Steps forward (and the odd backwards)

I started tuning the car this weekend. Running blended TPS and MAP. This means two tables (one is TPS versus RPM, the other is one is MAP versus RPM) that are multiplied together to determine the fueling. This is a quirk of ITB, as for many installs that use a single throttle body the MAP provides all the information needed. I guess that the flow isn't well formed after the ITB, so it isn't a simple relationship between the MAP and the amount of air going into the engine.

Anyway, tuning blended is tricky as all the tools assume a single table. Not impossible though.

So, I started with some kind of credible table.

Then Ran autotune on the MAP vs RPM, which improved things. Then I did the same again using more throttle and rev range, and things got better. Then I collected a datalog and used MLV (megalogviewer) to optimise the TPS vs RPM table.

All was reasonable. Then an autotune of the MAP vs RPM, and finally things were running nicely. It seems that I keep trying to change up to a higher gear. The car pulls nicely at 2kRPM and I get positive boost at 3kRPM even on a closed throttle, so the cruise is quite relaxed. There is a nice chunk of power at 4kRPM which builds in a quite scary fashion from 6kRPM.

I have put an dB killer into the exhaust. Two reasons. 1) neighbour annoyance reduction device 2) increased backpressure in exhaust. The second of these is counter to turbo logic, but it means that the previously inadequate wastegate now manages to control the boost almost adequately. Previously I was getting 0.7atm boost at 8kRPM, now I am getting 0.5atm of boost. In theory the spring should limit this to 0.33atm, but at least it is sensible.

The other thing that I wanted to do was to draw cool air into the intake. The intake is adjacent to the exhaust, and although the airflow should steer hot air away from the intake, it was always a worry. Now I have boxed in the filter and it only gets air from the outside (behind the front wheel, which isnt' ideal, but at least its cool). It turns out that this makes no difference at all!! The turbohousing itself gets pretty hot, and the pipework in the engine bay is aluminium and also seems to collect lots of heat, so although nicely made it seems to offer no benefit! According to the logs I was getting air temps of 58deg before and now they peak at 61deg. I am running a bit harder now, but either way it is rather disappointing. C'est la vie.

Back to the tuning. Car is now running like a demon to 7kRPM. Clearly I am a bit intimidated because the logs show me backing off the throttle as the revs rise. The problem is that at around 8.3kRPM things go a bit ugly. Stuttery with reduced power. Even the MAP drops off sometime. I tried adding more fuel. I don't want to lose the engine to detonation and there were some lower AFR readings, although they were all over the place.

Next I tried reducing the advance by a further 3 deg. At the moment I have it taking 5deg when at 8psi, and the base table is retarded by a couple of deg versus the NA set-up.

Here is a screenshot from Megalogviewer

The RPM gauge is wired into one of the coil channels (low tension) and it doesn't stutter. I don't know about the other channel.

The sparkplugs look alright, slightly red for 1 and 4, not sure what this is from (slightly worried because the anti-freeze is red) but hoping it isn't from that! The leads measure 16kOhm from spark plug to spark plug (dyna green 3Ohm coils) in both cases.

I tried increasing the dwell from 40% to 50% and that made no big difference, maybe some of the changes improved things a little, but nothing massive in any cases.

So, not sure what I have to do to fix this problem. I think it might be the spark blowing out, so have replaced the plugs with Iridium plugs, and closed the gap down slightly. But haven't tested that configuration as yet.

While I think about it I think I will fit an intercooler. From the Stoneleigh show I realised I could copy the way that Raw-Striker had fitted the intercooler to their show car, that is, it can be wider than the chassis at the front, which will make sourcing an intercooler much easier.

Also I need to fit heat shielding to the bonnet and paint the bonnet on the underside. Quite a bit of tidying to do here, but I measured the temperature at 96deg C (stopped after a run) which is scary hot and I don't want to burn my bonnet (or any of the car).

Car started reliably. Idled nicely. Didn't overheat. Really feel like I've turned a corner. Slight oil-leak, but it looks easily fixed! Happy days.

New timing wheel, glowing turbo

Reinstalled the 12-1 wheel, and started up. The throttle needed opening pretty wide to get a stable idle, which was surprising.

Went for a drive and it wasn't pulling very well. Fueling not great probably, so I optimised it at the side of the road a bit. Helped a bit, but still not great.

Then it started getting dark, and I noticed the exhaust glowing really brightly, this heat went all the way back to the headers which were glowing (less brightly, maybe blood red). There are air holes in the bonnet and in the dark this was really visible. So, I stopped by the side of the road and advanced the ignition by 5 degrees. Immediately the RPM shot up and I had to close down the throttles. Clearly the timing was way out and so I added another 5 degrees. Engine started much more crisply, and turbo glowed less, added another 5 deg which only helped.

I need to get the trigger wheel timed in properly, and also put some insulation on the underside of the bonnet.

