Marzocchi 40mm fork design and impact of oil level on spring rate, Part 3: The crap ones

.
It's amazing how a company with such a high end reputation for quality suspension, in many fields, can produce something as crap as these forks.  But, as those in business will understand, if someone asks you to make something, why not.  And clearly, when Ducati asked Marzocchi to make some upside down forks for the Monster and SS range in 1994, the word most emphasised was cheap.  Ie, "Yes, well, they're great, but do you have anything cheaper?"

So what we got were these 40mm forks with a damper cartridge arrangement that is fixed into the fork tubes by the bottom edge of the tube being rolled to fix them.  I've often poked quite a bit of fun at them, but it's not really Marzocchi's fault if it's what Ducati wanted.  As always, it's amazing the crap Ducati have managed to get away with over the years.

Functionally, the forks have normal style rebound damping, but compression damping in the final 1/3 (or so) of travel only.  This leads to quite a bit of initial dive, the sort of thing that catches you out when you grab a handful of front brake at low speed while working through traffic for example.

They were fitted to the Monster (M400, 600 and 750) and SS ranges (400, 600, 750 and 900SS CR) from 1995 until 1999.  I have record of two different dimension springs being fitted, but the overall rate calculates at much the same.  Being way too soft, and two stage to try to overcome it.

My set of these came from Minnie originally, before I fitted some 41 mm Showa non adjustable forks to her.  I did replace the springs at one point, but they've been unused for quite some years now.

I had a little time and curiosity, given my other fork playing, so decided to drag them out and have a looksee.

As I found them, they were fitted with 0.85 kg/mm rate springs fitted with 33 mm of preload and filled with oil to 100 mm.  Why they had 33 mm of preload I don't know, but it seems way too much.  So I tested them again with 17 mm of preload and the oil level set to 140 mm.  The preload tubes in this instance were pieces of 32 mm OD orange electrical conduit.  This is much the same size as the original tube, in both OD and thickness.  With a wall thickness of 3 mm, it does have a noticeable impact on oil level, especially when compared to the Showa thin wall steel tube preload spacers for example.

Then I dug up some original M600 springs from my stash of oldies.  I couldn't find any original preload tubes (long since cut up no doubt), so I cut a piece of orange conduit 120 mm long, which I'm pretty sure is the original length.  This gives 13 mm of preload on the original spring.  The std oil level is 90 mm, so I began with that, then added more preload (another 20 mm piece of orange goodness), then dropped the oil level to 130 mm with the same extra preload.  The orange conduit volume also had an impact here.  I calculated that the 20 mm long piece of conduit added the equivalent of 5 mm of oil height to the fork internal volume.  Which I'm mentioning as I didn't think of it when originally adding it, but later went "ah" when it appeared the air spring rate may have changed a touch.  As ever, it's all the errors you introduce without thinking that catch you out.

And so to a graph.  Blue is how I found it: 0.85 kg/mm spring with 35 mm of preload and oil set to 100 mm.  Black is with the preload reduced to 17 mm and oil dropped to 140 mm.  Red is with the original spring, preload and 90 mm oil level.  Green is with preload increased 20 mm and 90 mm oil level.  Orange is the same as the green, but with the oil level dropped 40 mm to 130 mm.

The orange and black curves show much more linearity, which is what I'm seeing a lot more of with oil levels reduced by 40 or so mm.  The shape of the yellow curve is also influenced by the spring rate, which changes from soft to harder at around 70 mm of compression.  With the preload set to 33 mm, this corresponds to 40 mm of fork compression, or just a bit more than the static sag would hopefully be.  I have found that the static sag load seems to be in the 45 kg range, conveniently occurring around 35-ish mm of compression for the black, green and yellow curves.

If I add the M900ie 'final' curve, with the 0.85 kg/mm springs running 5 mm of adjuster added preload and oil level set to 155 mm in purple, the orange and black look a bit light on at the top end.  Given the black and purple springs are both allegedly 0.85 kg/mm the difference in overall gradient is a bit odd.  I'd probably want to go back up in oil height by maybe 20 mm or so to increase the air spring.

Splitting them up for clarity, first a 20 mm increase in preload with the original spring.  The difference in static sag with this change should be in the region of 15 mm.

Then a 40 mm drop in oil level.  It's only in the last 1/3 of the travel that you'd notice any difference on the road.

The difference a heavier spring makes to the mid travel range is shown in the next graph.  Similar end points, but a difference in the middle of around 8 mm for a given load.  Black is 0.85 kg/mm, orange original.

Ducati Gold Coast special clearance sale

.
No affiliation, but Ducati Gold Coast are running a special clearance sale.  some of it is old DP accessories, including some Termi stuff that is now long gone for Monster and SS for instance.  Have a looksee here:

Ducati Gold Coast sale
.

Showa 43mm adjustable fork design and impact of oil level on spring rate, Part 2: 748 / 916 / 996 / etc

.
Today while I was doing a major service on a 748 I took the opportunity to throw a fork in the rig and see how it compared to the others.  I really wasn't sure what to expect, but was surprised anyway.  It's always nice to be surprised.

