Winter Layup - 1995 Ducati 900 Supersport

[OddDuck]

Thanks for that idea. Copper's good (excellent for RFI) but steel's about twice as effective for anything magnetic. Anything conductive will work, given enough thickness, but anything magnetically permeable works best. Which is why I've fixated on steel, difficult as it is...

The idea I've been toying with is to machine up a Y-branch housing (steel) and a bunch of fish spine beads (also steel) and then find a way to connect them mechanically and electrically, then shield the lot against the weather. It's big work and at the moment I can't think of a way to guarantee longetivity.

The alternative (as you suggest) is piping, in that case copper would be a hell of a lot easier to play with as a first try. Certainly 1mm-ish wall thickness copper would be a big step up from the braided shielding. It's also much more likely to age well... One piece construction, can be covered in heatshrink, the ground lead will solder to it without issues... it's got a lot going for it. Well worth a try as a first cut, if this does the trick then there's no need to spend weeks making beads by hand. Thanks again Pete.

The other path to take is separation - get the CDI's and pickup leads as far away from the coils as possible. That option's still open but having heard the motor finally synchronise sort of properly (with the earlier attempt at shielding) I'm curious; can it be done?

[OddDuck]

I've spent the last four evenings moving the oil cooler, as per Kickaha's suggestion - the idea is to provide a flow of warm-ish air to the carburettor bodies while riding.

I fabricated a pair of split-clamp brackets, intended to attach to the lower frame cross-bar, and finally got the oil cooler hoses from Improved Racing out. There was a fair bit of fiddling to get everything into a place which would work. The brackets aren't welded-on frame tabs; this turned out to be just as well. This way and that, rotated forward, back, into the frame cavity or out of it... it took quite a few goes to get to a position that I think will work. I've placed some rubber under the clamp brackets to protect the frame paintwork.

One of the unexpected constraints was providing for front wheel movement under a bump - accomodating the wheel isn't so hard, but one of the brake hoses moves rearward quite a bit. It wouldn't be the best to have this contact either the oil cooler or the edge of the air chute.

The air chute (a cheapo bit of sheet metal, shaped like a ski jump) is probably a work in progress, I can see this going through several iterations. The oil cooler simply wouldn't go to a place where it was in line with carburettor bowls and slipstream, so some kind of an air duct had to be used. It's meant to do three things:

- keep slipstream air from impinging directly onto the carburettor bowls
- guide oil cooler-warmed air to the carburettor bowls
- keep the rec-reg unit alive, by guiding hot air away from it during low speed traffic. Uprising air at the back will tend to suck cool air in at the front instead of letting air rise vertically off the front surface.

Finely-controlled air flow it isn't, there'll be all sorts of spill over edges and turbulence, but before it works well it has to fit into a compartment that is now a plumber's nightmare.

I replaced the anti-vibration mounts. I'd been concerned that the old ones would be corroded into the original valve cover and I'd rip them up while trying to unscrew them, but it turned out to be necessary anyway. The old ones had started to split and fail, with the rubber separating from the metal ends. It wasn't obvious with the oil cooler still in place and some dirt covering up the cracks.

The oil cooler hoses themselves have turned out to be far too long. I've had to place them where I could get them to fit. This is entirely my fault, measuring by guesswork in situ turned out to be a bad technique. Hose flex and placement turns out to be very dependent on length; a couple of centimeters can make a huge difference. I'll try ordering a roll of hosing, fittings etc, and making up my own oil cooler lines later.

One note for anyone else replacing oil cooler or hoses on an old engine... I had a nasty surprise. The aluminium nipples on the cooler had corroded, just behind the tapered sealing faces. It'd be very easy to get corrosion / aluminium particles / anodising etc into the new hoses by mistake, and this circuit is right in line with the main bearings and conrod shell bearings. I've ended up using rags, cotton buds, washing the oil cooler out etc, but it's still a risk. It may be a very good idea to order in new nipples as a precaution before starting this sort of work.

Right, time to go for a ride and see how it worked out. Shielding work is still on the list, I'll have a crack at that a bit later on.

[BMWST?]

i admire your tenacity and thoughtful approach sir.Always an interesting read .

[AllanB]

Nice job. Better place for the oil cooler as well - no crud from the front wheel and in a more direct flow.

[OddDuck]

Thanks guys... unfortunately although it looks like the oil cooler is helping with keeping the mixtures constant while riding, it's doing little or nothing for the vibration issue.

