Saturday, December 28, 2013

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

Disclaimer: I'm not a suspension expert, just curious.  If any experts would like to comment on my rambling, I'd be more than happy.  Plus I keep updating the text, now into its third day of writing and editing based on some more info from Rick.  Also with Rick's replies I have realised that this leg is (possibly, maybe it never had it) missing the inner spring guide, which is a plastic slotted sleeve that goes over the damper rod above the cartridge to help support the spring from the inside.  See the circled part in the parts catalogue diagram below.  Also note that this diagram is shown for all the 43mm non adjustable Monster forks from 03 - 07 that I looked at, and is possibly a Showa from memory with the spring located slotted collet.

There was a post on the Monster Forum about the Marzocchi 43mm non adjustable forks fitted to a M800ie.  I have an old M400ie fork leg at work, one with nasty external damage that I expect came from rubbing against something in the container on the trip over from Japan.  So I figured I'd strip it and have a look at the valving system.

Typical of the springs Ducati fit to many of the non SBK models, it is a two stage style.  I wouldn't call it progressive, as that implies something some see as desirable and useful.  Don't see either myself.

In terms of spring rate, you get an overly soft spring until all the tight coils have bound, then you get a fairly sudden change to a much harder rate.  Spring rate calculations are based on the wire diameter (bigger = stiffer), coil diameter (bigger = softer) and number of coils (more = softer).  So when the tight coils bind, you go from 22 to 14 and the rate changes.  The measured graph is as below, mm of compression along the bottom, kg up the LH side:

Adding two constant gradient lines to the graph shows the rate change more clearly, starting at 0.63 kg/mm and changing at 90 mm of compression to 0.91 kg/mm.  In a Monster I'd usually fit 0.85 to 0.95 kg/mm springs, depending on rider weight.  I don't understand why they choose these springs.  But, if you cut off the tight section, you get a 0.9 ish kg/mm rate spring that is quite usable (possibly a bit short though, the Showa ones are better to cut down in my experience).

The springs as fitted have 20mm of preload.  With what I find as your average rider on board they'll give 45 to 55 mm of sag.  Which means they'll be compressed 65 to 75 mm and have 42 to 49 kg of load on them.  More on that later.

On to the internals.  The cartridge has heavily swaged ends, so is very much non rebuildable without getting medieval.  And destructive.  But it's always nice to see what's inside, and the pipe cutter was happy to accommodate.  I didn't take a photo of the assembled cartridge, I forgot.  The following photos and text are a fairly basic description of the cartridge system.

The cartridge itself is a steel tube with a rod inside it.  On the end of the rod is a piston, in this case carrying a rebound shim stack.  The bottom of the cartridge is at the RH side, and the valve on it is simply to allow oil to flow into the cartridge through the large round hole you can see.  The small round hole is to allow oil to leave the cartridge on compression.  There isn't any real compression damping function in the cartridge.  When being compressed, apart from the small hole, the oil flows through the piston via large ports that have a flat plate over them.  The plate is to stop the oil flowing back through these holes and bypassing the shim stack when the piston is rebounding.  While this flat plate is held in place with a soft spring, it's only there to keep it seated.

The four small holes you can see are the rebound ports.  The four shims to the right are the rebound shims.  They act like a circular leaf spring, and the oil being forced through the ports as the piston moves bends the shims up to allow itself to pass through.  On the left is the compression plate.  The photo below shows the piston ports.  The large ones are the compression transfer ports.  The size of the rebound ports can be a restriction in themselves, and is why lighter oils are often used when revalving work is carried out.  One of the advantages of the Race Tech Gold Valves, for instance, was the increased flow area.  More flow area means you need heavier shims (usually just the same as shown previously, but stacked with more layers) to handle the increased flow, but also increases the ability to widen the range over which you can have effective control of the damping rates.

So, to summarise.  The fork cartridge has a basic rebound damping shim stack and no compression damping function.  I have heard of the damping being different left to right in these forks, but the parts catalogue shows only one cartridge part # for both legs.  Having one compression dedicated leg and one rebound leg is not a new idea, but it is becoming increasingly popular again.  For example, the Showa BPF (big piston fork).

I have read posts from Rick at Cogent Dynamics saying that, in his opinion, the valving in these forks is quite good for what they are, and certainly equal to the Showa non adjustable forks also used by Ducati during the era.

I thought I should ask Rick, instead of just misquoting him.  His reply was:

"Who knows what I said??? ;-)  It is true that while the Marzocchi fork is not serviceable or revalvable, the damping design is better than the crappy Showa, even the adjustable ones from the era.  The Showa can be fixed but the list of crap wrong with those cartridges is a long one.  The adjustable ones on many of the late 90s to even now for all I know are super crappy in that the rebound adjuster bleeds cartridge pressure into the cap where it effects compression damping as much as rebound.  Also, from what I see on the dyno, the damper rod pumps its self up with air and causes lag in VERY hard use.

We do have our own replacement cartridges that fit right in to the Marzocchi  forks making them very good and also adjustable if wanted.  We have a couple sets running around down there on the bottom of the world with you.

