Author Topic: "Doing it Right" or "How to Build a Functional Café Racer"  (Read 116185 times)

Offline Sonreir™

  • Moderator
  • *
  • Posts: 10137
  • Oregon
Re: "Doing it Right" or "How to Build a Functional Café Racer"
« Reply #60 on: Oct 15, 2012, 20:12:25 »
Applications - Top End Repair and Health

Well, I was going to continue with my semi-logical order of performance upgrades, but I've had an interesting couple of weeks and I figure I'll diverge a little bit in order to share what's been going on in my shop.

A couple of weeks ago, on the way to work, my daily driver (1996 Geo Metro, 1.0L) started missing on one of the cylinders at idle.  Once I revved it up to 3,000 RPM or so, the cylinder would kick back in.  The car got me to work, but on the way back home, the engine started overheating as well.

I suspected a blown head gasket.  A loss of compression will cause the misfire and leaking coolant system will cause the overheating.  Both of these things occur when the head gasket fails on a liquid cooled engine.

Turns out, I was correct.  This is the oil I drained from the engine and when it's in this condition it is known as "milkshake".  If drained from a recently run engine, it will be very frothy, hence the name.  I was expecting about three quarts of oil, and ended up with a gallon of this stuff; another sure sign something is not right.

Funnily enough, I've never really taken apart a car engine before, but having some experience with bikes I figured I'd give it a shot.  $100 in parts sure beats $1000 for the parts and labor, I figure.

So a few hours later, I'm at this point (this pic was taken after cleaning up the sealing surface on the top of the block).

Right away, we can see that cylinder one was the problem.  The coolant, leaking into the cylinder during operation, has steam cleaned the piston and combustion chamber, resulting in the shiny surface you see here.

While everything was disassembled, it's also a great time for a general health check of the top end.  I recommend this to anyone looking to restore a vintage bike.  I'm pretty sure I mentioned it earlier, but it bears repeating:  An engine is only as good as its weakest link.  You can have trick components, a hot cam, and high compression pistons, but if your valves are leaking then it's wasted effort.  A stock engine in good health will almost always outperform a built engine that's limping along on one or more questionable parts.

The Acetone Test
Now that I had the engine apart, it was time to start checking things over.  The first test I usually perform on the head is what many folks call the acetone test.  The idea is fairly simple.  You pour some acetone into each of the ports, one at a time.  Sit the head up on its edge and add a couple of table spoons into the first port.  Let it sit for a few minutes and inspect the combustion chamber for leakage.  Any acetone that makes it through indicates a failure of the valve to seal and this must be corrected.  Here's a pic of a failed acetone test on GS450.  The liquid leaking past is clearly visible.

It's important to note that other fluids may be used, but you're looking for something with very little surface tension.  Paint thinner, denatured alcohol, and gasoline are all viable fluids for this test.

Like the picture above, my Metro failed on all six valves.  Time for some work.

Inspecting the Valves
Now that we know that valve sealing is an issue with our engine, we can pull the valves and inspect what may be wrong.  Even if nothing is wrong and you pass the acetone test, it may still be a good idea to pull the valves.  This will let you check the valve stem diameters, stem-to-guide clearances, and also lets you replace the valve stem seals.  These next sets of pics are the intake valves, and the exhaust valves, respectively.  All six valves were cleaned using a combination of Simple Green and a soft wire brush, prior to taking these photos.

As you may have guess from the photos, the exhaust valves are all shot.  Each valve is clearly pitted along the sealing surface.  The intake valves are showing a little discoloration, but are in otherwise decent condition.  I reused them.

OK.  So we know that new exhaust valves were needed.  Time to inspect the valve seats within the head.

Again, the intake valve seats look to be in good condition (nice shiny surface, no pitting), but the exhaust valve seats have seen better days.  Though not immediately apparent in this pic, there was very little pitting, but they still weren't as good as I would like and carbon build up was a clear sign that we aren't getting as much metal-to-metal contact as is required.

Now... Sometimes compromises need to be struck.  In a perfect world, every job should be done perfectly, but this is not a perfect world.  We are limited by our time, our resources, and our own skill levels.  An argument can be made for not doing a job to perfection, but there is no excuse to not maximize the use of the time, resources, and skills we do have.  In this scenario, I was limited very much by resources ($$$), and so I chose not to have a machinist clean up these valves seats.  The price for a valve job usually runs about $20-$25 per valve, so I was expecting a bill of around $150 to have the valve seats cleaned up and that's something I could not afford at the moment.  A car that is running at a sub-optimal level is more useful to me than a car that is not running at all and so I decided to proceed without the valve job.  I do want to make it clear that is something that should have been done, but was not.  If this were to enter a mechanic's shop, they probably wouldn't guarantee the work without having the valve seats recut by a machinist. The overall health of my engine was affected by this decision, and I will get into that later.  But basically, know this:  Before you make a decision to stray from the beaten path, you should ask yourself one single question.  "What will happen if I do it this way, instead?"  If you can't answer that question, don't do it.  Decisions should be made from a base of knowledge rather than necessity.  In this case, I know that my approach may not fix the problem, but it wasn't likely to make it any worse, either, and so I choose to proceed.  Also, had this been on my motorcycle engine I would have done the job properly and not taken any chances, even if I had to wait a few months to get the money together.  My bike and my car serve different purposes in my life and so they get different treatments.

Clean up
Now that we know some work is needed, the next step in the process is to get everything cleaned up.  Even if no work is needed, it's still advised to undertake this step.  The only reason I list the acetone test first is that it helps to give you an idea of what problems may exist and it saves you at least one set of (dis)assembly routines.

Aside from pain involved in working with greasy and dirty components, sometimes it's just plain impossible to make the measurements needed with bits of leftover gasket, chunks of carbon, and blobs of grease all over everything, so it's time to make this thing shine.