I'd realised, while checking the oil hoses for leaks, that I'd left the tail end of the shielding on the pickup coil leads open. It's been scratching against one of the oil line banjo bolts, giving an intermittent, high resistance connection at one end of the shield. This is a no-no because it sets up a loop in the shield itself. I'd done a quick and nasty job of insulating it, using a piece of flat rubber sheet and some cable ties (was in a hurry for an upcoming ride), but this made no difference.

I've been trying one thing after another to isolate what's going on. It's been successful in the sense of now I know a few things that aren't causing the problem. Tried so far:

Engine to frame bolts: torques checked out OK. Nope.

Grounding my new-ish steel battery box: Nope.

Pulling, stripping and checking the primary leads to the ignition coils against insulation failure and cross-talk: nope, even at 20+ years, the wiring's still OK. I re-covered that part of the loom with heatshrink tubes over individual wires (to get extra insulation) and re-tried, but no change.

[OddDuck]

Carrying on...

Carburettor balancing on the flatslides is normally done with a mechanical gauge (the shank of a small drill for example) but I also checked it with vacuum gauges, normally seen used on CV's. No significant difference between them to indicate anything wrong with carburettor balancing. It was interesting watching the needles waving around with the bike running, though.

I realised that the existing setup had the HT leads directly contacting the underside of the fuel tank - damage can be seen on the outer sheathing of one in the photos. It had been like that since I'd installed the replacement battery box. I had a play with the setup and realised that there wasn't an issue with installing the coils HT lead down, since I'd left a gap in the right place. This also shortened the HT leads. So - tried running the engine - no improvement.

I bought some new HT lead and tried fitting that. Still no joy. So it wasn't degradation of the existing leads.

A quick check of play in the main bearings - that's possible via the crank turning tool and the crankcase cover - but that wasn't it either.

I'm growing convinced that the issue is cross-talk in the ignition. The coils are talking to each other and I'm getting sparks on both cylinders simultaneously, or so I think. It's also slowly getting worse, despite everything else I've been doing. The timing lamp was tried again briefly (on the vertical cylinder) and would only synchronise at idle. At any faster engine speed, the lamp would drop out. It wasn't that bad when I'd tried it a week or so earlier.

While riding, engine behaviour isn't any better - I'd get occasional backfires through the induction, and the engine would drop out completely for a revolution or two when under sudden acceleration. The vibration, curiously, is mostly there but sometimes smooths out, particularly after the bike's been stopped for a while with the engine shut down.

So... last roll of the dice, really, after this I'm out of ideas for now... my current theory is that the CA Cycleworks coils have suffered some kind of gradual insulation failure. This is very early, these coils are only around three years old. Anyway, what I'm guessing is happening is that the high voltage secondary on coil A is leaking voltage, either to the coil mounting bracket or somehow is cross-talking within the existing primary wiring. The leakage voltage either then couples directly to the secondary on coil B, or it energises the primary which then fires the secondary... anyway, a spark on one couples across and fires the other, too.

I'd made a list of crank positions for a full 720 degree cycle. There are four spark plug firings - one intentional firing for each cylinder, and one wasted spark each as well. If I look at where they cross over, there's one occasion where a cross-fire would catch the horizontal cylinder during its compression stroke, and another time where a cross-fire would ignite mixture in the vertical cylinder during induction. These would be weak sparks and would be swiftly quenched, but still... there'd be a bump on one and a burp on the other, hence the vibration and the intake backfires.

For the record, none of this would be a problem on a 360 degree parallel twin. Leakage spark would be lost in the intentional spark, since they'd happen at the same time. That's if they use separate coils of course, it might be a double-ended system, I don't know. This cross spark business is a V-twin or triple problem.

I haven't yet tested. Maybe there's a voltage turning up on the coil mounting bracket, or a look at behaviour on the primary side of the system will show something. I'm not sure yet how to test - if there's a secondary voltage present it may destroy the multimeter.

I did some reading today about dielectric insulation breakdown and it came to these salient points:

- enough voltage applied and anything breaks down
- the effect gets more powerful with higher frequency
- impurities in an insulation material lower the breakdown voltage - like, say, moisture ingression
- once a path is established through solid insulation (electrical trees etc form), it stays. Solid insulation doesn't heal.

It may be possible, as a test, to remount the coils on insulating spacers and see if that changes things. If there is a leakage path through the mounting bracket then that will tell me something, at least.

[OddDuck]

Insulating spacers tried - see first two photos - possibly a very slight improvement but nothing conclusive.

