GregBBRD

OK, if you say so.....I'm not looking to argue about what I'm doing.
But when I read what you are saying and what I'm saying.....I think we are saying the same thing.
Except, you are trying to explain what the waves are and how they occur.
And I'm not.
I'm not debating the existence of the different waves....
I'm trying to tell what I think (and have done, in the past) will help negate the "negative" effects of those waves and help increase the "positive" effects of those waves (which is the entire goal of a set of headers/exhaust.)

atb

That's actually not too far off. Putting aside Marc Thomas' HP claims for a moment, what made the DEVEK headers so popular is how well they fit. At the time, any other 928 headers had their collectors dump at the base of the firewall, AND one of the popular brands had one of the collector flanges welded "pointy side" down which reeked havoc on ground clearance. The DEVEK collectors actually reached under the car, well into the center heat shield and the collector flanges were "clocked" correctly "flat side down" so ground clearance was not an issue. Any other header at the time was a wrestling match to install. Not the case with the DEVEK header, which with one modular primary tube was actually pleasant to work with. Something to be said for ease of installation. LOL.

GregBBRD

These were better than the stock exhaust (and better than anything anyone else was making) 20+ years ago.
Is there room for significant improvement?
Absolutely.....but not for $600!

ptuomov

I think that anyone who's interested in spending some time on understanding how the pulses work inside an engine, Vannik's $400 EngMod4T has a really high return on investment.

Another thing that has personally helped me get comfortable with exhaust tuning is first observing others play with 10cc two stroke engines and then failing regularly tuning 50cc two-stroke and four-stroke singles. Engine designers used to commission production of a single cylinder prototype to validate their computer model. Yes, the interesting things about header design are about the multi-cylinder combos, but the basics are easily observed with a single cylinder engine.

I believe strongly that in NA engine, you need to size the exhaust port correctly for the peak power hp and then continue that pipe ID (really, the cross-sectional area) until about 12 inches from the exhaust valve. Then, step the pipe size up. It the exhaust port and the first primary "step" is sized to be just large enough for the peak power, then one in my opinion can step the pipe size up significantly. The best length of this second larger ID pipe in a single cylinder engine will depend on the intended rpm range. It's actually a reasonably easy setup to test on a 50cc motorcycle single cylinder engine because you can slide the big pipe over the "thick wall" small pipe with straight pipes and burn some electrical tape at the joint. What I believe (but don't know) happens is that the first step sends back a suction wave that should arrive when the exhaust valve is no longer sonic, and so the optimal length depends mostly on the exhaust valve flow area for the first 20 crank degrees vs. the torque that the cylinder makes. It's usually something like 12 inches from the exhaust valve with well-flowing 4-valve exhaust ports, I think. Then, the longer and larger diameter pipe is intended to send a suction wave as a reflection during the valve overlap. This longer pipe suction wave only works over a certain rpm range.

It's beneficial to combine the pipes with a collector, because all of the pulse energy can't be used productively to supercharge the engine. If the pipes upstream are sized correctly, the collector and what happens after it matters mostly for what happens below the peak power rpm. In a 90-degree crossplane V8, one needs minimum lengths for the primaries such that 180-degree interference and 90-degree interference don't happen. In 928, that means separating 5/6 and 1/3 with enough length to avoid 90-degree blowdown interference at high rpms and separating 1/4, 3/2, 7/6, and 5/8 to avoid 180-degree blowdown interference at low rpms. Those considerations give one a minimum length for the primaries before the 4-1 collector which ends up being about 30 or so inches from the exhaust valve (or a round trip from one cylinder to another of about 60 inches) for most V8s.

Beyond those considerations, I'd want to keep the pipes as straight as possible as in avoiding tight radius turns. I wouldn't compromise the straight flow paths for a tight radius turn just to get the primaries to be equal length in a crossplane V8.

GregBBRD

Maybe the trick is that instead of worrying about pipe length, you just stuff the pipes into the collectors in any haphazard convenient fashion....and make sure that the firing order isn't sequential in the collector and there is zero change of one cylinder "pulling" gasses out of the the previously firing cylinder....