Chances are, your head is made from aluminum and so you're going to want to follow some important procedures when cleaning.  First, don't expose the head to any heat levels that would be exceed normal operating temperatures.  No blow torches or crap like that.  For my own cleaning, hot water is about as hot as I get.  Second, try to avoid media blasting, if you can.  Soda is OK, but stuff like glass and shells can plug things up and coarser media like sand will usually cause damage.

The first step to cleaning is a soak in a citrus cleaner like Zep, Simple Green, or even Pine Sol.  For really caked on grease, you can use a plastic brush to work the cleaner into those areas.  The second step is a good scrubbing with soap and hot water.  I use a combination of another plastic brush, and occasionally a 3M scratch pad.  Any carbon deposits left after this stage can be cleaned with a carbon dissolver such as Zep Morado, but don't leave it in contact too long.  This stuff can hurt aluminum over time and it shouldn't be used on areas without carbon build up.  A wire brush may be necessary to fully remove the carbon deposits.  To remove any gaskets, use a razor blade, but scrape against the angle of the blade instead of into it.  This will avoid gouging the head and causing any scratches.  If there is coked oil at any point, kerosene or diesel can help get these off.  Finally, any left over gasket bits can be removed with some 400 grit paper and WD40.  Be very careful when sanding, however, as sealing surfaces are very intolerant of low spots.  Most specs allow for only .003" height differences across the entire sealing surface.  You may need to have the head resurfaced if you get too heavy handed at this point.  A final wash in soap and hot water followed by air drying should get you ready for the next steps.  3M Roloc pads in a die grinder can also be used for gasket removal, but use them sparingly.  Power tools get through aluminum in a hurry and what may save you time now, may cost you both time and money later.

The Valve Job
As I just mentioned, I chose not to pursue a valve job at this time, but I'm going to include the information here for future reference to those seeking to "do it right".

Many people, at this point, would hand the head over to a machinist for the necessary work, and that's OK.  I did the same with my own 360 build.  I'm going to cover how to do your own valve job, however.  You will need some special tools and this does take time, but it's also quite rewarding.  The pics for this section have been "borrowed" from around the 'net, so these represent what I would have done, rather than what actually happened.

First up, the tool set.  You will definitely need a valve cutter.  Neway is a company that makes these for a reasonable cost and no special power tools are needed for it's operation.  They usually cost between $250 and $400 for one of these things, so it's probably only worth buying if you plan to do your own valves at least half a dozen times (many vintage bikes will not have a three angle grind from the factory, so this tool can be used for performance and not just repair).  Make sure to buy the kit with the cutting heads of the appropriate size.  These kits are not one-size-fits-all, but they are one-size-fits most.  They look like this when they're new:

The first step in the cutting process is to insert the pilot.  This acts as a guide for the cutter to ensure the cut is precise.  If you are planning to replace the valve stem guides, you'll definitely want to do that before doing any valve cutting.

Next, slip the 45° cutter in place over the pilot.

And finally, the cutting handle.

After a few turns (only a little pressure is needed, better to cut too little than too much), some of the valve seat should be cut.

Repeat this process for the inside (60°) and then the outside (30°) angle and you should now have a nice, shiny, valve seat.

At this point, you'll want to cover your nice work with some layout dye (aka machinist's dye aka Prussian Blue).

And then repeat your 45° cut.

The thickness of the metal strip that is exposed should be around .040" for the intake valves and .060" for the exhaust valves.  This 45° cut serves two purposes.  First, this is the sealing surface.  When your cut is finished, all of the blue dye should be removed.  Any spots of dye that are left after the cut usually indicate pitting in the valve seat.  You're going to need to cut again, but slightly deeper.  The second purpose is that the thickness of the valve seat has to do with heat transfer.  Because fresh intake mixture is coming over the intake valves every time they open, they are cooled by this mixture.  On the flip side, it's all hot exhaust gases passing by the exhaust valves and so the seats need to be thicker to allow the heat to transfer from the (relatively) hotter exhaust valves to the cooler head.  If this seat is too thin, then you burn up your valves in short order.

Your 60° cut (the one further up the intake port) should be roughly twice the width of the 45° cut and the 30° should be about 75% of the width of your 45° cut.  If you've made your 45° cut too wide, if can be narrowed by recutting with the 30° cutter in order to make that width wider.  There are only so many cuts you can do in a head, though.  Cutting extra means the total lifespan of your head is reduced as most heads do not have replaceable seats.  Some guys are clever enough to be able to add more metal to this area, but don't count on simply finding someone who is capable of this type of work.  That said, it'll usually be cheaper just to get a new head, anyway.

At this point, you can reassemble things and place the valves, springs, retainers, etc, back into the head.  Repeat the acetone test and hopefully you will see zero leakage after around five minutes of waiting.

If leakage still occurs, it is important to know why.  Were you a bit haphazard in your approach?  Did all the measurements check out?  Did the 45° cut fully remove the dye from the valve seat?  If you're happy with the way things went during the cutting process, it may be necessary to lap the valves.

Valve Lapping
Many machinists will advise you to lap the valves as a matter of course.  For me, this seems unnecessary and I think this recommendation is one of those things that just gets handed down out of tradition rather than real need.  As a side note, if you're running with titanium valves, you don't want to lap them.  If the acetone test fails, then you need to recut the seats.  Titanium valves almost always include special coatings and lapping will remove them.