While the 748 forks (same as 916 and 996) are also 43 mm Showa adjustable, they differ internally to the M900ie forks by having the long rebound adjuster rod as discussed long ago in the 1000SS fork post.  All 916 on SBK have this style fork (except 749 Dark).  Assembly wise, there is a 100 mm long thin wall steel tube spacer and spring supporting washer below the spring and the usual Showa style preload tube above the spring.  I set the oil level with the lower spacer and washer fitted, as the manual doesn't specifically say to not fit them.  I'm not sure if that's right or not, but it's not mentioned and will result in a lower final oil level and I'm liking that these days.

The 748 forks have an oil spec of 132 mm.  I suspect this one was more like 142 mm before I pull them apart.  I couldn't get an accurate measure as the bottom plastic bush on the preload tube came off in the forks, making it pretty much impossible to get the spring out without tipping them upside down and shaking.  But the level with spring and bottom plastic bush in was about 10 mm lower before compared to after when set to 132 mm.

The spring in the 748 is a linear spring 285 mm long and the rate calculated at 1.00 kg/mm.  I didn't have time to measure it, but the calculation is generally very accurate.  So while it's a bit heavier than the others, it's fitted with only 8 mm of preload with the adjusters at the minimum setting.  Like the M900ie fork, the adjuster range is 15 mm.

To the curves: red is 748 with minimum preload, blue 748 with maximum preload, yellow M900ie as was originally with maximum preload and green is how I set the M900ie fork up at the end.  I don't feel so bad about my M900ie setting now.  Again, the coincidence of unrelated results is quite amazing.

Also again, it amazes me that you can hit (so near to) the same target with such variation in spring rate.  Although the rate variation in this case is a little less than the variation in the M400ie leg.  And the unknown is the variance in internal fork volume once assembled, as the different springs and spacers all contribute sometimes vastly different volumes.

This graph shows the collection of springs as encountered in these reports so far.  For this report, purple and green are the contributors.

One point that just occurred to me is that of fork swaps.  Often you'll read forum threads of where someone has fitted SBK forks to their Monster or other and the poster will be told of how the spring rates will be wrong.  I might even have said it myself.  But the above comparison of the 748 "as delivered" curve and what I decided was a good curve for a M900ie shows once again that, unless you actually check something, you really don't know whether you're talking through your arse or not.

Of course, all of this is separate to damping rates and their influence.  I'm not getting into damping rates at all.  That's completely out of my realm of experience, and it's an area where specialist experience is what matters.
.

Showa 43mm adjustable fork design and impact of oil level on spring rate, Part 1: M900ie

.
Or, keeping the same title so as to be consistent with the previous playing with forks.

So I got to play with a set of 43 mm Showa adjustable forks from a 2000 M900Sie today.  Well, I'm halfway through it.  I have serviced the forks in this bike once previously, and filled them with 5 weight oil to the spec height of 108 mm.

I thought to start with I'd do the same travel vs load graph (mm across the bottom, kg up the side) as for the Marzocchi forks, and then thought I'd add the original curve for the 43 mm Marzocchi with oil set to the spec height (105 mm?).  Turned out to be a bit spooky, frankly.

Red is the Showa with preload at maximum, yellow Showa with preload at minimum and blue Marzocchi.  Not really what I was expecting, and a wee bit eerie.  Obviously someone at Ducati thinks this is an ideal fork setting.  The difference between minimum and maximum preload is 15 mm, with this translating to 19 and 34 mm of total preload in the assembled forks.

More to come.  It's going to get some 0.85 kg/mm springs (rider 80 kg) and I'll drop the oil to 140 or so mm.

Ok, so on to the more.

I was very surprised by the fact that the air spring effect in the Showa was pretty much identical to the Marzocchi.  Looking at the parts, the Showa has a longer spring (349 mm) that sits at the bottom of the fork around the cartridge, with a long, but thin steel preload tube (277 mm) above it.  The Marzocchi has a shorter spring (249 mm) that sits on top of a stepped washer that itself sits on top of the cartridge, with a thick wall plastic preload tube (38 mm OD, 30 mm ID, about 100 mm long) sitting above (or maybe below, now that I think about it) the spring.  So perhaps it's just the visual weight of the shorter, but thicker Marzocchi preload tube that gives the impression.  Anyway, it was certainly not what I expected.

I replaced the original springs with a shorter, mostly linear rate spring (with a couple of tight coils at one end) which are meant to be 0.85 kg/mm.  I tested them on my amateur rig and came up with the curve below.  At 50 mm compression they average 0.85 kg/mm, but by 100 mm compression they're up to 0.90 kg/mm.  The OEM and ST series springs from the previous Marzocchi report are also included.  The two OEM springs are pretty close overall.  The Showa spring curve starts a bit higher, as it was a touch too long for how I had my rig set up, and also after midnight.  Dimension wise, the Showa spring has less coils (harder), but thinner wire (softer) and they're both 38.6 mm OD.