Lots of chasing my own tail going on with this problem, lots of trying stuff and getting nowhere... it was only purely by chance that I saw the hole in the CA Cycleworks coil potting. There's been some kind of a blowout around one of the Low Tension (LT) positive terminals.

Without testing beyond seeing, it looks like the HT side of the coil winding has been shorting directly into the primary side of the ignition. That's 30 kV looking to roam. No wonder there have been problems.

Right, replace coils... while I was at it I thought I'd sort out the nasty oil leak that's been forming on the front underside of the engine. My thinking with this was that the larger O-ring (sealing an oil return galley between head and barrel) had been fried. Horizontal cylinder head would have been running hot due to pre-fire during compression. I changed this seal out and reassembled the cylinder, then refitted exhaust system, inlet, fairings etc to the bike. That's about fifteen hours work but there's no shortcuts here. For some reason Ducati put the oil pressure feeds on the outside, where it's easy to get at, but not so with the drains.

On closer examination of the photos after the job was done, it looks like a worse problem: oil's been leaking out of the cylinder head spigot. It would have been carried to the top of the bore by the piston rings. The gunk path is quite clear. There's not really anything I can do about it now except to take the motor apart again and have a go at resurfacing the spigot. Somehow. Ducati do specify that it has to have a perfect surface, which might not be easily done in a home garage.

The extensive blackening on piston and inside head is what happens after lots of cold starts. It (mostly) came off pretty easily via scrubbing with a rag, a dishbrush and a witches brew of one part isopropyl alcohol to two parts white spirits.

[OddDuck]

Right, and now a result. This may be valuable information for other owners, I'm sure that this problem is wider than just my bike or even Ducati's in general. I didn't want to discuss this before I was sure.

Ignition's fixed.

Replacing the coils (with stock Ducati items) has helped, however I have finally understood a problem with the bike that's been there since day one of my ownership and have found a work-around solution.

There's a high resistance back to the battery, through the stock wiring loom, from the tops of the ignition coils. This resistance is important because ideally the battery absorbs voltage spikes off the tops of the coils. If there's impedance on the line back to the battery, a spike on the top of one coil will start to be shared, voltage and current divider-wise, between battery and the coil that isn't supposed to fire.

I measured the impedance on my loom by fitting an 80 ohm automotive relay coil where the coils and CDI units were supposed to draw their power and then connect back to earth. The system was powered up. I then measured voltage across the 80 ohm coil, and also voltage across the loom, from the top of the coil back to the battery terminal. I did it this way because having the loom powered up closes relay contacts etc, it also loads contact points up with realistic amounts of current instead of the tiny sensing current of a multimeter.

A look at the Ducati wiring diagram told me that this path - battery to tops of ignition coils - goes through quite a journey. There are multiple crimps, lengths of wire, push-together connectors, the main relay contacts, fuses, and the kill switch. Ideally the voltage across the loom should be as close to zero as makes no difference. It wasn't. I got 0.58 V across the loom and 10.7 V across the coil. Working out the resistance via V=IR, I(loom) = I(coil) and R(coil) was 80 ohms, it turned out that I'm running 4.3 ohms or similar in the main wiring loom.

4.3 ohms.

It doesn't sound that significant does it?

Unfortunately, with ignition coil primary winding resistance being 4.5 ohms, that works out neatly to nearly 50% of any noise on the 12V supply line coupling straight into the non-firing ignition coil. The resistance in the main loom has formed a voltage and current divider.

That brings me to Paschen's Law, the relation between spark gap, pressure, and the voltage needed to make a spark:

https://en.wikipedia.org/wiki/Paschen%27s_law

In a nice world for us bikers, it'd be easier to make a spark under compression. Unfortunately things go the other way. It's about 22 kV to make spark in an air-fuel mix at full compression, but only about 2 kV in a normal one atm pressure mixture of fuel and air.

This would mean that it wouldn't take much electrical noise to get a spark happening at the wrong time. To be sure that there isn't a false spark, the rejection ratio of the path back to the battery would have to be improved to the point where any surviving noise presenting to the non-firing ignition coil cannot generate 2 kV or more at the HT end of things. Reducing stray current will follow suit, and sparks also need current for their energy. So the voltage divider effect has to be reduced as far as possible.

I did this by simply taking the main loom out of the equation and fitting the fore-mentioned automotive relay in its place. The loom's ignition supply now runs the relay's switching coil. The relay has a direct, much shorter, path to the battery, and supplies coils and CDI's directly. The AMP Superseal connector was available through RadioSpares, although auto electrical places with access to the Hella network would have it too. The wiring shown in the photo isn't quite complete, I finished the reworked loom off with spiral wrap to protect against abrasion. Spiral instead of heatshrink was chosen because it'll come off again reasonably easily and also it allows breakouts.