It is interesting if you also plot the top out spring effect at the beginning of travel (before the sag).  When we fit spring to those forks we reduce the diameter of the spring guide (depending on the wire diameter of the new straight rate spring).  The cartridge you took apart has a base valve that looks much more simplistic than the one I had apart.  The one I had used a standard type shim stack/ valve and check plate on the base valve.  I have some intact 800 SS cartridges here but I have to dig through my Damper Dynamometer files to see if and what I have of dyno reports on those…

The spring/oil height combo really can improve the fork feel, those progressive wound springs kinda work some for everyone but not too good for anyone.  Not too a bad choice if you're Ducati, I would say."

Moving on, the next stage, and something that I took the chance to spend a couple of hours doing today, was to check the effect of oil height on effective spring rate.  In this case, effective spring rate refers to the impact of increased air pressure inside the fork leg as the fork compresses.  The air gap inside the fork leg decreases as the leg compresses, leading to a pressure rise.  Well, unless the seals are leaking.  This pressure acts equally on and perpendicular to all surfaces, so it tries to expand the tubes as well.  But what we are concerned with is the pressure acting on the top of the oil (assuming in this instance the top of the outer tube is the fixed portion of the spring system), which pressurises the oil (which is not compressible, unlike air) and thereby transmits that force to the bottom of the fork leg, just as the spring does.

I have an Ohlins chart for this from their manual for the R/T 43 mm forks, as below.  In this application, the oil level is specified with the springs fitted.  Compared to the Monster settings below, where the oil level is set without the spring fitted, the spring fitted will change (raise) the oil level, probably in the range of 40 to 60 mm from my experience.  This system works in the Ohlins R/T 43 because the spring sits at the bottom of the forks and is covered by the oil.  In many other forks, the spring sits at the top and protrudes from the oil, both making measuring difficult and introducing another variable (being spring design/configuration).

I did this by reassembling the fork (tack welding the cartridge) as if it were being fitted to a bike and then compressing the leg in the same way as I test springs.  This way I measured the overall load on the leg as it was compressed, again in 10 mm steps.  I ran 3 different oil heights with the std springs, 105, 125 and 145 mm, as measured without springs, preload tubes or seats.  105 mm is the specified oil height for these forks.  Then I replaced the springs with some linear rate ones (oem ST4 from memory) and ran the test again with an oil height of 125 mm.  This was as much as time allowed.

There is a small amount of obviously incorrect data in the results (bumps in the graphs), but overall the results were rather interesting.

The first graph shows the overall results, then I'll break it down.  The "spring only" curves are for the springs tested out of the leg, and this is the base rate so to speak.

The three oil level settings with the original Monster springs is as below, with the Monster spring only as a comparison.  I have offset the spring only curve based on the 20 mm spring preload, to try to be a little more accurate (?).  And I have modified the data so all the curves start at 0, just to make it easier to read.  I'm not sure if how I have done it is 100% correct, but it gets a bit confusing.  The 125 mm curve is a bit high up to 50 mm of compression, so there's a little error there.  But the shape of the curve is the main point of all this.

As you can see, even 145 mm oil height increases the spring rate at 120 mm compression from 0.91 kg/mm to around 3 kg/mm.  The 105 mm oil height curve ends at 110 mm compression simply because before I got to 120 mm compression the total load went over 200 kg and my scales turned off.  At that point the effective rate is over 4 kg/mm.

Next I replaced the original spring with a linear ST series spring measured at 0.83 or so kg/mm.  It's shown in comparison to the Monster spring below.  As the photo shows, the linear spring at the bottom is longer.  In terms of material volume, a basic calculation shows the volume of the linear spring is 41.5 cc and the Monster spring 42 cc, an almost negligible difference.  But the black plastic oem spacer at the RH end of the Monster spring is 110 mm long, 38 mm OD and 4 mm thick.  Its volume is 47 cc, which is significant.  This sort of thick plastic spacer is common in the Marzocchi forks, whereas the Showa have a thin walled steel tube with nylon end supports.  The piece of aluminium tube I cut to preload the linear spring is 32mm od and 1.6mm wall thickness.  It's what I use for new preload tubes in place of the steel originals when replacing springs in Showa forks.  The piece shown has a volume of 4 cc, so the total volume of the Monster spring and spacer combo is around 89 cc, as opposed to 45.5 cc for the linear replacement.  That 43.5 cc difference is significant, as the next graph shows.  Assuming the ID of the fork tube is 39 mm, it calculates to an oil height difference of around 40 mm.

Both these next curves are for 125 mm oil height, green is the ST series linear spring, purple the oem Monster spring.  The 27 cc greater spring/spacer volume of the oem parts relates to maybe 30 mm in oil height at a guess, which is another variable when replacing springs.  Dropping the oil level with the oem setup to 160 mm or so might give an equivalent air gap change.  Although, as the oem spring volume is more concentrated at the top (you always fit a non linear spring with the tight section at the top to reduce unsprung weight), this may also lead to the air gap volume decreasing at a higher or accelerating rate with increasing compression compared to the linear spring, again compounding the difference.  As ever, it's a case of the unknown bringing you undone.  Or just confusing you.