Lapping is a lot like precision sanding.  Lapping compounds are usually some sort of lubricant with different levels of grit mixed into it.  The idea is the rub the valves back and forth across the seats until both the valves and the seats are nice and smooth.  This is accomplished by adding a bit of lapping compound onto the valve head and pressing it into the seat (by hand).  A valve lapping tool is then used to spin the valve in place and the compound cuts into the valve and the valve seat.  A lapping tool is essentially a wooden stick with a suction cup on each end.  Don't buy a cheap one of these off of eBay as you want a good suction cup for this not to end up being a huge headache.  Auto parts stores usually stock the good ones.  Lapping should not be done by attaching things to the valve stem and spinning (such as I've seen suggest with power drills and such).  Also, lapping should be done prior to the replacement of the valve stem seals.  It's OK to lap with the old seals in place or removed, but not OK to use the new seals during this process.

For the lapping process, itself, just do what this video is showing.

Repeat this process until the "sound" of the grit starts getting quieter.  Switch to the next finest grit until it gets quieter and so one and so forth.  Grits commonly used for valve lapping are 120 grit, followed by 220 grit, followed by 400 grit.  Some people even go out to 800 grit, but I think this is overkill.  Also, you may not need to start at 120 grit unless you skipped the cutting process, like I did.

As another side note, lapping compound is useful to have around the shop to clean up sealing surfaces on heads and jugs prior to assembly.  Spread a liberal amount over a sheet of plate glass and work your part in a figure eight pattern for ten minutes, rotate it 90°, and go for another ten.  This clears up microscratches and the like and is also a good way to correct any scratches you may have introduced to sealing surfaces during the removal of old gaskets.  A bit of diesel or kerosene can be added to the lapping compound to thin it out a bit.  Thinned out compound is called "slurry" and is available for purchase directly, but I just prefer to keep the compound around and thin it myself, if needed.

When you have finished the lapping process, the shiny valves and valve seats will have developed a satin gray stripe along the middle.  Lapped valve on the left, new valve on the right.

After running the valves in your engine for a few hundred miles, this gray satin finish will eventually be worn into the shiny finish like you see in the valve on the right.  The lapping process helps provide an initial seal, but like cross hatching on cylinder walls, it eventually wears away and creates an even better seal after doing so.  You should not count on the engine to do all the work in this process, though, the idea behind a good lapping job is to minimize the time this wearing-in process takes and doing it poorly can result in a poor final seal as well.

After this lapping process, you should have a seal good enough to pass the acetone test.  If not, you need to make a decision.  What the problem with the cutting or was it with the lapping?  More Prussian Blue can tell the tale.  Paint it around the valve seal and place the valve into position without twisting it.  Pulling the valve back out should result in a coating of dye all the way around the valve.  Any missed spots indicate a failure in the cutting process and you'll likely want to repeat it.  If you get dye all the way around the valve, you can try lapping it again.  A failure after another round of lapping will mean you'll want to recut, though.  Make sure your seat widths are within spec, as well.  Too narrow and you could have sealing problems.  Other things to check are bent valve stems or damaged valve guides.

Back to my Decision
As I mentioned earlier, we'd revisit my decision not to cut the valve seats.  After getting the new head gasket into place and reassembling the engine, I decided to run a compression check.  The spec on a fresh engine for the 1.0L Geo Metro is 195 PSI with the lower rebuild limit coming in at 165 PSI.  My first test netted me values of around 130 PSI.  After discovering some cam timing issues and retorquing the head (this should be done after covering a few miles so as to compress the gasket a bit further) I'm up to 150, but still below spec.  There are a couple of things which could be working against me.  First up, I mentioned that I was doing this work under tight financial constraints and so I choose not to have the valve work done.  One thing I hadn't yet mentioned is that I also skipped out on the head surfacing (which was probably needed as well, but would have been another $100).  Here's a pic of the sealing surface on the head and the pitting here is quite clear.

So... how does that affect my compression?  Well, it is not uncommon for head gasket kits (especially those in the auto industry) to come with slightly thicker-than-stock head gaskets.  It is assumed that resurfacing the head after 200,000 miles is something that will just be done and so the head gaskets come slightly thicker than they would, otherwise.  I don't know that this is for certain in my case as I did not measure either the new or the old head gasket.  This is simply a possibility.

Assuming the head gaskets had no difference in thickness, then I suspect the lapping I did on the valves either helped only a little, or not at all.  The car's power output does seem a bit low, but that could also be from me being used to riding on two wheels these past few months.  I have no data to which I can compare and I know only that it could be better than what it is right now.  With any luck, the valves will continue to seat in and my compression will slowly rise over the coming miles.  But having to choose between a car that doesn't run and one that doesn't run as well as it could, the choice was pretty clear.

OK... so your head is screwed up and you're not sure where to begin.  Start with some simple tests and some cleaning and go from there.  Remember guys, you're working with technology that has been around for a while now.  This isn't rocket science and it's not as hard as you'd think.  Take your time, make your measurements, implement your solutions.  An engine that runs always puts out more horsepower than an engine that doesn't, and so making sure you're up to spec should be job number one.  The reliability and performance of your engine will depend a lot on the time you take in diagnosing and correcting any problems you encounter.  Also, be aware, that when it comes to engines, new parts aren't always enough.  Some parts (even new ones) may need machining in order to fit, but not all machining needs to be done by a machinist.  There are a lot of hand tools available that will get the job done and you can learn a lot of new things along the way.
« Last Edit: Jul 08, 2015, 18:25:36 by Sonreir »
Sparck Moto -

Audaces fortuna iuvat.

1977 Honda CJ360 - Café SOS - Stage One™, Café SOS - Stage Two™
1976 Puch Maxi - APuchalypse Now
Suzi T500 Cobra Resto

Custom Gauge Graphics
Custom Wiring Harnesses

DTT Red, White, and/or Black 360 Club - Better than those Blue guys

Offline teazer

  • *
  • *
  • Posts: 8132
Re: "Doing it Right" or "How to Build a Functional Café Racer"
« Reply #61 on: Oct 15, 2012, 23:55:24 »
Good run down there Matt.