This isn't just a Ducati problem. The arrangement of wiring running from battery, through loom, to nose of bike, kill switch on handlebar, then back through loom to coils, is pretty much universal as far as I know. A high resistance through this arrangement will affect any bike with non-180 / 360 degree ignition timing.

That's all V-twins, triples, unequal parallels, big bang fours, etc etc. The only bikes fundamentally immune are 360-degree twins and inline fours, because a false spark happens at the same time as a wasted spark.

From what I've seen so far, warning signs that this is happening are:

- occasional backfires through the induction system
- white spark plug on one cylinder, persistently near-fouled plug on the other (identical fuelling)
- exhaust sooting occurring on one cylinder (the one with the near fouled plug)
- lots of really nasty exhaust noise
- rough running, hesitation or stumble on throttle opening

Tonight's test run, incorporating a turn on the motorway, showed massive improvement in noise control and running. This was the best that the ignition has ever worked.

Unfortunately the vibration problem, while greatly reduced, has not gone. It's about 25 to 50% of what it was, enough to be alright on hands (mostly) but painful on feet and toes. I think there's something mechanically wrong with the engine due to false sparking happening halfway through compression. So, halfway there. Mechanical stuff is a pain but at least I can see what's going on.

[Kickaha]

There's not really anything I can do about it now except to take the motor apart again and have a go at resurfacing the spigot. Somehow.

fine valve grinding paste and lap it on

[OddDuck]

Thanks Kickaha - I'll try that the next time I have the head and cylinder off the motor and on the bench. Head studs in the way, want to be sure everything's square, don't want lapping paste trapped between piston and bore etc etc. I'll have to live with the oil leak in the meantime.

More work today. Apologies, no photos this time.

I did something I should have done earlier and got the timing lamp onto the ignition, checking both cylinders.

Vertical strobed in cleanly and advanced smoothly to the maximum. The horizontal didn't. The timing lamp wouldn't light at all, at any RPM. After some swapping of CDI's, checking pickup coil voltage etc, I realised that I was running the old HT leads that had come with the CA Cycleworks coils. Unscrewing these showed nasty cracks in the silicon inner insulation, on the horizontal cylinder coil end, with one or two copper strands making direct contact with the HT lead's jacket. There were no marks of any kind on the outside of the lead.

Swap old HT lead for new (the ones that came with the Ducati OEM coils) and bingo. Timing lamp strobed in cleanly and stayed with the cylinder through to full advance.

Vibration was still there, though...

First things first, was a bearing disintegrating in the engine? Checking the oil mesh filter and magnetic drain plug showed some fine powder and a couple of tiny flakes, nothing significant. I don't have to get into the motor, at least not yet. This is good news.

The AFR gauge showed me a rich mixture on the vertical cylinder (by a couple of points) and slightly richer on the horizontal cylinder (a point or so more). The plugs are suddenly coming out of the engine a lot blacker than they used to be, so probably carburettor re-tuning is the next thing to attend to if I want to smooth the running out a bit more.

While I was at checking things, I had a look at compression, screwing in a manual car gauge using a threaded adaptor. Results were 90 psi H, 95 psi V.

The gauge is pretty clearly marked green - ok - and red - issues - with both the above results well inside the red zone. Before ripping anything apart I decided to run the numbers. What (theoretically) would the cylinders return as a compression test, given that I was using a threaded adaptor which had a pretty decent amount of dead volume?

Compression gauge adaptor: 2.6 cc
Cylinder bore: 92 mm
Cylinder stroke: 68 mm
Compression ratio: 9.2
Nominal capacity: 902 cc, 451 cc per cylinder.

Compression ratio is defined as: (volume swept plus volume of combustion) / volume of combustion.

This allowed me to work out that V (combustion) is around 55 cc. So the adaptor volume (although it'll affect things) probably isn't that significant.

Then I realised that the engine features late inlet valve closing - this valve closes 60 degrees after BDC. So on low RPM intakes and compression strokes, the piston's actually pretty far up the bore again before the intake valve closes. This reduces the swept volume and the compression ratio.

The conrod (on the i.e. models, I hope it's around the same on the earlier carburetted bikes) is 130 mm long. It's a reasonably simple geometry to sketch out. Trig time. After some calculations, it turned out that volume swept after that intake valve closes is 360 cc, not the nominal 451 cc.