Adding the Monster spring with 145 mm oil height curve in black shows this more clearly.

In hindsight I should have pulled a new 0.90 kg/mm spring from the shelf to use for this comparison.  The springs I usually use are between 260 and 297 mm long, depending on brand.  But, unless I used a thick plastic spacer, I'd possibly end up with an even more varied result if I had a longer aluminium preload tube.  The plastic tube is hard to find in this sort of size (38mm od ish) and suitably thick wall thickness, and the closest thick wall orange electrical conduit is on the small side OD wise for my liking.

The next graph sums up my frustration with the way the Ducati forks are set up as they leave the factory.  The red curve is oem at the 105 mm oil height spec and orange the ST linear spring with 125 mm oil height.  The green line represents 30 kg loading, which equates to 40 and 30 mm compression respectively, chosen because it represents an approximate loaded sag setting.  The blue line is 145 kg, which is the force at 120 mm compression with the linear ST spring.  I chose this as it is probably a good representation of the max load the fork will see (they bottom at 122 mm travel), and it illustrates my point nicely.

If we assume 30 and 145 kg represent our end points in on-road use, then the oem setup operates between 40 and 105 mm of travel, or 65 mm total effective travel.  Over the same load range the linear ST spring uses nearly 90mm of travel.  This is the point that I don't understand.  Why make use of a little over half of the available travel?  Using more of the travel with an overall softer rate at the end point will help it absorb bumps when loaded heavily.  Specifically, braking hard into a bumpy corner.  With the oem setup, the effective spring rate in that case would be over 3 kg/mm.  With the linear setup, it is half that or less.  This will allow the suspension to work, whereas the oem setup will be more likely to just bounce over the bumps and unsettle the bike.

I believe this is why sport bike forks run linear springs and are going to lower oil levels, increasing the compliance at close to full compression.  From a design point of view, it may be the separation of influences that is the big plus.  Much like the Showa BPF damping separation, anything that can be done to clarify influences and reduce interference between them is desirable from an engineering stand point.

Bringing the progressive spring point up again, the above example leaves me with no understanding as to why they are so popular.  The effective spring rate in the oem Monster fork changes by a factor of 7 or so over the total fork travel, and leads to a reduced range of travel.  It seems to me that most progressive springs start out too soft, which just gives you excessive sag.  Excessive sag is just using up one end of the travel range, for no reason I can see.  Even the linear ST spring gives an effective spring rate change of a factor of 2 or so over the total travel range with 125 mm oil height.  So I guess the question is "how much progression is desirable?"

The linear spring gives an effective progressive rate as tested, and raising the oil level would increase that just as effectively as changing the spring would.  The big qualifier there being "within reason".  The relationship between internal component volume and oil height will also be relevant.  All forks will be different in that aspect, and the results in this report apply only to these forks.  I'm sure the Showa forks, with their reduced spring and preload spacer volume, will require higher oil levels to show the extreme increase the 105 mm oil height does in these Marzocchi forks.  Going back to the 90's Monster 40 mm Marzocchi and 41 mm Showa forks, oil levels in them were 80 to 90 mm.  And there is an early M900 service bulletin to add another 30 ml of oil to them to reduce dive under hard braking.

The other side of that is lowering the oil level until it starts bottoming out, then go back up 10 mm.  But the spring is the first thing to get right.

One thing some do to try to help a too soft spring is to increase the preload.  I have been told that a soft spring overly preloaded will give a harsh feeling towards the end of the travel, so I made another graph with the oem Monster spring preload increased.  I think it's representative and reasonably accurate.

I've added 13mm of extra spring force to the measured load to give curves for 105 and 145 mm oil height and compared these to the ST spring with 125mm.  As you can see, with more preload and less oil it's not too dissimilar to the ST curve in the working range.  I might try to physically measure this setup, and maybe drop the oil level even more.

The extra preload may help reduce the dive under brakes too.  Without any real compression damping function in their design, any help you can get there is a bonus.  Raising the oil level will possibly help that too, but given these results it's a variable I'd be going the other way on for all the other requirements.

Which is not to say that the ST curve is ideal.  It's just more like I would do myself, but now I may think about that more too.

I'm very pleased with the test results.  I wasn't expecting such an extreme rate change with the higher oil levels.  I have heard of people saying you can restrict overall fork travel with oil level, but I think personally I'd rather do it with spring rate.  On a bike that is designed from a marketing viewpoint, such as the Ducati Hypermotard for example, the suspension setup (soft, long travel) is what is expected by the market, not what is desirable.  and you could certainly raise the oil level on those forks to reduce the total usable travel.  But you're still stuck with the excessive sag and oft associated dive under brakes, and that's the thing that annoys me the most.  Replacing the springs and leaving the oil level at a normal height would give a much better result in how the bike performs.  In my opinion, anyway.

As I said, I'm no suspension expert.  But I do find it interesting up to the points that I can understand.

Tuesday, December 24, 2013

Closed for the holidays

I'm closed for the holiday season until Monday, January 13.

Best wishes to all for whichever holiday or not you choose to celebrate.

Many thanks for your business and support over the past year, and looking forward to not stuffing it up in the future.