We get all our valves and seats cut on a Serdi machine.  It's more accurate and doesn't leave striations on the seat- aka chatter.  With a Serdi cut it is neither desirable nor necessary to lap the valves.  We just pop them back in and they are perfect.  Whereas if I do them, they are not so perfect.

Offline Sonreir™

  • Moderator
  • *
  • Posts: 10137
  • Oregon
Re: "Doing it Right" or "How to Build a Functional Café Racer"
« Reply #62 on: Oct 16, 2012, 00:16:49 »
Please correct me if I'm wrong, but with a Serdi you also get all three angles in one cutting?  That'd be nice.  :D
Sparck Moto -

Audaces fortuna iuvat.

1977 Honda CJ360 - Café SOS - Stage One™, Café SOS - Stage Two™
1976 Puch Maxi - APuchalypse Now
Suzi T500 Cobra Resto

Custom Gauge Graphics
Custom Wiring Harnesses

DTT Red, White, and/or Black 360 Club - Better than those Blue guys

Offline bradj

  • *
  • *
  • Posts: 6166
Re: "Doing it Right" or "How to Build a Functional Café Racer"
« Reply #63 on: Oct 16, 2012, 09:30:38 »
I was a machinist for 8 years and take alot of what you posted for granet but its not. kool write up it will prove helpful to many im sure. O yea we had a vaccum guage to check valve seating it was pretty simple and very telling if there was a problem

Offline scottyfunj

  • Posts: 104
Re: "Doing it Right" or "How to Build a Functional Café Racer"
« Reply #64 on: Oct 23, 2012, 13:39:41 »
Thanks for all the extremely valuable info.  This thread is becoming something of a motorhead's bible.

Offline mohabie

  • Posts: 121
Re: "Doing it Right" or "How to Build a Functional Café Racer"
« Reply #65 on: Oct 23, 2012, 16:31:14 »
Thanks for the valve job step by step... Though I will still pay someone to do it, I understand it better than before.

1984 BMW R65
1981 Honda CM400
1980 Suzuki GS850S
1973 BMW R75
1977 Suzuki GS750

"You dont graduate from leadership, you graduate into leadership"

Offline Corsair

  • Posts: 150
  • Not a Purist...just Traditional
    • Ground Up Cafe' Racer
Re: "Doing it Right" or "How to Build a Functional Café Racer"
« Reply #66 on: Oct 23, 2012, 21:21:30 »
Excellent post, and well written. Also, perfect timing, since I'm just getting into the engine work on my CB350f. Looking forward to more!

Sent from my Nexus S 4G using Tapatalk 2

Offline Sonreir™

  • Moderator
  • *
  • Posts: 10137
  • Oregon
Re: "Doing it Right" or "How to Build a Functional Café Racer"
« Reply #67 on: Nov 07, 2012, 13:16:40 »
Applications - Exhaust Design
Time for a new post, I figure.  As you may have expected, one of the major areas still to cover is the exhaust system.  I've
saved this item until now because your design at this point is going to be very dependent upon earlier selections.  Furthermore,
there are several options you can take in order to change the overall characteristics of your engine (for good or bad) with only
a little regard as to your previous choices.  In short, the exhaust system is one of those areas where you can really make a
difference in the performance of your bike, especially in bikes as old as ours.  Even in newer machines, it is not uncommon to
see exhaust changes as the primary means of modification in engine performance.  For those of you guys riding smokers, the
exhaust design is, arguably, the single most performance consideration.

As with previous posts, the plan is to break this up into manageable chunks.  Each subsection will deal with a different
consideration and will hopefully tie it all together as we reach the end of this post.

Before we begin, there are a few things you'll need to know in order to make the best use of this information.  First, you need
to know your displacement as well as the number of cylinders on your bike.  If you don't know these first two things, please hand your keys over and go buy a Toyota.  ;D  Second, you need to know the exhaust timing on your selected cam.  Finally, like with most everything else, you need to have a general idea as to where you'd like your peak torque to occur.

Exhaust Valves and Ports
The first part of the exhaust system actually begins inside the head.  Like intake valves and ports, a three or five angle grind will provide benefit here.  But unlike intake valves, less emphasis is placed on things like tapered stems or backcutting.  The reason being is that the gases exiting the engine are under much higher pressure than the mixture when it enters.  This created a much greater pressure differential between the cylinder and exhaust system than that is between the cylinder and the intake system.  This greater differential (often called "delta P") means that the little details become less important and the bigger picture become more important.  In this case, the bigger details are exhaust valve and port sizes.  The traditional approach is that exhaust valves should be sized at approximately 80% the size of the intake valves.  Hemispherical heads with a sharper include angle will benefit from slightly larger exhaust valves due to the tight radius the exhaust ports must follow.  This design philosophy is clearly visible in the ratios seen in the Honda 350s and 360s (34mm intake valve, 28mm exhaust valve, 82.4% sizing ratio).

For exhaust ports, the diameter should be the same as the valve size (or only slightly smaller).  Usually they are smaller and so taking some metal out could be a good idea.  Pay special attention to the port ceiling as this is where most of the gains will occur.  Also, like intake ports, a 'D' shape with a flat floor will generally out perform purely circular ports.  Exhaust port walls should be as smooth as possible and can be sanded down as fine as 800 grit paper.