According to my calculations the turbo is getting up to between 745 and 790deg C.

I need to sort this out before I cook the oil and block the turbo from getting any oil.

This also reminds me that I need to sort out the insulation for the underside of the bonnet.

trigger wheels

I ended up going back to the original trigger wheel. This is a 12-1 wheel (that is teeth spaced at 30degrees but only 11 of them, so 12 teeth minus 1).

I rebuilt the LM1815 trigger circuit, this time putting in the proper input filtering 15kOhm with 2nF(why do they refer to these as 2000pF) capacitor. This yields a low pass filter with a cut-off of f = 1 / (2*pi*2nF*15kOhm) = 5.3kHz. As there are 12 teeth and max RPM is 12000rpm, this corresponds to 200 engine rotations per second, and hence 2400 teeth per second, each tooth gives sinusoid so the cut-off seems to make sense.

I am also using the adaptive functionality of the LM1815. My hope is that this will be less sensitive to slow cranking (something that can happen as the battery goes a bit flat). This does seem to work, and I have blipped the throttle to give just over 6kRPM.

Keen to test, but too many other things on at the moment.

In other news:

Refitted the exhaust and bolted in a baffle at the back end. The car is a bit loud at the moment, and also I am having some boost creep problems. A little more back pressure may help with the boost creep, and the baffle does help with the noise.

Also welded up a couple of holes in the down pipe next to the Lambda probe.

Next plans:

Sort out the tuning. Bit worried about the spark timing, as it is a pain to configure.

Paint underside of bonnet and fit heat resistant shield

Install intercooler brackets, work out pipe runs and get intercooler welded up.

Post Stoneleigh fixes

Bought a replacement fuel pressure regulator. Too many question marks over the cheap McGill one, so got one from FSE (Malpassi) which seems to be a reasonable quality product.

Need to vent to the engine inlet (for environmental reasons mostly, but also because with present venting it gases the driver).

Compression isn't great on this engine. 8.6, 9.6, 8.0, 9.3 but all engines are test engines, so it doesn't matter much!

Need replacement manifold bolts, presently running a grab-bag of random bolts.

Need to sort out the trigger wheel above 6kRPM. Its a bit strange as this worked before, and now doesn't. Probably plan to go back to the old 12-1 trigger wheel design as that worked pretty well and ran smoother than the suzuki wheel ever did.

Bonnet insulation requires sorting because the bonnet got super-hot the other day. I've got the stuff, but want to combine it with painting the bonnet, fixing the damage, and insulating it.

One of the clamps on the exhaust fell off on the Stoneleigh run, so I need to sort that. Probably also fit an exhaust baffle to quieten things down that may even help with the over-boosting that I have. Not sure, but it was cheap from a mate and easy to fit.

Stoneleigh success (ish)

I got to Stoneleigh and back.

Trigger wheel failures above 6000rpm.

Tune isn't great.

Exhaust clamp fell off.

One fuel injector is gummed up

Engine vents gases into cabin

Problem with hot starting (trigger wheel related I think)

Nothing that isn't easily fixed. The only bad one is the "heavy breathing". It seems to be blowing off something more that air. Not sure what is going on there, but it may improve. COmpression test wasn't great yielding 9.6, 8.6, 9.3, 8.0. Which is disappointing.

Interestingly I did a massive datalog all the way to Stoneleigh (50mins) which showed up some interesting tuning features.

I had read that you could run a tuning map that was based on MAP and RPM. It actually turns out that TPS plays a significant role (my finnish buddy Arttu told me this, but he is in the minority of 1). I have put the whole log file into the computer and simulated different maps. It turns out that Alpha-N with multiplicative MAP scaling is pretty close to perfect. I am going to run a MAP table so that I can tidy up the low speed running (where MAP works really nicely).

If you look at the megaviewer screenshot, you can clearly see the AFR (air-fuel-ratio) modulating with TPS even though MAP stays pretty constant. When running the TPS didn't affect fueling, and clearly it should.

Falling towards Stoneleigh

The Stoneleigh kit car show (Warwickshire showground) is the "largest kit car show-- in the world", and is a good place to meet up with virtual friends from the internet world with a common interest in car stuff. It is free to attend if you arrive in a kit car, and I have never managed to take advantage of this (this is only the second year that I have had a road legal car).

Anyway, the plan had been to put the new engine in map it on the road, and then drive to Stoneleigh. Simple.

That is still the plan, but there was set-back last weekend when the triggering failed and I ended up getting towed home. (BAD).

I have since fixed the ECU. It turned out to be a dead ignition driver, that combined with carboned up plugs and a low battery. I still don't really understand why the triggering wasn't working, but when an ignition driver fails I think it can stuff up the earthing, which can corrupt the trigger. The triggering circuit was designed by me (and hence a bit random), so I assumed it was that, and adjusted a couple of things. Anyway, I tidied up some of the cable routing too, and sorted the shift-lights, also I built a megasquirt simulator, and used matlab to output representative trigger signals (via the headphone output) for the MS to trigger from.