I fitted the 0.85 springs with 10 mm of preload (adjuster at minimum).   These springs are 282 mm long, requiring longer 335 mm preload tubes.  This is usual when replacing fork springs.  The only springs I know of that come at the same length as the originals are Ohlins, where they generally have a uniform length for a given application.  For the preload tubes I use 32mm aluminium tube with a 1.6 mm wall.  Unfortunately, the original steel tubes have a slightly larger ID, which means, when using this aluminium tube, I have to machine down the plastic end pieces Showa use to allow them to slide inside the aluminium.  The chuck ID on my little lathe is just a touch too small to allow the tubes to slide inside and be held at the machining end, so it's much less destructive to do the plastic pieces.

I set the oil level to 140 mm, thinking this would be a good starting point.  I did check the difference in oil quantity between 120 and 140 mm, so I could make changes based on quantity once the forks were all together.  This saves having to strip them all apart again.  At 120 mm oil height, the fork had 475 ml of oil in them.  Dropping the level to 140 took out 27 ml.  So if I wanted to go back to 120 mm I would just add 27 ml.  Or, conversely, if I wanted to go down to 160 mm, I could just suck out another 27 ml.

So, to the curve for 0.85 kg/mm springs with 140 mm oil height.  Not exactly what I was hoping for comparatively.  Red is OEM with max preload, yellow is OEM with minimum preload, blue 0.85 with max preload, green 0.85 with minimum preload.

I had figured that the 0.85 kg/mm springs would need about 5 mm of preload to give the right sag, which would place them about 1/3 of the way between the green and blue lines.  But the overall result is still a rather high rate of increase in effective spring rate near full compression.  Adding two further curves from the Marzocchi fork report, being the OEM spring with preload increased by 15 mm in orange and oil set to 145 mm and the ST series spring with oil set to 145 mm in purple, shows what I ended up with there.

So we come back to the question of how much load will the forks see?  Peter at Promecha told me that the maximum load (non impact if you like) the forks will see is if the bike is standing on its nose under hard braking with the rear wheel in the air.  And also that you don't spring them that hard, as it doesn't really work that way.  Which makes sense now that I've seen the impact of the air spring.

If we assume this M900Sie weighs 200 kg ready to go, and with the rider weighing 80 kg, plus 5 kg for gear, and taking off 5 kg to allow for the unsprung weight of the front wheel and fork lowers, that's 280 kg or 140 kg per leg.  These forks have around 115 mm of travel.  If we sprung it assuming the springs themselves supported all the load, 140 kg at 115 mm of travel gives a rate of 1.22 kg/mm.  But once we introduce the air spring effect, the required spring rate drops quite a lot.  With the 0.85 kg/mm springs, 145 mm oil level and my planned 5 mm of preload, that load will only give around 100 mm of compression.

How the bike was set up when it arrived, with the preload set to maximum on the OEM springs (the red curve), the same load would have given 95 mm of compression.

So it looks like I need to lose some more oil.  I'm quite disappointed with how this is turning out.  Not at all as I had expected.

Back again after losing some oil.  I sucked out 40 ml of oil, which should relate to 30 mm.  The change is somewhat as expected, as below: purple 140 mm, red 170 mm.  Both curves with minimum preload.

With my intention of running more preload, I made another calculated curve based on adding 5 mm of preload (1/3 of the maximum) and raising the oil level to 155 mm, 1/2 way between the two tested levels above.  The result wouldn't be exactly 1/2, as the change in air spring rate is slightly exponential.  I gave it 60% as a (random) guess, and came up with the curve below.  It's the yellow one.

I haven't ridden this one as yet, but given the chances of me exploring the end limits on a customer's bike during a suburban road test are non existent, I'll have to wait and see how the owner finds it.  The graph below shows how it arrived in red and how it is now in yellow.  There's not a lot of difference to show for my several hours of messing around, but the change in the last 1/3 of travel is what this has all been about anyway.  I would hope it's much more compliant, and less likely to make the big bumps feel big.  If that's not a poor way to describe it.

I got to thinking today that I had seen a fork oil level spec somewhere that always seemed pretty odd.  Turns out it was the 888.  The 888 has 41 mm Showa adjustable forks, much the same internally as the 43 mm forks fitted to this Monster.  The carburetor model 900SS used the same forks as the 888, albeit with the soft crappy dual rate springs in place of what I recall are much firmer, linear rate springs in the 888.  The oil level specified for the 888 is 162 mm and for the SS it's 108 mm, the same as this Monster.  I wonder why there was so much difference?

The 916 oil level is 135 mm, and the manual says "Oil quantity affects fork behaviour at full compression.  Compression load increases or decreases with oil level."  It also gives a minimum level of 150 mm and a maximum of 106 mm.

Which brings us to the end for now pretty much.  I'm curious to try some of the other forks I have kicking around, being the 40 mm Marzocchi and the 41 mm Showa non adjustable forks fitted to 90's Monsters.  Both of these had quite high oil levels specified, 90 mm for the Marzocchi and 80 mm for the Showa.

Plus there was a service bulletin for the M900 with the 41 mm Showa that recommended adding 30 ml of oil on top of the 80 mm spec.  This was aimed at helping the fork performance at full compression.  I'm curious to run them on my test rig to see how they behave, as it just seems like way too much oil.