Finally, the transition to the header should occur with a sharp step.  The header will have a significantly larger diameter than your exhaust port and the change to this larger diameter should be sudden, not smooth.  This fast change helps to prevent reversion, especially at lower RPMs.  Reversion is a killer of torque and so you can expect a bit more power down low and in the mid range with this step in place (especially if you're running a hotter cam).  If, for some reason, this step doesn't exist in your system, an insert can usually correct the problem.  I know MikesXS sells them, and they're not too hard to make, either.  They look like this, and you can clearly see the sharp step created as you transition from the head to the header:

Header Design
When it comes to exhausts, there are two types of tuning with which you should be familiar.  The first, called inertia tuning, which involves overall design of the exhaust and its length.  The second, known as acoustic tuning (very similar concept as pulse tuning for intakes), is very important for two strokes, but also has an effect on four stroke engines.  Both, however, are of some importance and so you should be familiar with the basics.  As with my previous posts, I'll stick mainly to the four stroke applications.

Also, the majority of the design in an exhaust system will go into selections on the header.  The gases in the exhaust system are the hottest just after they leave the cylinder and enter the header, so this is where many systems fall down.  Correct sizing and lengths can make several percentage points of difference in peak numbers.  Tweaking things here can also change the characteristics of the engine without having to open up the cases again, so if money isn't an object it's definitely worth exploring having several different header configurations.

Inertia Tuning
The concept behind inertia tuning is that the exhaust gases leave the cylinder in regular pulses and that these pulses tend to stick together as they travel through the exhaust system.  This means that each pulse is a discrete unit of high pressure traveling through pipe and due to inertia, these pulses leave a low pressure zone behind them as they travel.  The efficiency at which your cylinder is emptied of exhaust gases is directly related to the delta P between the cylinder and the header and so keeping the exhaust system in a low pressure state will translate into better volumetric efficiency and more torque.

The best design for inertia tuning has been shown to be a four-into-two-into-one (4:2:1) for best mid-range torque or four-into-one (4:1) for max power.  Obviously, twins only get the 2:1 option.

Finally, we need to decide on a total length for the exhaust system (we'll handle lengths for the headers and collectors and such in the next section).  For inertia tuning to work best, an exhaust pulse must be able to travel the entire length of the pipe before another pulse enters.  It's OK to have two pulses in the pipe at once, but zero is bad.  As a pulse leaves the end of the pipe, the low pressure zone in its wake will pull in gases from wherever possible.  We want this low pressure zone acting on our cylinder to pull out more exhaust gases and not acting on the atmosphere to pull fresh air into the end of the muffler.  Basically speaking, our exhaust needs to be long enough so that an exhaust pulse remains in the pipe up to the point at which another pulses is added.  At 10,000 RPM, a new pulse is entering the pipe every 0.012 seconds.  This means our pipe needs to be long enough for a pulse to travel its distance in .012 seconds (or more).  If the pulse can travel the distance of the pipe in less than .012 seconds, we need a longer pipe.

Exhaust gases will usually travel at between 200 and 300 feet per second.  Overly large exhaust pipe diameters will slow this speed (which isn't good) and pipe diameters that are too small will speed this up (also not good).  I prefer to use the 300fps number as it's a bit of a worst case scenario.  The actual speed of your exhaust gases can be calculated and I'll run through the math for that toward the end of this post.  Just remember that if you set things up correctly, your exhaust gas velocity should fall somewhere in or very near this range.

So... now we know how fast our gases are traveling, the next thing we need to do is decide upon the ideal RPM at which we want our exhaust to work.  Just like everything else, the total exhaust length should be tailored to the optimal RPM of your engine.  For this example, I'll use 8,000 RPM.  At 8,000 RPM, a four stroke engine is creating an exhaust pulse 4,000 times per minute.  That reduces down to 66.7 times per second. One divided by 66.7 gives us an exhaust gas pulse every .015 seconds.  We can then multiply our 300 feet per second value by .015 seconds to get our resulting length.  We end up at 4.5 feet.  If your desired RPM is higher, this length will reduce.  If your desired RPM is lower, the pipes with be longer.  When in doubt, err on the side of length.  Chopping pipes too short seems to be a common design decision these days (looks?) and hurts power up until the point where the engine revs fast enough to take advantage.  For pipes chopped up near the foot pegs, few bikes can even rev high enough to take advantage of this.  Even if your bike can rev up to 15,000 (in theory), short pipes are hurting power until you get to those speeds and so your redline (in practice) may be lower.  Looking at Honda's iconic RC166, we can see the pipes extend all the way to the back wheel, despite having a redline of 18,000 RPM.

Bear in mind that this calculation method applies to a single pipe per cylinder.  For the more idealized approach of collected and merged header, total length will be shorter, again.  For merged headers, the idea is to use the amount of time elapsed between each pulse of all the cylinders rather than each pulse from each cylinder.  For bikes with more than two cylinders, use the Acoustic Tuning methods listed below.  For 360° twins (XS650, Brit bikes, etc), take the total length and divide by two.  180° twins like the vintage Hondas should multiply the total length by a factor of .75.

Acoustic Tuning
Acoustical tuning in the exhaust system works on the same theory as pulse tuning in the intake system.  When the exhaust gases explode from the cylinder and head through the exhaust system, a high pressure sound wave is generated.  A low pressure wave reflects back from the high pressure wave as it exits the pipe and the idea is to time this low pressure wave to come back to the exhaust valve at the right time.

Strangely enough, the total length isn't so much as calculated as it is derived.  By summing the lengths of the properly designed components, we end up at a more-or-less ideal total length.  The length of the headers sort of changes the slopes of the peak torque curve.  Longer headers will begin to generate more torque, sooner, but will cause a rapid drop off as RPMs increase.  Short headers will do the opposite; a rapid increase in torque with a slower drop off.

In order to accurately size the header pipes, you will need to know your cam timing as well as the displacement of each cylinder.  We'll get to the cam timing part in a minute, but we'll talk about displacement right now.  The displacement of each cylinder dictates how much exhaust gases are generated.  The amount of exhaust gas and the diameter of the header work together to dictate the speed of the exhaust gases.