There's pictures of both sides of Devek headers in this thread. Look at both sides, and number the pipes in the collector, from the cylinder head. Then get out your firing order sheet on a 928 engine and explain that science.

Strosek Ultra

Header primary pipes into collector by Mr. Brown.
Åke

Thom

This is perhaps the most straighforward way to proceed, at least with most cross plane crank engines which seem happy enough even with log manifolds...

ptuomov

How the 4-1 exhaust collector works in 90-degree crossplane V8 is a mystery to me.

The pulse traces in the collector choke don't show four pulses, they show two small pulses and one big pulse. I don't know what's the constrained optimal way to deal with that, given that the ideal is not possible.

Some collectors are modeled after 4-2-1 headers, except there's virtually no secondary length and the usual 4-1 primary length. In those collectors, they pair 1&2, 3&4, 5&7, and 6&8 to separate pulses before merging with a third Y. I think this is an attempt to make the third Y see less of a big bang from the combined pulses of 1&3 on the passenger side and 5&6 on the passenger side.

With rotating firing order, the usual 4-1 collector exhaust will have a swirl to it. There's a related issue in the turbo world. Turbine outlet produces strongly swirling gas flow as it comes out of the rotating tubine. You'd think that coming off with a megaphone diffuser from the turbine outlet would work great, but it appears to work about equally well regardless of which way you install the megaphone!!! Car factories like sharp expansion of cross-sectional area right after the turbine to kill the swirl and then shrink the cross-sectional area afterwards. Aftermarket tuners like to gently straighten the swirl with an expanding cone. How does this relate to the NA exhaust collector design? I think (but don't know) that rotational firing order will help when the collector choke is too large. This is because swirl causes the gas to flow a longer flow path in the same space, which means that the gas has to flow faster. If the collector choke is not small enough, then the swirl will make the collector look smaller and gas flow faster. I am speculating here.

In terms of the Devek headers, what would I do differently? I would come off from the exhaust port with a pipe that is smaller than 1.75" OD, trying to match the exhaust port size as closely as possible and port matching the whole deal at the flange. I've yet to see a 928 four valve engine that has a too small exhaust port, so I'd try to hit that. Then after about 8 inches from the flange, I'd step up to larger size, say that 1.75" OD pipe. I'd try to make the bends longer radius than in the current design and wouldn't worry about getting the primaries to be equal length The collector is above my pay grade, and in any case what's right for the collector probably depends on what's coming after the collector.

belgiumbarry

the 4 in 1 collectors seems to me very symmetrical between the primarys , so no importance how 1234 and 5678 come in . Or i'm i wrong ?

Strosek Ultra

Barry, the rotating firing order will create a swirl inside the collector as Tuomo describes above.
Åke

GregBBRD

Pretty much every professional in the header business will have the cylinders firing sequentially in the collector (in a circle)....so that the exhaust can "swirl" as it goes down the exhaust pipe. (The idea is that an adjacent pipe helps "draw" out the exhaust from the cylinder next to it in the collector (this is called scavenging.) Not only did I do this, but I reversed the direction of the two "swirls", so that they can merge together better at the "X" pipe. The Devek pipes (the subject of this thread) have opposite cylinders firing in the collector and then jump back to adjacent pipes. I'm sure the exhaust gasses will seek a lower pressure area, but I don't see how one pulse helps the previous pulse (which is what "scavenging is all about.)

A couple of drawings and just tossing a bit of logic at what is happening in an exhaust system goes a long way.....

belgiumbarry

thanks Ake …. i compared it to "my" M headers . Than 5678 side is OK ( rotation firing but clock wise) but not 1234 . Those come in 13 and 24 diagonal .