So, after installing and testing, and testing and installing the engine fired up from its own battery this morning. I think the batteries (both the normal bike battery that I run, but also the backup battery, which I use when starting is a problem) were flat, so have charged those, and have ordered some Anderson connectors (these will allow me to use the bigger battery whilst driving, as the jump leads can fall off, so I presently can only use the big battery for starting).

Anyway, back to the plan. Here is a list of things I need to do before I can leave:

1) Oil change (the present oil is contaminated with everything, it has seen over fueling, a track day, about 1k miles, and most recently a cracked barrel), I get lots of interesting vapours from the vent which are very petrol based). So a new filter and oil fill are a necessity.

2) New lambda sensor. The cracked barrel put water into the exhaust which knackered the old Lambda sensor. This is a bit annoying because they are expensive, but it is essential to have these signals to map against, so I got a new one. Hopefully this will last.

3) Mapping. This is the big problem at the moment. It seems that it should be possible to map it using the MAP (manifold pressure) (also known as Speed-Density), but then again some folk say you need to take the TPS into account. At idle MAP is fine as the turbo isn't doing anything, at high RPM its fine as the throttle is open and its all going on, at intermediate revs is where the problems happen (4kRPM) as the throttle is fairly closed but the turbo gives a bit of boost. I think I need to include TPS (megasquirt allows this).

The problem is that although I can probable tune using MAP before Sunday, but tuning to include TPS would take too long. So I am hoping that I can manage the "cruise" settings. A 70mph cruise is about 5.8kRPM so I think the turbo will give about 5psi of boost, so that should take me out of the difficult region. But it could make for an interesting drive!

Hopefully, I can go and do a few datalogged runs, which should allow me to get the tune to a point where the car is driveable. If not then its Stoneleigh 2012.

4) Other tasks. Bolt down the ECU. Set up the sat-nav. Install speedometer (optional as sat-nav). Get together a tool-kit and spares. Tyre pressures. Top up water levels.

A bit nervous (silly really as it doesn't matter!)

Engine runs, but got home on wrong end of tow rope

Lots of progress. Mostly good, some bad.

The last set of cylinder barrels got cracked owing to boost combined with already high compression, and me being an idiot by not stopping sooner.

So, I built up a new engine.

1192cc barrels (up from stock of 1074cc)

New head gasket

3mm base spacer (to lower the compression)

Stock head and valves (not the big ones)
gsxr600 TBI's

Sorted out all the little things (like valve shims), and valve guide seals (gsxr1000k5 ones are the same, wemoto sent me the gsxr1100 oil cooled ones, which are wrong, so look out for this).

Got the engine in, topped up with fluids, bolted together.

Fuel leak, fixed.

Fuel pressure too low (fuel pump leak inside tank) fixed.

Throttle position (too wide open due to sticking choke cable) fixed.

Got it to idle, and seemed to idle lower than the last (Kent) cams at around 1100rpm. Throttle is still more open than is ideal, but its a detail.

Finally (4pm Monday) got out, went to fill up. Car ran poorly. Not smooth (felt like 3 cyls) but reved a bit and seemed to sort itself. Filled up with fuel and the car did restart (a bit surprising). Thought about heading back, but decided to do a longer up the bypass run. All went fine, rubbish jumpy spot just as boost comes in (3-4k), so that needs some mapping work (unfortunately the O2 sensor isn't working nicely at the moment, maybe needs calibrating).

So, I datalogged, and got 10psi at 8.8kRPM, which should be worth a few BHP. At the moment I don't have an intercooler and intake temps got to 65degC, but the lower comp engine didn't seem to mind (no detonation). The engine gave decent power from 5kRPM, and popped and banged like a crazy thing on the over-run (quite nice, but probably should turn it down a bit). I think the fueling is rich up top, not sure what is happening at 3-4k though.

Bad news, going around the Woodstock roundabout there was a bit of a plastic burning smell, not sure what that was. Then coming back along the dual carriageway I stopped at the traffic lights and the engine idled fine. Then it cut out, and the lights changed. Tried to restart, but failed. Eventually pushed to side of road and tried to fix the problem.

Not sure what it was but ECU wasn't happy, and it dumped gallons of fuel into the engine (bad news). Couldn't get the ECU to recognise the trigger, which is a wierd fault. I have a few suspects:

Blocked injector (I was messing with the fuel lines, so could easily have blocked an injector).

Megasquirt hardware problem (need to bench test it)

I'm keen to get the car to Stoneleigh.

List of things to sort:
Lambda sensor (I need a better map if I am going to go any distance)

Megasquirt (I need to get the engine running again)

Speedo (this is needed as at present I have to guess the speed). Not essential.

Oil and filter change (maybe fix the odd tiny oil leak whilst I am at it)

So, mostly good, but lots to do.