Rafi: thanks for inviting me to the discussion!
One of my favorite topics, actually...
First, from above: Bwaller has it right on the $. Ilbikes proved it, too. (Beautiful bike, Gordon!). I have seldom had a dyno to use, except for a couple of summers in 1972-3 when a local truck shop had enough interested parties to let me in on Sundays to play. What I learned fit perfectly with a theory book on airplane exhausts I had been reading (from the Merlin Engineers of WWII, Britain). During those 2 summers I tested most of the then-available 4-1 and 4-2 pipes while trying to beat the "100 MPH wall" we had encountered with the K3/4. In the end of that, it was NOT the pipes that caused it, but that's a different topic.
Here's what I learned that may help you muse on: the exhaust flow MUST be in line with the intake valve size. Any other items chosen as the basis for the pipe size, length, backpressure, etc., with cause confusion, flat spots, or simply poor performance (Gordon: for Strokers, this is the total intake port size, adjusted downward according to crankcase pressures). The original 750 pipes were based on the intake valve, with even the final exit port still being the same diameter (minus the anti-finger bar across it).
The next most important thing is: the temperature of the exiting gases. As they get hotter, their pressure rises, because the cylinder's "dump" is a fixed volume and the exhaust valve size is fixed. So, this creates a "pulse" of exhaust pressure. If you leanout the fuel mixture a little bit, it cools off the exhaust: if you richen things up, it increases temps until the pipes (and valves) burn. That's where it gets thorny, if you don't have a Detroit V-8 feeding one pipe. You can either use the energy of this pulse to:
1. create a suction for the next exiting pulse (4-4 pipes)
2. time the pulse expansion via a spread in the pipe to cause a suction (4-2 pipes) for that next exhaust stroke
3. join the pulses (you got 4 here to work with) to create a suction over a wider RPM range (4-1 pipes)
4. tame the pulse, via baffling, so as to make the exhaust pressure (and suction) more even through the RPM range, while still creating a suction moment at the RPM range of interest (hint: this takes a late intake valve opening and an extended exhaust valve open duration, all else being equal - which we know as the "F" cam). This method tends to become more 'peaky' (at higher RPM, too) than any multi-pipe method. It also tends to make the engine overly sensitive to carb size and efficiency in the RPM range where the cam works (sound familiar to you "F" guys?).
So, what's a boy to do?
First thing: realize the length limitations of this bike, and consider using foldback baffling (like BMW pipes often do/did) if you have room for large "expansion chambers", relative to the exhaust volume. Or, consider a balancing pipe between the 2 "pair" of pipes to spread out the time duration of the pulse by dissipating it a little bit (not as efficient, but works over a wider RPM range). Chances are, this pipe would block your oil pan's drain plug...
If you're doing 4-2, you've at least eliminated 2 other possibilities to struggle with: you have 1 chamber for a 90-degree Twin, because you probably won't make a complex crossover under the bike to get the 180-degree cam timing working for you: now, you're in Harley territory (i.e., 90-ish engine degrees (I know, Harley is even less...) between exhaust pulses, then nothing for 270 degrees). This gives the opportunity to let that 270 degrees duration get a long way down the pipe, with all that energy, to 'pull' on the next [exhaust] valve opening. But, unless you are making a Swordfish-piped Chopper, there isn't enough length behind you to make this pulse create suction above 4000 RPM. So, you've narrowed the variables a lot.
In the 836cc bore (9.81 cu. in.), the airflow past the intake valve at [normal] atmospheric pressure with the 5 BTC/35ABC cam and optimal lift (which is about .305") matches very closely the air volume in the pipe before the muffler * 2.5 (with the stock 31" long headers) headers. So, you have 2.5 engine cycles to consider, and they are stacked as 2-at-90-degrees apart and can be considered as 1 pulse above 4000-ish RPM. The next 270 degrees being quiet, the moving pulses will drop the pressure in the pipe until they are either slowed down or broken up. If the pulses enter a larger chamber, it should be tapered to support the cooling gases at about the same rate they ARE cooling, to maintain the suction: this is where dynos come into play, adjusting the tapers. If they simply exit the pipe (drag pipes), so does the suction. If they taper too slowly outward, the pulse pushes back toward the engine.
A diffuser as the first baffle is one way to spread out the RPM range where the suction remains, and this is EXACTLY where most commercially-available pipes fail us today (especially 4-1 pipes). After that barrier, the sound pulse can be used to help create additional low-pressure zones because the pulse front of the exhaust is lost. Honda used this to make the length of the HM300 pipes almost spot-on for 7000-7600 RPM use, with the sound pulse traveling the open section of those pipes up to the fiberglass diffuser, pulling suction out of the diffuser at the pipe's upward bend. (Those of you riding the sandcast bikes probably know the sense of "screaming" from the pipes at this RPM).
In the end, on these bikes, pipe length is everything, in a sense. The only guys who got this right, besides Honda on the K0-K1, was Dunstall, who upped Honda by several HP, while adding about 10 lbs to the bike. Their pipe length nearly doubled the path overall, inflicting deep enough suction at the exhaust valves that we had to richen up the needles (even the mainjets) in some of them to keep the engine cool. A few chopper riders I knew added Swordfish-length pipes on their Amen frames to accomplish this, too, and their 1000cc-equipped roadsters dominated anything they wished!
Now...if you're not going racing, you can probably ignore most of this stuff, and just make sure the mufflers start with a diffuser at the end of the header, have a significant tapered portion up to the first sound-breaking port (which can be up to 4 each 1/2" holes in a plate across the whole muffler) at about 20" from the header, and then figure out how to break up the crackle as it exits the muffler. Honda used a 19mm hole in a long pipe, which we know as "the plug" in the end of the HM341 pipes. By the time the energies reach this part, they aren't making much suction anymore, so it matters only a little, unless you make the pipes stick out as long as the Dunstalls, from the end of the bike. The 836 engine is going to peak at about 7000 RPM as the [stock] intake valve maxes out anyway, so it will feel much more like a Kawasaki 900, with its 7000-ish RPM redline, anyway.
