Winter Layup - 1995 Ducati 900 Supersport

[OddDuck]

Yep, back into the engine again. I'm not quite happy with how it's been running post rebuild, and guess what, I'd failed to examine or replace the conrod big end bearings.

Oops.

That's about thirty hours of work which I could have avoided if I'd just taken an extra night or two and gotten into the crankshaft / conrod assembly.

As to why it happened, it was late in the rebuild, I was getting fed up with doing this stuff as well as very tired, and my mate was onto me constantly about when we'd go riding again. I let these things make a decision for me. Engines need their details done right, and how I am doesn't really come into that. Next time, if in doubt, I'm taking the night / week / month off, then coming back to it.

The other issue is that post rebuild, and installation of nifty fuel hose disconnects plus inline secondary filter, the bike's gone from bad to worse with running lean under any kind of throttle and load. A quick run up the Wainui hill saw the AFR readings go from their usual awful 16 - 17-ish to clear off the gauge at 18+. It should be 14, plus / minus 1.

That's really bad news. Ultralean running will very rapidly trash engine components like valves, cylinder heads, barrels and pistons. I'm reasonably sure that the major reason the horizontal cylinder has a crack in its spigot is sustained lean running.

What I haven't been able to do, at least so far, is sort out what's going on and why with the bike's carburettion. It's always had issues running. I've tried almost everything to tune the carburettors and sort the ignition out. Valve timing and compression have been attended to as well, plus exhaust header sealing and inlet manifold vacuum leaks. None of it has sorted the issue out.

For a long time I was thinking I'd have to get a selection of slide needles and play with taper angles, then somehow I bumped across an idea I should have had a play with much earlier: the carburettor float height directly affects the fuel mixture, and is dependent on both baseline (static) height setting and also fuel supply pressure.

If the fuel pressure is changing, so is the float height. More pressure means more force needed to close the needle valve, so the float height will increase. Less pressure means the opposite, the floats will need less force to close and therefore will sit lower.

Carburettors are very sensitive to fuel level in the float bowls. It doesn't take much of a height change to affect the air-fuel mixture proportions. It's very difficult to see what's happening with float height while riding the bike, of course. This could have been going on for years and I'd never know it.

The behaviour that happens, again and again, is that I get the bike out of a 50 kph zone and into a 100 area. The AFR gauge ticks down from a healthy ratio to a lean ratio, over roughly 30 seconds or so. This happens no matter what I've been doing with main air jets, main fuel jets, needle position etc. About the only carburettion possibility I haven't tried is going to a more tapered needle, but these are hard to obtain.

I've been thinking that this has to be thermal, but it's also consistent with a float bowl draining to a low level and then holding there.

Then there's the very real possibility that with all the junk I've installed along the fuel lines, the fuel pump is having problems delivering enough fuel volume at any pressure to keep up with demand anyway... the brief run over the Wainui hill would seem to confirm this.

The lesson to all this is that if you want to tune your carburettors, the very first thing that should be done is to ensure that at all throttle settings, you've got consistent fuel line supply pressure (at the carburettors) and also enough supply volume at that pressure to keep up with fuel flowrate demands.

I'll hopefully be getting into this later. For now, it's sorting the engine. In a way it's kind of a good thing I went back in, I've already found a couple of mistakes with the base cylinder gaskets. I've used too much Loctite 510 on the horizontal, leading to smear gunking up and blocking an oil return line, and it looks like somehow I managed to completely forget to use any at all on the upper face of the vertical cylinder's base gasket.

There's discolouration over the conrods, too. I think the reddish colouration at the big ends is that nasty steel-on-steel contact at high pressure, similar to what I've been seeing on the chain sprocket splines on the ST2. The brownish colouration on the vertical piston's conrod, well, not sure. Maybe bad combustion from the ignition issues discussed earlier, but I really can't be sure.

[OddDuck]

Well, here's the crankshaft / conrod assembly, in bits... right, this was worth doing after all. It's been getting hammered.

Scoring is visible on the crankshaft main journal. It's centered at BDC on both pistons and running between 1/4 to 1/2 the circumference. It isn't that deep or over the full contact area, but it didn't look like this the last time I had this apart.

The big end bearing shells themselves are showing quite significant loss of bearing metal, with what looks like original turning marks in the substrate metal showing through the remaining bearing surface. The flanks of the shells are marked, showing that angles of contact have been changing, too.