More rambling to come.  Lucky you lot eh.
.

Marzocchi 43mm fork design and impact of oil level on spring rate, Part 3: More fork details

.
Just a short update.  I had an S2R800 in today for service and had the forks apart.  While they looked identical to the Marzocchi ones, I noticed that the markings in the inside of the lowers were different.  On closer comparison, it's the brand part that is different, everything else is identical.  I have no idea who makes them, but it is a symbol I see on a lot of these late model forks.  Photos below show them, Marzocchi with 'M' cast logo on left.

I did notice that the RH leg, which is the one I had played with previously, had rebound damping and no compression.  But the LH leg had compression damping and no rebound.  A little for most of the travel, a lot more in the last 30mm or so.  So as a combination they provide both, and that way you can tailor them as desired individually with different oil weights for example.

I added 15mm more preload, using pieces cut from an original spacer.  I'm not sure 15mm is enough now that I think about it, given it had nearly 55mm of sag before we started.  But we'll see.  I also dropped the oil level to 135mm.
.

Some photos for Sarge

Nothing exciting, just some crankcase photos.

Lift graph for 4V Desmoquattro cams using strada inlet cam as exhaust cam

I thought I'd post a graph here for a conversation that I've been having with someone about using the 4V Desmoquattro strada inlet cam as an exhaust cam.  I haven't tried this personally, but have been told it works well.

In the graph, you can see the exhaust cams on the LH side.  The yellow curve is the G, the corse exhaust cam used from 1992 (888 SPS had it too) until the last of them in 2000.  They obviously liked it.  The 916/999SPS exhaust cam (blue), while having much less duration, has a fairly similar shape and compared to the pink strada exhaust curve, is completely different.  The strada is much slower opening, with what is called an asymmetric profile.  The point of this graph, the red line, is the strada inlet cam being used as an exhaust.  Profile wise it is very similar to the SPS up to peak lift, although it doesn't hold the peak or closing profile as long.  But much better than the strada exhaust, and the SPS cam set does work very well.  It's not a straight fit, the closing lobe needs to be narrowed and the opening rockers need some grinding at the feet.  But well worth the effort given the cost and complete unavailability of the SPS cams.

How much input the SPS inlet has over the strada inlet in comparison I don't know.  The extra high lift duration of the SPS inlet is something the specs don't show on their own, and that sort of extra area under the graph can only be good.  The orange curve is the 748RS '506' cam, shown because there's some on Ebay at the moment and duration wise they only have 5 degrees more than the SPS and 11 degrees more than the strada.  They do have a lot more lift, and are quite aggressive acceleration wise.  But if you kept them to around 10,000 rpm, not the 13,000 or so they would have seen trying rather vainly to keep the Japanese 600's in sight, they won't be too destructive I'd think.

Doug Lofgren's cam page has more info on them too, with cam doctor graphs.

Timing belt pulley positioning on belt drive Ducatis

There's a thread going on one of the forums about replacing belts, and the point of marking the belts has been raised.  I made the comment that I found the marking of belts an over complication of the procedure, but some people do find some comfort in it.  In particular for locating the vertical camshaft correctly (vertical inlet on a 4V), due to the load the vertical inlet closing spring places on the camshaft when the cam is at the correct position.  Coupled to the fact that on most of the 2V models the vertical cam pulley has a surround covering it and making it hard to get a good hold of it.  I can understand the issues this brings to those who do this very infrequently.

So, to confuse people further, I thought I'd introduce another method to bypass this.  I did teach this method to a previous workmate, who them went and bent some valves, so if you don't understand it, don't use it.  But it seems pretty simple to me.

The timing shaft pullies traditionally have 4 holes in them, and the locating dot on the outer edge.  See the picture below:

In this instance, the locating dot is lined up with the mark on the primary drive cover.  This denotes the engine is at TDC firing on the horizontal cylinder, which is the master position for most setting procedures.

As the timing shaft turns at half crank speed, it rotates 180 degrees for one crankshaft rotation.  From the RH side the crankshaft rotates clockwise and the timing shaft rotates counter clockwise.  Vertical TDC firing is 270 crankshaft degrees after horizontal TDC firing, or 135 timing shaft degrees.  If you look at the above, you can see 4 holes spaced at 90 degrees, and conveniently located at 45 degree spacing either side of the locating dot.  This means that the second hole clockwise from the locating dot is 135 timing degrees from horizontal TDC firing, and as such, it's a locating dot for vertical TDC firing when aligned with any of the aligning marks on the primary drive cover or camshaft surrounds.  As below:

On the engines with adjustable cam pullies, these holes aren't present.  But of the three locking screws that hold the halves together, one is in the correct position.

The locating dot still gives the position anyway.  On the 2V pre DS motors, when the timing shaft dot is at 3 o'clock, the vertical camshaft dot will be at 4:30.

On the later 4V Testastretta and 2V DS style engines with the 20 tooth pullies, the small holes are still there.  But there are many larger holes, just to confuse the issue.

And, as always, when done, the locating dots on the timing shaft and all cam pullies must align with all the correct marks once set back to horizontal TDC firing.