Time for some math...  First up, we need to know how much exhaust gas is being generated by your cylinder (not cylinders, we're only interested in one of them right now) at your ideal RPM.  Falling back to my previous examples of my own CJ360, I'm aiming for an ideal engine speed of 8,000 RPM with a per cylinder displacement of 189cc.

The first thing to do is to decide on the inside diameter of the primary headers.  The diameter of the header pipes is what determines the exhaust gas velocities.  Smaller pipes increase the gas velocity and cause peak torque to occur earlier, while larger diameter pipes slow down gas velocities and cause peak torque to occur later.  I'll touch on the heavier portions of the math later in this post, but for now, just take it on faith that each cylinder is generating 866cc of exhaust gases every period (two rotations per period).  At 8,000 RPM, we're getting 4,000 pulses every minute and so this translates to 3,464,000cc of exhaust gas per minute.  This is almost exactly 122.33 cubic feet of gas per minute.

Mr. A.G. Bell has been kind enough to test things thoroughly for us, and so we know that peak torque, from the exhaust system, generally occurs at 250 feet per second.

Using our ideal gas velocity and the amount of gas being generated at our desired RPM, it is now possible to calculate the inside diameter of the header pipe which will meet these requirements.

The formula for calculating gas velocity through a duct is v = q/A.  v is the resulting velocity in feet per minute, q is the gas flow in cubic feet per minute, and A is the cross sectional area of the duct in question.  We know the velocity is 250 feet per second and so we convert to feet per minute and get 15,000.  Plug the values into the formula and we get 15000 = 101 / A.  Algebra allows us to reconfigure the equation to solve for A.  We now have A = 122.33 / 15000.  A equals 0.008155 square feet.  Convert back to inches and we have 1.17437.  Because we're working with a circular pipe, it's time to break out the geometry skills.  1.17437 in² = pi*r².  1.17437 divided by pi give us .3738 = r².  Take the square root of both sides and we now have r = .6114 inches.  Double it to get our diameter and we're at 1.223 inches.  The nearest sized standard pipe is 1.25 inches and so we'll make use of that.

Now that we have our pipe sized correctly, it's time to get the length of the header.  The calculation for the length is fairly simple.  Length, in inches, is equal to (((850 * (180 + EBBC)) / RPM) - 3.  EBBC is the number of degrees before BDC that the exhaust valve opens.  In my own engine, this number is 54° and so the calculation becomes (((850 * (180 + 54)) / RPM) - 3.  Unlike other RPM selections, the RPM value used in this calculation should either be the redline of the engine, or the midpoint between peak torque and redline.  I'll be using my redline of 11,000 RPM.  Working the math through gives us an ideal primary header length of 15.1 inches.  Should you ever arrive at a length of less than 15 inches, just go with 15.  You don't want these things too short.

Before we talk about the next aspect of exhaust design, I'll give you a slightly easier method of pipe diameter selection.  This method tends to result in a slightly larger diameter pipe, but it's a lot easier to follow (especially after I get to the "heavy" math section).  After calculating the primary length, you can solve the issue of diameter by D = sqrt((cc / (pL + 3) * 25) * 2.1.  Basically, take the square root of the cylinder displacement divided by the primary length plus 3, times 25.  Then take that result and multiply by a further 2.1.  Using the same numbers are my above calculations we end up with an inside diameter of 1.35 inches.  When it comes time to match up to a standard pipe size of 1.375, this is only one step up from my original choice of 1.25 inches.

Secondary Headers
OK... with that out of the way, we can calculate the length of the secondary header.  This part only applies if you're planning on running a 4:2:1 system (or basically any system that doesn't converge directly into one pipe, so this applies to 6:3:1 as well, but not 4:1 or 3:1 or even 6:1).  The "secondary" nomenclature applies to the section that falls between the primary headers and the tail pipe.  In four cylinder applications, this is the set of two, after the four, but before the one.

The length of the secondary header is calculated in very much the same way as the primary header, but with an important difference.  The total length of the primary plus the secondary should be the value from the formula, multiplied by two.  If the formula yields a value of less than 15 inches, the primary header should remain at fifteen inches and the secondary head should be shortened accordingly.  For instance, if our formula gives us a length of 13 inches, we get 26 inches for the total length of both the primary and secondary headers.  15 of those inches are reserved for the primary header and the secondary header (and collector) should take up the remaining 11.

When adding a secondary header into the system, it is important to recalculate the diameter of the pipes being used.  The secondary headers should be slightly larger in diameter.  To calculate the value, raise the original value to the power of two and then multiply by two.  Square that result and then multiply by .93.  Round to the nearest standard pipe size and you're good to go.  Going back to my original example of 1.223 inches we follow the math through like so:  sqrt((1.223 * 1.223) * 2) * .93 = 1.609.  Round to standard pipe sizes and the secondary headers should be 1.625.

Exhaust Tuning
One of the nice things about saving the exhaust system until last is that we can fine tune the engine to the desired levels without having to open it back up again.  Changing primary and secondary header sizes and lengths will move and reshape the torque curve to match your desired characteristics.

For instance, if you want to lower the RPM at which peak torque occurs, you can reduce the diameter of the primary header (because secondary header diameter is derived from primary header diameter, don't forget to resize the secondary as well).  A reduction in size of a single step will drop peak torque up to 1,000 RPMs lower.  An increase in header length will result in lowering the peak torque as well.  It's even possible to use unequal length headers to broaden the torque curve as one cylinder will hit peak torque before the other(s).  On a four cylinder engine, you may decide to go -2 inches on one header, keep the second the same, and then go to +2 and +4 on the other two cylinders, respectively.  This approach is clear to see on the CB400F pipes.  They look good because they work well.