GregBBRD

Well,as long as we are just tossing ideas around....
Here's what I'm trying to answer (and I can, for a few thousand dollars, build two/three different varieties of headers to test):

My exhaust ports on Andy's engine are very close to 1.625" (which is what it took for me to get the exhaust flow to mathematically go with the intake ports (968 valves and 15 years of port development on the intake side.)
In my theory, the 'returning exhaust pulse" is laminar and is traveling back to the port along the walls of the primary. As I've pointed out, having it "arrive" when the exhaust valve is opening/already open is the worse possible thing that could occur.
If I make the primary pipe out of 1 3/4" pipe, and then go to 1 7/8" pipe 8"-12" down, the only "step" to redirect the reverse pulse is right at the primary pipe size transition....some 8-12" away from the port.
If I make the primary pipe out of 1 7/8" pipe, there will be a "step" right at the port exit, to "turn" the reverse pulse very close to the actual exhaust valve....but no step 8-12" down the primary.
I also have the option of exiting the port with 1 7/8" and then "stepping" the primary pipe to 2" pipe....which would make two "steps" to reverse the returning pulse. (2" pipe is a major pain to get past the steering, but it may be possible....I'm working on that.

Which is actually more beneficial, in your mind...and why do you believe that?

ptuomov

My opinions (not facts):

If the engine has high volumetric efficiency and a lot of camshaft overlap, the header can make a difference and the pipe diameters need to be right. If the volumetric efficiency is low and there's little camshaft overlap, just separating the interference pulses with any size pipes is going to work about as well as any header.

I recently read that the Ford Coyote super stock drag race car engine that makes 750hp has 37.6mm/1.480" exhaust port diameters. This I believe is because there's a lot of camshaft overlap in that race engine and it runs high compression (race gas). High compression makes exhaust cooler because of the expansion ratio, so the denser exhaust gas fits thru a smaller pipe. High camshaft overlap means that the engine can accept some significant pumping losses in order to generate a strong suction wave during the overlap.

In the case of the 928 engine in question, I would personally go off the flange with 1.75" OD 0.063" wall stainless steel tube. This is because 1.75-2*0.063 = 1.624" nails the port size. Flange would have to be port matched. I would go with that as straight as possible (long radius bend) for 8 inches off the head, and then step it up to a larger size. I would go to 2" OD in two relatively close steps (like an inch or two, but opinions vary). I would not worry about getting the primaries to be exactly equal length but I would worry about the bend radii being long and the pipes as straight as possible. I would go with a primary length target of 25-28 inches, so max about 20 inches after the first step. I'd merge those 2" OD pipes into a 4-1 merge collector that would have a 2.75" choke and and that would expand gradually to 3" exhaust pipe. I'd try have the collector end at 15" after the choke point, whether that's opening in free air on a dyno or an H-pipe in a more drivable setting.

That's just a guess, or an opinion. Might not work. However, there's some logic (possibly faulty) behind that spec in that I want the first step to send back a suction wave after the exhaust valve is no longer sonic at all rpms when the engine is making torque and then another suction wave around the camshaft overlap at the tuned rpms, which in this case would match a relatively high peak power rpm (like 7000-7500 rpm). If the first step is at where the head ends, then the first suction pulse arrives too early.

I'm sure that there are other people with better handle on things, I'm just giving my own honest best guess without much experience with V8 headers.

GregBBRD

I respect the amount of research and thinking you've done on exhaust, that's why I'm asking....
Clarification:
My exhaust ports are 1.625". A matching port exit pipe size would be 1.750" (1.625" ID), which would allow me to make a step to 1.875 downstream, very easily.
If I use a 1 7/8" exit pipe, I would end up with a step at the head.....and getting 2" pipe past the steering rack is a very tall task".....so there might not be "room" for a second step.
Your thoughts?
I have a plethora of BB Chevy stuff. Equal length has always "trumped" (which may have a whole new meaning, these days) unequal length. Getting equal length pipes into the rear two cylinders of a 928 engine requires multiple, fairly tight radius bends. You seem to be adamant about making the pipes as straight as possible, but if I do that, exhaust pulses will certainly be arriving ahead of the adjacent cylinders, in a large percentage of the time. (Granted, because of unequal firing order one each side of the engine, this also occurs in the collector, no matter what. Unequal pipe length exasperates this issue.
Your thoughts?