Fitting a 2003 onwards 696/796/800 6 speed gearbox into a pre 1998 600/750

One of the motors I have in bits is a 600SS motor, which came from the bike I sold the original engine from Minnie into.  I actually sold more of this engine last week, just to make it harder for me to pointlessly waste a heap of money rebuilding it.  This was mostly due to a mock up of the bits I had done the week before, trying the 68 mm stroke 900SS crank (a narrow one) to see what happened.

There was contact in a couple of places between the crank webs and the cases, at the inside of the starter drive idler gear boss in the LH case, and in a similar position on the RH, with a casting ridge that I'm guessing would be for an oil gallery.  So I can't really cut either out, meaning the crank would have to be relieved at the outer edge, which reduces the most significant section of balance weight (furthest from the centreline).

The same section of crank web also hit the timing shaft on the RH side, where the later model timing shafts don't have a locating ridge.  The pre 98 small blocks have a timing shaft that is shimmed between the cases to locate it, unlike the big block or 98 on small blocks where the shaft is held against the LH bearing by the timing gear.

The bottom of the 600 cylinder did just clear the web, with the underside of the piston quite a distance away.  And the rods cleared the timing shaft.

And the 620 pistons I was going to use (it was going to have 750 heads, similar chamber to 620 and much larger than the small chamber in the 600 heads, which I had sold anyway) stick out a bit, but a thick base gasket shaped spacer would fix that.  The 6.5 mm stroke difference between 620 and 900 (61.5 mm to 68 mm) would suggest the spacer would need to be 3.25 mm thick.

Anyway, I digress.  The point of this was trying to put a late model small block 6 speed into pre 98 cases.  The complete 6 speed box I bought from Ebay was from an M696.  The easiest way to show why it wont drop in is with some photos.

Input shaft:
The input shaft has a larger nut on the clutch end, like all other late model boxes.  So you'd need a bigger nut and possibly to do something to the clutch hub to accommodate it.  The length of the clutch snout looks to be the same from threaded end to first gear.  I didn't notice any difference there, but I didn't measure them.  I might do that.  6 speed at the top.

The LH end of the input shaft has an extension that goes through into the recess which the clutch pushrod passes through.  As the pre 98 doesn't have the clutch pushrod passing from left to right, there's no need for the extra snout, and no room in the LH case for it anyway.  It has to be removed, and I figured the drop saw would take care of that.

Output shaft:
Not a great deal of difference here.  The shaft outside of the LH bearing is 10 mm longer on the 6 speed, but you'd just turn the sprocket around to fix that.  The only other thing I noticed is that, while both shafts are 25 mm diameter at the sprocket spline, the 6 speed wouldn't slide into my 600 cases like the 5 speed did.  Maybe it just didn't want to be there.

Shift drum:
The shift drum is the biggest issue, with quite significant differences.

The RH bearing area is much smaller diameter on the 6 speed (1 arrowed).

The RH end of the 6 speed drum has the 6 groove snout (2) for the sprung detente lever the later bikes use.  You can see at the bottom left of the snout that one ridge is missing its top, that's neutral.  The sprung detente lever at the RH end of the shift drum was introduced with the 916 from memory, and every new model since has adopted it.  The sprung ball detente at the LH end of the shift drum is now only for neutral, so there is only one corresponding groove in the LH end. 

The neutral light activation has gone from a 2 wire switch that is pushed in by the lump on the 5 speed drum (3) to a single wire switch that is earthed by the drum itself.  The light brown covering on the drum, with the space at the (3), is an insulating coating that keeps the earth circuit open at all times except for when the drum is in the neutral position.

The difference in the shift fork groove pattern is obvious, so clearly you can't use a 5 speed drum.

I had various thoughts for overcoming these issues, from cutting and shutting the end of the 5 speed drum onto the 6 speed drum to making a spacer ring and welding on a switch pushing lump and cutting off the detent snout with the drop saw.  Attempting the shut once the cuts were made is the only way you would be able to determine if that was a viable option.  If not, you have two scrap shift drums.

But you still have the issue of the LH end to deal with.  The LH end of the 5 speed shift drum shows the old style detente grooves.  You would have to machine these into the 6 speed shift drum to use it, which for me starts to make the whole thing much more of a hassle as I don't have a mill.  One option around this is to use a non RH snout detente style shift drum from a pre 916 big block as a starting point, ie, 851, 888, 900.  I'm sure I have one of these around somewhere.  I did compare a 900ie shift drum, which does have the RH snout detente, and the grooves appeared to be functionally the same.  They weren't as smooth flowing, with a step at each transition point, but all appeared to do the same thing at the same setting.

So I didn't get any further with this fitment.

The other issue I have with the concept is that while a 6 speed is nice, the niceness comes from the ratio spread.  The 5 speed ratio spread is 2.59, the 6 speed is 2.67.  The big block wide ratio 6 speed is 2.88, a comparatively much bigger change.  To overcome this there is the possibility of having another gear made for 6th, either 1 tooth bigger on the input or 1 tooth smaller on the output.  The 26/24 ratio gives an 8% gap between 5th and 6th.  27/24 gives an 11% gap, 26/23 12%, with a ratio spread of 2.78.  That makes about a 3 tooth on the rear difference between the top gear ratio of the 5 and 6 speeds.  The shape of the current teeth determines if the 1 tooth idea will fly.  The input shaft gear will be the easiest to make, as the output shaft gear has a shift fork groove on it.
.