Something I should have realised earlier was that loose main bearings, with off-center torque through the transmission which changes direction because it's one cylinder and then the other, will mean that the crankshaft is changing angle relative to the conrods. It'll be rattling a bit while the engine runs. It might very slight, but this means that a bearing surface which should be parallel will start being angled. This will stuff the big end bearings in very short order, since they're dependent on oil float to keep the bearing surfaces separated from physical contact.

This angling might explain the steel-on-steel contact between conrods, too. Some wear marks on the flat, parallel faces were visible, on close examination.

As a side note, I finally noticed that the conrods and end caps are stamped with numbers. If parts are disarrayed on a bench, it's possible to match them up again using this.

[pete376403]

Oil supply / flow / pressure / filter problems? Have you plastigauged the bearing clearances?

[OddDuck]

Oil supply / flow / pressure / filter problems? Have you plastigauged the bearing clearances?

No, no issues with the oil pump. I think this is a casualty of the main bearing failure earlier. The crankshaft was loose and side flex took out the half-bearings.

Plastigauge on order for reassembly work, hasn't arrived yet.

[OddDuck]

Cleaning up the crankshaft bearing journal was done relatively quickly with strips of 1000-grit wet-n-dry and CRC. I deliberately didn't go for a mirror finish.

The reason for that was a lot of chatter from American engine builders about surface finish, RA, and boundary layers in the oil film between crankshaft journal and bearing shell. Most of these guys talk about hand-finishing with 600 grit and leaving things there. Certainly after I was done the scoring was greatly improved, and if it's managed to run on the scoring without gross damage then this should work.

Fluid boundary layers are critical with journal bearings. There has to be fluid 'grab' on both sides, journal and shell, so that there's turbulent swirling between them and thus the journal effect can build up. Too finely finished might be too slippy and thus lose this boundary layer. Of course I could be completely wrong and find the crankshaft and shells get destroyed in short order once the motor fires up - aargh - but the crankshaft itself is now on borrowed time anyway... more about this below.

Regrinding crankshaft journals (what has to be done when things get really bad) turned out to be really involved. Not only is a weird-looking dedicated rig used (looks like a giant lathe, with an equally oversized beltsander) but the crankshaft itself might have to undergo post grinding heat treatment, including nitriding. Grind direction is important due to micro-somethings in the cast iron matrix leaving directional sharp edges. Fillet radii are critically important, if they're too sharp then the shaft can crack in use. And so on, there's a book's worth of similar considerations to take into account about this stuff.

In the meantime it's possible to get wrecker's parts with half or even quarter the mileage off Trademe or Ebay...

A mate told me to check the conrods against having gone out-of-round, with shells removed. I did this via transfer gauge and micrometer, finding no issues with either rod. I wasn't able to check parallelism, unfortunately. At this stage I'm not sure how the home worker could do this properly, since a pair of ultra precise diameter shafts are needed for the procedure discussed in the workshop manual.

I had a look at measuring journal clearance using micrometers as well. It turned out that there were a couple of non-Plastiguage techniques:

1) use micrometers, measure crank journal, bearing shell thickness, and conrod big end inner diameter.
2) use a transfer bore dial gauge, measure crank journal, zero the dial gauge on this reading and then transfer the gauge to the assembled conrod, thus reading the clearance directly.

There's a trick for using a drill bit shank to measure the bearing shell thickness, as photographed. I found that it had a distressing tendency to mark the bearing material and it's fiddly, no matter how careful I was.

Anyway, method 1 had problems with tolerance stacking. The plus / minuses rapidly add up to the point where the measurement is just too fuzzy to be trusted. Method 2 involves purchasing another pricey piece of gear which would get used about once every couple of years at most. It'd make sense if I was building engines all day every day, though.

As to the crankshaft itself being on borrowed time: I noticed, later during engine reassembly, that there seemed to be a bright line on the transmission side main bearing shoulder. A close look showed that it's been getting worn down, and measurement confirmed it.

The alternator side is running with a clearance of 10 microns between crankshaft and bearing inner race. On the transmission side, it's 30. This side of the crankshaft sees nearly all of the output torque and power, so it's highly loaded. The crankshaft has been rolling in ever-so-fine orbits inside the bearing's inner race and getting very slowly ground down.

The clearance is too fine for a shim. About the only fix is Loctite 641, and then it'll never come apart again, at least not without ruining that main bearing.

[OddDuck]

The idea behind Plastigauge is simple enough: it's a wax strip of a known cross sectional area. Squash it between two faces and although the width / height will change, the cross section won't. This means it's possible to infer the height (ie bearing clearance) from the width, using the handy scale printed on the fold of paper that the wax strip is sold in.

