really don't want to as it was bought new, and I've still not proved that it's the most likely cause of woe.
A friend, Greg, has also emailed me with some suggestions. In particular, he's confirmed a suspicion I've got - the lean spikes at fast idle could well be due to one of the cylinder misbehaving, rather than an electrical problem. A duff regulator, for example, would cause all the cylinders to misbehave.
Before investigating these things further, I measured the voltage that the ECU sees, and it was a touch low. I disconnected the little bulb that tells me the ECU is on, and got a little bit more voltage for the ECU. I disconnected the ignition light and got a bit more. I changed to the new battery, and got a lot more (we're up to 12.5 volts now, from 11.2)
Next, I stuck on a starter cable from the engine to the chassis to act as good earth.
I tested the regulator diodes - they're all as expected. I retested the alternator coils - they're all well within spec.
Running the engine, I get a good solid 13.8V across the battery, at all revs. With the previous battery, the voltage wasn't even making it to 13V. So it would seem that Cam7 are at least a little bit right.
Poking the multimeter around at various connections, everything seems in order, so I turn my attentions to lambda readings
And, a bit of a relief, we seem to be at a good stable platform to work from. The lambda reading is unaffected by the fan kicking in. Hopefuly we've eliminated any electrical problems.
After quite a time of trying to fiddle with fuelling at normal and fast idle, I come to realise that I'm not going to win though - there's still something wrong. Normal idle seems to be good and stable, but at fast idle there are still a plentitude of lean spikes - nothing like as bad as the previous graph, but still pretty bad.
I decide that the next thing to do is get some more petrol, and do some more investigating tomorrow
- August 7 -
Today's stroke of inspiration was to disconnect each injector in turn whilst running the lambda logger, and observe the effect. Once I turn off the one that's causing the lean spikes, I should get a stable reading - brilliant, huh?
It so nearly worked out that way as well - what I actually found was that whilst disconnecting most of the injectors caused a big increase in lamdba (engine appearing to run much leaner due to excess oxygen coming from the unfuelled cylinder), disconnecting #3 made almost no difference. The conclusion I drew from that was that some, but very little fuel was getting into the cylinder
I found this a bit weird, as the temperature of each exhaust header (measured using an infrared remote thermometer thingy) was the same.
There was no reduction in spikiness.
Anyway, I swapped the injectors for #2 and #3 around, and found to my surprise that now all were behaving the same - disconnecting any one injector causes the lambda reading to go very lean. Slightly worryingly, not each by the same amount, but definitely by a healthy amount. I'm hoping this means that there was some sort of blockage or goop stuck in #3, which has now been dislodged. As for the difference between the readings per injector, I'm hoping that's due to the position of the lambda sensor - I'll investigate that another time
OK - let's look at fuelling again. Normal idle, no problem, I have a stable lambda value, I fiddle with the numbers and the lambda goes where I want it.
But fast idle - woe, woe, woe. We still have lean spikes, but in a curious pattern - there's a cluster of them for 5-10s, every 30s or so. I stare despondently at the lamdba log scrolling past, and the pattern starts to get less clear - the spikes clusters get closer together, merging into one continuous spikey mess
But after a few minutes I notice an interesting thing. A spike cluster always preceeds the fan kicking in. Once the fan turns on, the spikes vanish, until the fan turns off, when they come back. As the engine has been running now for about an hour, the fan is kicking in more and more often to control the temperature, which could be why the spikes clusters are getting closer together.
The most obvious candidate is the Crankydash - it's the only device which can know that the fan is about to kick in, because it's the thing that controls it. But disconnecting it makes no difference.
But then something else happens. As the fan is not under automatic control, I switch it to continous operation, and the spikes vanish (theres just a very occasional short spike). Turning the fan off, they come back, and don't go away. I turn the headlights on, and the spikes vanish completely. Not a hint of a spike.
So it would seem that the spikes aren't a pre-cursor to the fan turning on - they're always there, but turning the fan on supresses them. Turning on the lights, and increasing the load supresses them further.
I've asked Cam7 for their opinion, but I suspect they're going to blame the regulator.
What I'd really like to be able to do is log the voltage and confirm that the lean spikes coincide with a voltage fluctuation, and then trace the cause of the fluctuation. William's LM-1, which I'm using to log the lambda stuff appears to have the capability to do this, but isn't currently set up for it. I'd better ask him before I fiddle with it...
- August 8 -
I rang Hallens today to order a new regulator, and was a little gob-smacked to discover it would be £141. I think, for now, I'll stick to the strategy of keeping the lights on whilst doing the emissions test.
No other bad news to report today though. I went out to the garage to set the fuelling, and the fuelling was set. Only tricky thing was that everytime I increased or decreased the fuelling at a given engine speed, the engine would change speed. Grrr.
- August 21 -
I bought a K1 regulator (from an 8000 mile bike) off eBay a couple of weeks ago, and thought I'd try it out instead of the new K2 one I bought before. No hints of any spikes, so I'll use the new (old) one and keep the old (new) one as a spare.
This page last updated on: Sunday, Jul 16 2006