Valve train article from Kevin Cameron

This atricle by Kevin Cameron is the second in a collection of 3 regarding valve trains.  Some of the comments regarding Desmo operation and its limitations may come as a surprise to some.

Facebook Page!

.
I've started a Facebook page for Brad The Bike Boy.  My main plan is to use it to get to the local customer base in a short time frame.  We'll see how it works.

Of course it could just be another thing I don't get round to spending as much time on as I should.
.

996R and 998 (and sort of 748R) Non Linear TPS Baseline Adjustment

Summary: How to set up throttle bodies and idle mixture on the 996R and 998.  And 748R too, which are somewhat the same.

Photos with thanks from Phillip at Two Wheelers in Griffith.

The 996R and 998 use a unique set of throttle bodies (although the 748R is similar) with shower injectors.  The photos below show the features of these throttle bodies.

In the photos you’ll notice I’ve numbered all the parts we need to know about in series across all the photos.  That way, if I give a number, you know which part I mean.

Items viewed from the LH side on the horizontal throttle body.  Ducati call it the master.

  

1/ TPS

2/ TPS adjustment screws (only one visible), usually T20 torx.

The TPS (1) is the small PF3C Non Linear part used on the 2V models from ST2 onwards in addition to the 748R, 996R and 998.

Items viewed from the RH side on the horizontal throttle body.

  

4/ Throttle cable wheel

5/ Throttle cable mount

6/ Throttle linkage rod

7/ Horizontal cylinder air bleed (obscured by wiring)

8/ Horizontal cylinder idle stop screw

Items viewed from the RH side on the vertical throttle body.  Ducati call it the slave.

6/ Throttle linkage rod

9/ Vertical cylinder idle stop screw

10/ Vertical cylinder air bleed (behind throttle arm)

11/ Throttle linkage adjustment screw (accessed from below)

12/ Fast idle arm

13/ Fast idle adjustment lever screws

14/ Fast idle cable mount and adjuster

Although it is hard to see in the above photos, the throttle body balance adjustment set up is a hex head screw (7mm) with a flat blade slot accessed from underneath.  You can see the end of the screw numbered.

748R: The 748R doesn’t have # 11 to 14.  The throttle linkage adjustment is done by loosening the little nuts and rotating the rod, which has a left hand thread at one end and a right hand thread at the other.  It’s as frustrating as it sounds.  Fast idle is done with the little button at the twist grip, like all the other 748 – 996 models.

Fast idle adjustment for 996R and 998

The fast idle actuating lever is obscured in these photos, but if you look directly from the side you can see the ends of the slot in the fast idle lever and the boss that stops it.  This means the ends of fast idle lever (12) travel are constrained, and this needs to be remembered when you adjust the cable (14).  There’s no point adjusting the cable so the lever doesn’t move, as the adjustment for when the fast idle arm starts opening the throttle bodies themselves is controlled by the fast idle lever (13) adjustment, which is done by undoing the two screws (13) and sliding the assembly forward or back, depending on whether you want more or less.  As such, adjusting the fast idle lever (13) can be a little tricky, and realistically there’s not much point, as below.

On the 996R they went to a fast idle cable, but this cable acts on the vertical throttle body.  The throttle cable acts on the horizontal, so when you open the throttle cable you open the front throttle, then take up the free play in the throttle rod, then open the vertical throttle.  So when you set the running balance (sync for the Americans) by holding the throttle open, all the free play is allowed for.

But when you pull the fast idle lever on, the above action is reversed and so the free play is introduced as an imbalance.  The means the vertical throttle is open more than the horizontal.  As the TPS is on the horizontal throttle body, the vertical is open more than the ecu expects and so will run leaner than the horizontal.  All of which is not conducive to reliable cold starting.  It’s a completely stupid design, and it’s why most 996R and 998 will start from cold better without any fast idle and with the throttle closed or opened a little at the grip.

To set the “Non Linear” TPS

Preparation: you need to get to the top of the throttle bodies to wind out the stop screws and disconnect the linkage rod ends, so you really can’t do this unless the air box is removed.  Removing the air box on these involves removing the trumpets with the injectors and fuel hoses built into them.  Be very careful not to crimp the plastic fuel lines when twisting the trumpets.  Remove the wiring to the injectors first, and don’t pull to hard on the connectors, as you can pull the plastic centre out of the steel body.  When you lift the air box out you need to feed the wiring and it’s grommet down through the hole in the base.  You can leave the fuel lines in the air box base, but one of them usually gets hooked under the vertical throttle body tabs.  So when you’re pulling upwards on the air box and it just won’t come out, remember this.  And the drain hose at the front LH of the base.

TPS base line setting procedure (as I do it)

Every step is as important as any other, and the predominant aim of this procedure (apart from being “correct”) is repeatability and consistency.  I normally do it with the engine and throttle body cold.  This is simply the convention, I do it hot if I have to.