I got this off STA Parts, who delivered in about 36 hours from placement of order. Each 30 cm strip costs the princely sum of $4 or thereabouts.

I used this three times. First, with the worn bearing shells, both to learn how to use the stuff and to see what the old clearances now looked like. Second, with new Ducati bearing shells, in a red / blue combination, as I'd read recommended on another forum. Finally, with red / red bearing shells, as I'd built it last time.

Some notes from the experience...

Get some silicone oil onto metal surfaces as a releasing agent, prior to use. I had an old can of CRC 808 Silicone and this worked well enough. Engine oil didn't, and bare metal certainly ended up with waxy Plastigauge stuck fast. It took quite a bit of scraping with rags and fingernails to get it off the crankshaft journal, the first time around. There would have been no rescuing a nice new bearing shell if this had happened. The bearing metal's too soft to scrape it clean. I think I'd read somewhere that the wax will dissolve in engine oil, but I really wouldn't want this stuff interfering with a bearing's operation during startup and initial bedding-in.

The gauge itself is much more a yes / no type of measurement than anything giving clear numbers. It's possible to say that it's under X but over Y. That's about it. This is actually OK for this kind of work though. The numbers on the Plastigauge scale line up nicely with Ducati's quoted tolerance band limits. I'm not sure if that's luck or design but it's very handy.

It's important to be careful to not crush it any further while removing components, too. Obviously this would ruin the measurement, but it can be difficult to do. The conrod caps use guide pins and things can jam up.

The results were quite telling. Ducati's tolerance limits for the conrod bearing clearances are 0.024 to 0.051 mm, ie 24 to 51 microns.

Both of the old sets of bearing shells were outside spec, coming in at something between 51 and 76 microns.
The red-red shell combination was just on the limits of the spec, coming in at 51 microns or just over (or as near as I could tell).
The red-blue shell combination was a match, coming in somewhere midway between 25 and 51 microns. This confirms the advice given in the Ducati ST2 manual, by the way. I know it's a different bike but it does use a lot of the same components. There's a table of A and B crankshafts and conrods, with suggested bearing shell combinations for these.

Engine reassembly has started. I haven't worried too much about photos, having covered this earlier, but at time of writing progress is:

Crankshaft / conrod assembly done, with assembly lube and fresh conrod cap bolts
Case halves back together
Pump and transmission flank done
Alternator, starter and gearchange flank done
Vertical cylinder piston, barrel and head refitted

I'm quite keen to get this done in a reasonable time and then move on to sorting out the fuelling.

[OddDuck]

Carrying on... the engine's back together, as previous. Today was about lowering the engine from the stand and then carrying out major assembly of frame, engine and swingarm so that the bike's rolling stock again.

The photos say most of it. Previously I've replaced frame onto engine by hand - it's always been a pig of a job if working alone. On my own, it's been a desperate juggle. Cables always find a way to snag on something or get in the way while trying to fit frame to engine, then as soon as I try to fit a bolt, the handlebars will flop to one side or the other and then the frame will tip. Doing this job single-handed always ends up being about getting a bolt through and engaged, while trying to hold the frame steady with one hand. Somehow. I've managed it on the few occasions I've had to, but it has got pretty sweary at times.

Obviously with helpers this'd get a lot easier. However it can be difficult sometimes to get help... for anyone else out there just trying to get this job done, I'd found my tiedown set worked pretty well to lift, lower, and stabilise the bike. It was a real luxury being able to methodically clear cables etc as the fit was made. Keeping the bike steering locked at center using a couple more tie-downs also helped greatly.

The photo of the tie-down on its own shows a technique for doing controlled lowering. Just pressing the release button might let it explosively release. Looping the free strap back through the eyelet makes a rough block and pulley arrangement, with lots of friction controlled braking, and it becomes very easy to slowly let it down.

I'd also found earlier that it paid to fit the swingarm to the engine and torque the swingarm pinch bolts first, before major reassembly. The reason for this is that the pinch bolts can't be got at with a torque wrench while the swingarm is at a normal angle, part of the engine cases blocks access for a straight shot at the fastener.

Cue dramas with swingarm axle shims, while doing this. It seemed to have suddenly lost 0.4mm of width, until I realised that this particular assembly has to be done with the rear wheel fitted and torqued up. Pulling the rear leaves of the swingarm inwards as the axle is nipped up will open the front leaves up, and so properly running shim clearances have to be set in this condition. I left 0.205 to 0.276 mm clearance, as discussed earlier (the engine cases will expand once the engine is at normal operating temperature).