One thing that hasn't yet been mentioned is collectors.  The collector is the section of pipe that joins the primary headers to the secondary headers and the secondary headers to the tail pipe.  The design of the collector can help to promote low-end torque.  The most common type is what's known as the merge collector.  The concept is that each of the header pipes smoothly transition into the collector rather then end abruptly.  This type of collector is best suited toward maximum horsepower, but does little to promote any boost in torque in the lower RPM ranges.  You'll also see these types of collectors in use for turbo manifolds.  In this particular example, a venturi has been added in order to help scavenge gases.  The venturi makes this design slightly more efficient from an inertial stand point.

Another type of collector is known as a baffle-type.  In this design, each header ends abruptly and is joined by a single cone to the tail pipe (or secondary header).  This design is simple to implement, but suffers for pure racing applications due to some turbulence as the exhaust gases merge. For street use, this is just fine.

Split interference collectors are kind of a halfway house between the two designs listed above.  They're what you usually see in stock collectors.  Each header will end abruptly, but instead of a simple cone for the collector, you get a smoother transition like you would see in a merge collector.

I'll talk a bit more about collector design in a future post, but for now, just be aware that there are different types and they don't all behave equally.  It's not life-or-death differences though, either.

Selecting the length of the tailpipe is fairly easy.  Though I prefer to use the more complicated method I first listed a dozen paragraphs ago, you can always take the easy way out.  If you've designed a system with secondary headers, simply double the length of the existing system and you'll be at the correct length.  If only primary headers are being used, you'll want to quadruple the length.  Going back to our original example of 15.1 inches for the primary headers, our total exhaust length ends up at 60.4 inches, or just over five feet.  The length of the collector for the tailpipe should be included in this number.

The final diameter for the tailpipe is a fairly simple calculation as well.  First, double the displacement of a single cylinder.  Then divide that number by the primary length, plus three, and then multiplied by 25.  Take the square root and then multiply by two.  Now round to the nearest standard pipe size.  For our example of 189cc per cylinder and primary length of 15.1 inches, the equation should look like this sqrt((189 * 2) / ((15.1 + 3) * 25) * 2.  The result ends up at 1.83 inches which would round to 1.875 inches.

Unless you're making a full on race bike, you're going to want these.  All things being equal (and I mean equal in the performance sense) then open pipes will generally produce the greatest level of horsepower.  On the street, though, you're probably going to want some level of noise suppression.  The drone of open pipes gets pretty old after the first few times.  When selecting a muffler, you're going to want something that provides a decent level of noise control with the minimum amount of flow restriction.  Unfortunately, these values are rarely listed by the manufacturer.  If performance is more of a concern than money, try to stick to name brands with dyno-proven improvements over stock.  Also, the length of the muffler should count toward the overall length of your exhaust system.  Adjust things accordingly.

Tips and Tricks
There are a few little tricks you can pull to help things out even more.  One of those is called "stepped headers".  This is the same theory that is used to help two strokes out with exhaust design.  The concept is that as the header lengthens, you widen the pipe once or twice along it's length.  You only go up a step at a time and only need to do this once or, sometimes, twice.  The first step is usually located between 10 and 14 inches from the head.  Larger diameter header pipes will tend to favor the longer side of the 10-14 range while smaller diameters favor the shorter side.  If a second step is desired, this one is usually located just an inch or two before the first collector.

Also, like the intake tract, the exhaust tract doesn't much care for lazy welds or rough spots.  To keep the gases flowing, you want smooth surfaces with smooth transitions (steps, aside).  Welds, as much as possible, should be ground down on the inside, especially around the flanges and the head where gas velocities are the highest.

Next, the header pipes should exit from the head in a relatively straight manner without too much of a bend.

Finally, if your design necessitates headers that are of a smaller ID than your exhaust ports, try to have the headers flared to match the exhaust port diameter.  The ideal situation is one where the headers are of larger diameter than the ports, but if that ideal cannot be met, at least aim for the same diameter.

The Math
OK... I said I'd get to this part...  This is how I calculated out the 866cc of exhaust gases per cylinder, per period.  The reason I prefer this calculation over the simpler one is that it takes less for granted (or, at the very least, includes some extra variables which can be tweaked).

If you recall, the majority of the expansion of gases in a cylinder is due to the heat created by the combustion of the gasoline.  We can take advantage of this fact by using the ideal gas law to figure out how much expansion the gases have undergone.  All we need to know is the temperature of the mixture before the combustion event and the temperature of the gases after the combustion event.  For these two values, I am going to assume 72°F and 1800°F.  These numbers, of course, can be accurately measured with the correct equipment and I encourage that approach.  When these temperatures are converted to the Rankine scale, the quotient between the two becomes the factor of expansion.  For instance, 72°F is 531.67 Rankine and 1800°F is 2259.67 Rankine.  2259.67 / 531.67 = 4.25.  So the gases will expand 4.25 times due to heat, alone.

Though that number alone is probably a decent facsimile, lets get a bit more accurate and also account for the expansion in volume due to the combustion of gasoline (liquid to a gas).  For gasoline, we'll use C8H18 for its chemical representation and this gives us a molecular weight of 114.  Stoichiometry tells us that 12.5 moles of Oxygen will be required to combust every 114 grams of gasoline and will result in an output of 8CO2 + 9H2O.  Equal moles of gas are also equal in volume and so we know that the expansion in volume of the gasoline is going to come in at a factor of (8+9)/12.5 = 1.36.  However, the presence of Oxygen in the atmosphere is around 21% and so only 21% of our intake gas is affected by this calculation.  The final expansion in volume due to combustion can be calculated as (100 - 21) + (21*1.36) = 107.56%.