Disconnect the throttle cable at the throttle cable wheel (4).  Easiest to just take the cable out of the wheel rather than play with the adjuster.

Disconnect the throttle linkage rod (6).  You can pop off the plastic connector at either end.  I use a pair of side cutter pliers with the taper side against the throttle cable wheel side – it pushes the plastic connector off nicely.  Disconnecting the throttle linkage rod (6) removes the dumb arse fast idle system from having any impact.

Connect to the TPS by whichever method you are using to measure the TPS output voltage – back probing wires, Mathesis, DDS, etc.  I think you need to go into the outer two wires if back probing.  You'll either get 5V, nearly 5V or 350 - 450mV.  Given the connectors fitted to the 5.9M ecu, back probing at the ecu can only lead to much agro.

On the 748R you can back probe the 1.6M ecu quite easily.  See the MV TPS for the procedure.

With the throttle closed before you start you should be seeing voltage in the range of 350 – 450 mV, depending on who’s been there before you.  I usually record this voltage, just in case I need to go back to it.  Remember, you need to have the ignition on when doing this, and you don’t want to short out any wires.  Disconnecting the headlight or removing the headlight fuse can also be a good idea if you don’t have a "lights on/off” switch.

Back off the horizontal cylinder idle stop screw (8) to allow the throttle butterfly to close fully and jam if snapped or pushed.  These screws are 8mm hex.

Open and close the horizontal throttle butterfly gently until the voltage stabilises at a minimum.  Do not snap the throttle butterfly shut, as this will jam the butterfly and may give a false lower voltage reading.  Usually the voltage at this point will be in the range of 90 to 170mV, most likely around 100.  Again, I usually record this voltage out of suspicion.

Once you’ve ensured the horizontal throttle butterfly is closing properly, reset the TPS to give an output voltage of 150 mV by loosening the two screws (2) and rotating the TPS as required.  These screws are usually T20 torx head.  Tighten the two screws (2) and ensure the setting remains consistent.

Wind in the horizontal throttle body stop screw (8) until the TPS output voltage is around 390mV.  The specified idle setting for the 748R is 2.2 degrees and for the 996R and 998 is 2.3 degrees, which theoretically equates to 380mV and 390mV respectively.  I always use the throttle angle the ECU is reading to set the “base idle” setting, as what the ECU is reading is more relevant than the actual voltage, as there is always variation in the relationship between them.  However, you need some sort of diagnostic tool or software to read the throttle angle, so for many the voltage is as close as you can get.  It’s still quite valid.

Open and close the throttle a few times to check the idle setting is stable and adjust if required.  The TPS is now set and the horizontal throttle body stop screw (8) should not be touched again.

Reconnect the throttle linkage rod (6).  Remove the vertical cylinder idle stop screw (9) and turf it.  This way, the throttle linkage rod (6) is loaded by the vertical throttle butterfly all the time and free play is not an issue.

Reconnect the throttle cable to the cable wheel (4) and set the free play in the cable.  You can do this at the throttle body (5) or the twist grip, up to you.  On the 748R the free play sets the fast idle speed with the button pressed.  Turn the bars fully left and right while doing this just in case the cables move when doing that. 

Refit the air box and fuel tank.  You can get to all the adjusters you need to from the RH side, though they can be a bit fiddly.  Start the engine and get some heat into it.

Wind the air bleeds (7, 10) fully in.  Again, I record how far out they were, just in case I need to go back to where I started.  You will need the fast idle on to get the engine to idle, but remember it will be unbalanced.  Set the running balance (synchronisation) via the throttle linkage adjustment screw (11) using mercury sticks, etc at various RPM.  But remember to raise the rpm by holding the twist grip open, not using the fast idle lever.  This is often a case of best compromise, and remember that cruise at 4,000 RPM has more throttle than free rev at 4,000 RPM.  I usually go up to 6,000 or so RPM free revving to check for variation.  It’s not uncommon to see one cylinder lead then the other lead at differing RPM.  Remember the word “compromise”.

Next set the idle speed using the air bleeds (7, 10) as required.  Set the idle speed at approx 1100 – 1150 RPM once the engine is hot.  There is no general rule for how far out the air bleeds should be, although I find the 748R will be nearly closed.  At this point most will set the air bleeds so the cylinders are balanced at idle.  I don’t worry about vacuum balance at idle though – we balance the mixture using the air bleeds and idle trimmer, as set out below.

That’s the end of the throttle body fiddling.  Next is the setting of the idle mixture.  I often say to people that if they can’t carry out this final step then all the previous steps may have just been in vain.  You might get lucky and be close.  Or not lucky and still have a poorly idling or running bike.  The idle mixture can have a very large impact on general running at low throttle openings.

For example, if the bike is too lean and idles low, winding out the air bleeds (7, 10) to try to raise the idle speed will only lean it out more, so the idle will most likely not increase.  You can sometimes make a bike idle better by winding the air bleeds (7, 10) in and richening the idle mixture if it was too lean to start with, even though the idle is low.  As an indication based on experience, a full turn on the air bleeds can change the mixture from 1% CO to 4% CO, so the air bleeds have a big affect.