The big bits of the reassembly are done, there's a bit of fiddly messing about with cables and hoses, plus the exhaust system has to go back on, but that's it for things mechanical for now.

[OddDuck]

I've had the chance to take a look at Ducati's fuel supply arrangement to the carburettors.

The basic layout is:

Mesh filter and pickup at base of tank, fuel pump, proper fuel filter, metal piping inside tank to petcock, hose to Y joint. The Y joint then branches, with the center line going to the carburettors and the other leg of the Y returning to the fuel tank. The return piping runs to the highest and most central point in the tank before discharging into free air within the tank, above the fuel surface. The tank has a breather valve of course.

At first glance, it's a simple way of ensuring consistent fuel supply to the carburettors. Back pressure on the return line due to height should ensure consistent supply to the carburettors, even as the fuel level in the tank drops. The top legs of the Y are a high-speed flow circuit, with the carburettors sipping fuel from the center.

Unfortunately it appears that things aren't that simple... there's more than pressure due to height involved. There's back pressure due to flow resistance as well. The flow resistance is going to depend on flow velocity.

This flow velocity, even in the main supply / bypass line, will change. The carburettors will take only a little fuel at idle, but they'll want more as the throttles open. That change in carburettor demand will come out of the flow in the return line. That in turn will affect the back pressure in that return line from the Y joint onwards. That back pressure pushing fuel back to the tank is the same pressure supplying fuel to the carburettors.

As mentioned earlier, change in fuel pressure will affect the needle valves, the float height, the fuel level in the bowls and thus the fuel-air mixture strength.

I took an evening and made some measurements of internal tank piping, the Y joint itself, and fuel hoses - length and diameter for everything. I also checked the fuel pump's petrol flowrate per minute. Then I calculated (for a theoretical 900cc engine) expected fuel demands for perfect 14:1 operation.

Measurements:

Internal tank piping: roughly 5mm ID, 300 mm length on the return pipe
Fuel hose: 6.0 mm ID, 540 mm length on both Y branches
Fuel hose from Y to carburettors: 6.0 mm ID, 340 mm length
Y joint: 4.5mm ID all legs, branches 30 mm, center leg 34 mm length
H, height between outlet of discharge pipe and carburettor needle valves: roughly 300 mm
Fuel pump volume: roughly 1 litre in 38 seconds, flowrate affected by height of outlet
Carburettor demand: varies between 10 mL per minute to maybe as much as 300 mL per minute

So the first concern - that the fuel pump isn't keeping up with demand - is satisfied. The pump is delivering over 5 times the maximum expected flowrate, so that isn't a problem.

The second concern was that varying flowrate to the carburettors, via the 340 mm length of fuel hose, was causing enough varying pressure drop to affect supply at the needle valve seat. Similar to below, I calculated flow resistance at 300 mL / minute to be just 4 mm of height drop. There is an effect but it isn't more than 1% variation. A 6 mm ID fuel hose is plenty.

As a side note, a purely gravity fed carburettor arrangement will have similar behaviour to this. Flowrate in the line will cause a supply pressure drop, but it'll be minor compared to pressure head variations due to the tank emptying. On this bike, if it was purely a gravity fed arrangement via petcock at the base of the tank, I'd get roughly 40% variation. That's simply because the carburettors aren't all that far below the tank.

I then had a first shot at calculating back pressure in the return line due to flow resistance, using Reynold's numbers, tabulated values for petrol specific gravity and viscosity, and formulae for pipe flow resistance from an old mechanics of fluids textbook. It got complicated fast, unfortunately, but here's the gist of it:

Pressure due to height = fluid density x height x g (acceleration due to gravity)

Reynold's number Re = flow velocity x pipe diameter / kinematic viscosity

f, friction factor, depends on Re and tube roughness and can be found from experimentally derived charts

Height pressure required to drive flow = ( 4 x f x length of tube / diameter ) x (velocity^2 / 2g)

Any obstacle in the flow, like a pipe expansion / contraction, sharp edge, bend, abrupt edge etc will also introduce some sort of velocity squared flow resistance, but for now I've neglected these.

Anyway... what I worked out is that the pressure supplied to the carburettors via this arrangement won't be very high. Less than 1 psi, if I can trust my numbers. The other key finding is that small variations in flow velocity lead to large variations in back pressure, due to the velocity squared term.