This reduces CR to 7.55, at least during starting while gas flow is slow enough that cylinder over-filling etc won't be significant.

So... if 1 atm pressure is 14.7 psi, 14.7 psi x 7.55 CR = 110 psi.

This is still in the red zone on the gauge markings, with the green only completely taking over from the red at 120+. This 110 psi is a theoretically perfect compression result from a healthy cylinder at low enough RPM that gas flow effects don't come into things. If the 60 degress ABDC valve isn't taken into account, the expected compression pressure would be 14.7 psi x 9.2 CR = 135 psi.

This could be achieved, you'd just have to spin the motor pretty fast to do it. In the meantime I'm happy with the 90 to 95 psi measured. I still have the oil leak, but I know I'm not getting rough running from bad compression on one cylinder.

[OddDuck]

Back into it... I went for a test ride and finally decided that it's early failure of the main crank bearings. Vibration was linked directly to engine RPM. It didn't correspond to engine loading, tested by feathering the clutch but maintaining RPM while rolling. I will have to get back into the cases after all.

The bike uses an unusual main bearing scheme, at least compared to Japanese bikes: it's got roller element ball bearing mains, opposed taper, with preload onto the crankshaft via the cases. There isn't supposed to be any free axial or radial movement whatever.

Symptoms: vibration when running despite good fuelling and good ignition. Initially noticeable but OK, gets worse as the motor warms up. Gets better again if the bike is parked up briefly, rapidly goes back to what it was. It's steadily getting worse ride by ride too.

I think the variation once parked and cooled is due to differential thermal contraction, cases versus crankshaft. If the engine's been warmed up (ie with a warm crankshaft) but then is shut down, the cases will cool first. They'll contract a wee bit onto the crankshaft, restoring some of the preload, or at least taking up some of the open clearance.

Anyway, back to the big work... pull motor, split motor. To do this, the sequence is:

Fairings off
Exhaust system off
Carburettors off
Rider footpegs off, rear brake caliper too
All engine wiring loom disconnected
All hoses, wires etc cable tied to frame freed up
Starter motor cable disconnected
Rear shock removed
Battery and fuel tank removed
Engine stand plates fitted and bike positioned directly under lifting point.

Then frame bolts (just two) undone and frame lifted off and walked aside with one hand on a clip-on and the other on the rear frame rail. It would have helped greatly to have had two people for this, but I was able to do it alone, just.

The swingarm removal needs some snap ring pliers - circlip pliers just won't work here. Once done, it's very easy to pull the swingarm axle and bag the shims.

The lifting straps are just my standard ferry tie downs. The manufacturer points out very clearly that they aren't rated for lifting, but 4 x 250 kg straps used on an engine like this should be OK. Tilt (if needed) can be done by slackening straps while the engine is standing on the 2x4's at ground level. Taking time, checking everything prior to the actual lift, lifting in stages etc was done. The engine stand plates would be guillotines or shear blades in the event of a drop.

[OddDuck]

Confirmation - it's worn out main bearings, or at least, they've gone loose.

The engine supports the crankshaft at four points: the two mains, a bushing on the oil pump cover, and a ball bearing on the alternator cover. To quickly check play in the mains, I had to take both covers off. This could be done with the engine still in the frame.

It was possible to push the crankshaft by hand - I measured around 0.5mm to 0.6mm of free play. It was disconcerting to see how much the flywheel would move if I pushed it. It looks like the main drive gears provided some support on the oil pump side of the engine - the gear teeth still look OK, at least by eye.

As far as why this has happened:

Non-Ducati main bearings
Pre-ignition and increased pressure on crankshaft etc
Too much initial preload
Too little initial preload
Pre-ignition, overheated engine cases, and too little preload leading to rolling element skidding and premature wear of the races
Permanent flex and set of the engine cases
Cracking of the engine cases around the bearings
Cracking of the crankshaft
etc etc, maybe it's better to stop thinking now

I really don't know. Could be any of these reasons, could be any combination too. Only way to be sure is to separate the halves and check everything. The clutch is pictured because it had some oil coming out of that compartment, but that's a very minor issue for later.

[T.W.R]

Get some DPI to check the gears for stress cracks

[SVboy]

The saga continues......

[OddDuck]

SVBoy: Yeah, saga would be about the right word!! Perfect riding weather too.

TWR: thanks - have found this:

http://tradetools.co.nz/products/2845550

Unless you have another recommendation?