To adjust the idle mixture you need to get into the idle trimmer

I use 4% CO as the idle mixture target for the 748R and 5% for the 996R and 998.  This tends to work well and give good low speed running and fuel economy.  It's a bit leaner on the 748R to try to stop the off idle hiccup that they often have with Ducati based eproms, although because the problem isn't really fueling related (it's more ignition advance) it's a bit of a badly applied band aid.  

Take a sample from both header pipes using the little 1/8 gas threaded bungs.  You can buy from Ducati a tube to go into this bung – it’s a piece of copper pipe with the appropriate fitting on the end – that is surprisingly cheap.  You could also get some made up by a local hydraulics supplier.  I have a few of varying lengths and bends for all occasions.

Given there is only one idle trimmer for both cylinders, I check the mixture and then adjust the trimmer to get the average between both cylinders where we want it.  Then I adjust the air bleeds to give the same mixture in both headers.  This means that the manifold vacuum balance or synchronisation at idle will often not be equal.  That’s just how it is.  Equal mixture CO% is more important in my experience.

748R with 1.6M ECU

The idle trimmer is a trim pot inside the 1.6M ECU, which is under the seat.  The 1.6M ECU has a rubber access plug that is hopefully hidden under a “do not remove” sticker of some sort or some race tape if someone has been in there before.  Once you’ve finished, cover the plug again – they can leak water into the ECU if you leave the plug uncovered.  Either remove the whole sticker, tape, etc or just use a sharp knife/blade to cut around the plug and gently pry it out with a little screwdriver.  Once removed you can see the eprom directly under the access hole.

The idle trimmer is a little metal square just next to the eprom with a plastic rotating pot inside.  This adjusts the idle mixture, working the same as the electronic idle trimmers in the 1.5M and 5.9M ECU.  It adds or subtracts a given pulse width from the map fuel number across the whole map, therefore having a much greater percentage effect at small pulse widths.

This trimmer is adjusted by rotating it between the ends of its travel.  The total travel is 270 degrees (3/4 of a full turn) as shown in the next photo so if you manage to make it go all the way round you’ve wrecked it and it’s time for a new ECU.  Be very gentle!  As with a mixture screw, clockwise is lean, anti clockwise is rich.  The mid point, where the trimmer slot points directly at the eprom socket, is nominally zero.  Best to use a non metallic screwdriver when adjusting this especially when you have the engine running and the ECU circuit board is live.  You don’t want to be shorting anything out.

In the photo you can see the trimmer is set at about 60 degrees lean (which looks identical to 120 degrees rich, unless you look really hard and can see the little dots that act as arrow head edges).

If you don’t have a gas analyser, as a bit of a rough guide, I’ve found that the best way to set the idle mixture is to move the trim pot slowly clockwise (leaner) until the revs begin to drop.  Then go richer until the revs begin to drop.  Now you have these two end points, set the trimmer 1/3 of the way from the lean stumble point towards the rich stumble point.  This usually gives 4% CO when I’ve tried this method and then checked with the gas analyser.

There is no default or ideal setting for the trim pot (except for the ones I specify as guides for my eproms), just wherever it ends up.  You can manipulate it somewhat with air bleed position if you need to, and it can have quite an impact on low speed running and fuel economy.  And it does act over the whole fuel map, so can affect WOT running too.

996R and 998 with 5.9M ECU

The idle trimmer on these models is adjusted electronically via diagnostic tools or software.  There is no way to get around it.  Mathesis, DDS, Technoresearch Centurion, Navigator, etc.

There is no default or ideal setting for the idle trimmer, just wherever it ends up.  You can manipulate it somewhat with air bleed position if you need to, and it can have quite an impact on low speed running and fuel economy.  And it does act over the whole fuel map, so can affect WOT running too.

.

Marzocchi 43mm fork design and impact of oil level on spring rate, Part 2

.
I tested the theoretical 15 mm more preload result today, and it was pretty good.  Green is my theoretical prediction, and red is the measured.

Compared to the ST 0.83 kg/mm spring and 125 mm oil level, the end result is fairly similar.  It would be interesting to see how this rode, and what sort of rider it would suit.  Overall, the difference between the heavier ST spring and 125 mm oil level and the original spring with more preload is fairly minor.  Minor enough to make me re-evaluate my believe that you need to replace the springs.  I have always been wary of just adding preload, but maybe that view was wrong.  The linear spring certainly gives a more linear result from no load to bottomed, but the linearity is much more dependant on oil level than I would ever have expected prior to having this play.  Yellow is the ST spring, red the Monster spring.

Next I took the oil level thing to another extreme, just to see what would happen.  With the spring preloaded an extra 15 mm as above, I dropped the oil level to 215 mm.  As you can see, it's moving closer to a linear rate overall, but still the rate at full compression is around double that at the beginning.

So to give a visual summary of where I started and where I sort of ended up in an untested guess, red is the start, green the end.  I still don't get the desire for such an aggressive progression on effective (assembled) spring rate.  Maybe if the initial travel wasn't so soft (giving so much sag) I'd see more sense in it.

.