I'd also found this page during the week:

http://members.iinet.net.au/~petercu...lorto_3_2.html

The key point I took from it was that the needle valve seat had to be matched to expected flow and expected supply pressure. I found I was running 2.0 mm seats, with 190 main jets; these seats may be too small for such a low supply pressure.

The other thing I found was that the variation in pressure due to variation in flowrates isn't trivial. Running the numbers showed that, even with as little as 300 mL being taken from a 1.58 L flow, a variation in back pressure of 18% occurs. This 18% variation includes the fixed contribution from height.

This is purely for flow resistance in the above lengths of 6 mm tubing and 5 mm piping. I haven't included any terms for the Y joint, the contraction area where tube goes over pipe, bends in the piping inside the tank, etc etc. These terms will make the variation worse, everything with a flow velocity squared term in it - like in-line filters, quick disconnects etc - will add to this variation.

I have no idea of the relationship between supply pressure and fuel-air ratio for my setup, but suspect it's complicated and non-linear. What I do know is that after installing the pictured hardware on the fuel lines, my fuelling problems became much worse. It worked badly with smooth lines, it has worked much worse with clutter in the way.

I think it's real. The fuel supply design is flawed.

What I'm not sure about is the fix, I'll have to take some time to think this through properly.

[pete376403]

Being a Ducati, of course they made a complicated solution to a simple problem. Why not a header tank (within the fuel tank, at the highest point, just a simple weir, nothing special. The pump supplies fuel to that and any surplus flows over the weir and back to the main part of the tank. The fuel petcock is supplied from the header tank and the carbs are supplied from the petcock. Constant pressure from the header regardless of the level in the main tank. Everything else, line sizes, float valve sizes etc are determined by maximum demand as you have calculated.

[HenryDorsetCase]

I have a stupid question.

On the crank there are two brass slothead screws. Are they something to do with the counterweights or are they blanking a drilling for (say) an oilway?

[OddDuck]

Being a Ducati, of course they made a complicated solution to a simple problem. Why not a header tank (within the fuel tank, at the highest point, just a simple weir, nothing special. The pump supplies fuel to that and any surplus flows over the weir and back to the main part of the tank. The fuel petcock is supplied from the header tank and the carbs are supplied from the petcock. Constant pressure from the header regardless of the level in the main tank. Everything else, line sizes, float valve sizes etc are determined by maximum demand as you have calculated.

Yep I like it. Should be possible to build into the existing tank setup without hacking / welding anything, I'll have to fabricate a couple of components though.

[OddDuck]

I have a stupid question.

On the crank there are two brass slothead screws. Are they something to do with the counterweights or are they blanking a drilling for (say) an oilway?

That isn't a stupid question at all. They're oilway blanks. I've never attempted to remove them, after finding that they were good and tight during the first engine stripdown. After all this time, chances are very high that they're firmly locked into place with engine ash and any attempt to remove will break the heads.

[Ocean1]

I have no idea of the relationship between supply pressure and fuel-air ratio for my setup, but suspect it's complicated and non-linear. What I do know is that after installing the pictured hardware on the fuel lines, my fuelling problems became much worse. It worked badly with smooth lines, it has worked much worse with clutter in the way.

I think it's real. The fuel supply design is flawed.

What I'm not sure about is the fix, I'll have to take some time to think this through properly.

I admire your dedication to the scientific method. If you look you'll find flow value data for commercially made valves and fittings using the unit Cv. (Or Kv for if you're picky). You can also find flow values expressed in terms of equivalent pipe length in diameter units, which makes for easy calculation.

Or, having discovered your system is too small you could simply double it's capacity by going from 1/4" to 3/8" tube and fittings.

[Ocean1]

Being a Ducati, of course they made a complicated solution to a simple problem. Why not a header tank (within the fuel tank, at the highest point, just a simple weir, nothing special. The pump supplies fuel to that and any surplus flows over the weir and back to the main part of the tank. The fuel petcock is supplied from the header tank and the carbs are supplied from the petcock. Constant pressure from the header regardless of the level in the main tank. Everything else, line sizes, float valve sizes etc are determined by maximum demand as you have calculated.

If the lines and fittings are correctly sized you don't need the header tank, if the lines and fittings are undersized a header tank won't help.

[BMWST?]

If the lines and fittings are correctly sized you don't need the header tank, if the lines and fittings are undersized a header tank won't help.

but wont a header tank remove the fuel supply pressure variations ? A header tank will always have say 200 mm of "head"...wont vary (much)