Going back to our rate of expansion due to temperatures, we can multiply that by 1.0756 to include the expansion due to combustion and multiply again by the displacement in our cylinder to get the final volume of the exhaust gases.  4.25 * 1.0756 * 189.58cc = 866.63cc.

One final thing before I finish this up.  Use stainless steel wherever possible.  A lot of the exhaust gas velocity is dependent upon the volume of the exhaust gases.  As we know, as the gases cool, the volume decrease.  This decreases pressure which decreases velocity.  In order to help keep the exhaust gases hot, stainless steel is the material of choice when building exhaust systems.  Other metals tend to be better conductors of heat and so stainless steel gets the nod in this application.

If, you find, during your sizing calculations that you regularly had to round up when choosing standard pipe size, the application of fiberglass header wrap can also net you a few extra ponies, especially toward the upper end of the RPM band.  If you've routinely had to round down on sizes, you may be better off skipping the header wrap (unless you're trying to shield something from the heat).

Despite the length of this post, I'm hoping that the details behind a decent exhaust system seem a bit easier to comprehend.  Like everything else, it's not black magic.  Selecting the correct lengths and sizes of your tubing is 90% of the battle.  Quality welds and smooth bends are the rest.

If you have access to a dyno, experiment with different lengths and sizes.  Pretty much everything listed above is just rule of thumb kind of stuff.  It'll get you into the ball park, but the home runs are only hit after empirical data is collected.
« Last Edit: Jan 08, 2014, 11:52:24 by Sonreir »
Sparck Moto -

Audaces fortuna iuvat.

1977 Honda CJ360 - Café SOS - Stage One™, Café SOS - Stage Two™
1976 Puch Maxi - APuchalypse Now
Suzi T500 Cobra Resto

Custom Gauge Graphics
Custom Wiring Harnesses

DTT Red, White, and/or Black 360 Club - Better than those Blue guys

Offline teazer

  • *
  • *
  • Posts: 8132
Re: "Doing it Right" or "How to Build a Functional Café Racer"
« Reply #68 on: Nov 07, 2012, 15:33:26 »
Nice write up Matt.

Collectors are interesting.  In order for them to work the way they are described in all the books, is that they have to end in an abrupt change in section.  Merge collectors do the opposite.  They try to make it a smooth transition. A sharp change produces the wave effects that Matt described but flow is not smooth.  Merge collectors have much better flow but generate much weaker pulses.

Current pipe designs favor flow over pulses and F1 uses merge collectors.

When calculating tuned lengths you must remember that the waves travel at the speed of sound and gas travels much slower.  We use the speed of sound in those calculations and that is effected by density and temperature of the molecules in the pipe, so it changes.  And we use gas velocity when working with the pipe sizes.

Anti reversion headers are interesting in that the flow out is primarily on the roof of the port down to the center of the pipe.  Reversion typically is on the lower side where there is less outward flow. The perfect shape is a D shaped port with raised floor.  A simple step simply restricts flow at all speeds and typically hurts top end power.  On a street bike that isn't always a bad compromise.

Jim Fueling had a patent (4,206,600) in 1980 with that simple design that Matt shared with us.  I have an article here that shows HP gains across the range with a special exhaust using aspects of that design. It was pretty clever for the time.  Subsequent development by guys like David Vizard and AG Bell took that one stage further and came up with the D shaped port which is more common now.

Stepped headers attempt to achieve a similar effect and crudely mimic a 2 stroke tapered header.  You will also see tapered sections in some MotoGP pipes and some Akropovic and Yosh pipes.  The idea there is to create a long duration, low amplitude negative pressure wave to assist in evacuating the cylinder.

In general, any sharp change in section creates a pressure wave and disrupts flow. The trick is to harness one while minimizing the harmful effects of the other.

Stainless is nice, but hard to weld and costs more. Mild steel is cheap and easy to weld but rusts.  Zinc coated steel lasts longer but generates toxic gases when welded, so be careful.  I only use mild steel and a can of VHT paint.

See the way that collector above flares out after the outlet?  That allows the gas to flow correctly.  Parallel pipe there hinders flow there somewhat.  merge angles also have an effect on gas flow and wave generation. Lost of stuff to think about if you want every last 0.1hp out of your project.

« Last Edit: Nov 07, 2012, 15:35:55 by teazer »

Offline crazypj

  • Posts: 12464
  • Split personality, I fake being smart
Re: "Doing it Right" or "How to Build a Functional Café Racer"
« Reply #69 on: Nov 09, 2012, 19:13:55 »
  I only use mild steel and a can of VHT paint.
 merge angles also have an effect on gas flow and wave generation. Lost of stuff to think about if you want every last 0.1hp out of your project.

I use mild steel, it's a lot cheaper.
Isn't the optimum angle 14 degrees? (included angle)
I remember reading about it on a NACA or NASA site years ago
 Last time I checked, SERDI 3 angle cutters were around $90.00 a pop, the tool to mount onto head (hand version) was around $1,500.00
Of course, it's cheap compared to industrial version ($2~$300,000, but it does 4~8 valve seats in one cut)
 BTW, like your math on the carb sizing for 360, I had hell of a time getting people to listen when I said 30mm instead of 32mm Mikuni's, I knew they didn't want the truth that 28mm was optimum on 350/360  ;D
« Last Edit: Nov 09, 2012, 19:24:39 by crazypj »
'you can take my word for it or argue until you find out I'm right'
I gave my girlfriend an orgasm the other night, but, she spat it back at me
 Some humans would do anything to see if it was possible to do it. If you put a large switch in some cave somewhere, with a sign on it saying 'End-of-the-World Switch. PLEASE DO NOT TOUCH', the paint wouldn't even have time to dry
 It’s not worth doing something unless someone, somewhere, would much rather you weren’t doing it  (Terry Pratchett)