S54 Header Design talk

[3staxontheradio]

euro/CSL header I'm getting primary lengths on this bank of about 304mm, 275mm, and 301mm. Pipemax has that as being on the short end of tuned for the 5th harmonic for a peak torque RPM of about 4900. Also if I scale VE in pipemax for 346 crank horsepower it wants a diameter very close to the CSL header. To me this suggests that the pipemax configuration is really close to dead on since the CSL header I have scanned is apparently dialed right into correct for 5th harmonic.

The issue is that the 5th harmonic is far from ideal. Its going to be a much more intricate bundle for snakes for the much longer 3rd harmonic runners but the scavenging pulse will be much stronger. If there weren't manufacturing cost constraints and challenges packaging longer runners I don't think BMW would have chosen this route. They did because its a mass market car and mass manufacturing is a serious consideration.

[digger]

Euro headers 🤓

Actually it does work pretty well for a stock/mild engine (probably just needs a pipe size bump) and there is some science behind the length of primary (higher order harmonic) and collector and how they work together (not surprising) but like always you get to a point where you need to decide where in the powerband to compromise to make a gain

[digger]

Absolutely not. The euro headers are miles off. I can measure them out but they are waay too short and clearly living on a smaller harmonic than is ideal. It's not even close.

So far Kromer Kraft are the only off the shelf headers that seem to plausibly tuned well but NLA. I don't think supersprint can even really claim their V1 headers are well tuned when the runner lengths are literally all over the place. Not really a thing to tune runner lengths and then just use a 5" spread of different lengths. Which of the many lengths is considered the tuned one?

My takeaway so far is that S54 off the shelf headers have mostly been skipped over for off the shelf products. I'm sure people like achilles motorsport have been making much better custom headers but building headers from Jbends is probably too expensive for an off the shelf product compared to CNC bending used for other products. 3d scanning makes packaging precision way easier than it was 20 years ago. OEMs, even for the euro cars have a bunch of competing constraints that we don't have and CNC bending tube comes with a bunch of limitations for bend placement.

I'm pretty confident we can do better and also produce equal length secondaries at the same time so the car sounds better.

If you average the pipe lengths on the V1 that is pretty much how it behaves from a power perspective per graph above, IMO there isn't a need to get it perfect equal length to the nearest 1/2" that has become a marketing thing for the most part.

One thing i found is that pipemax calculates a primary length and 3rd harmonic that works well as a zoomie (open primary), but when you add collector it throws the rpm where the scavenge wave is in tune off (it tunes in at a lower rpm since the primary and collector start to interact) and the shorter primary (higher 5th or 7th harmonic) does seem to correct it to get it to tune the scavenge back in at higher rpm and probably one reason why BMW did this along with packaging. It can still be optimized for sure.

[3staxontheradio]

This is good to know for the next set of testing. When you say primary and collector start to interact can you explain what is happening there? Is there a reason not to just dyno test and then shift the length if its off instead of jumping to a 5th harmonic? Also hitting equal length dead on is a pretty different concept than having that much spread. That irks me especially that the price point of the V1s. I also really doubt there is a good reason for it other than CNC bending limitations.

I did a quick dig into Bry5on logged data to look at EGTs. Since EGT has a big effect on how fast pulses move I wanted to get a better sense for how close pipemax is and try inputting manual EGTs. This is what I got for EGTs at the end of a WOT pull. Then for sanity I did a rough division for how many RPM moving the length changes. The EGT spread here (690 to 730) would equate to moving the peak power about 150 rpm. It would be good to get more logged data from people but it seems like a 700 degree EGT at peak power (7900 as configured) is pretty close. Also I don't think this tune (bry5on mullet tune?) is pushing the limits very hard on timing advance at WOT, that's just my understanding from how the tune was built so bry5on feel free to correct me. That also probably applies to anyone else who doesn't running a fully custom dyno tune which I think is the majority of people. EGT is a tricky factor because its super transient and also different setups, different VE, and different ignition timing will have a direct effect. It doesn't seem right to just handwave EGTs since the impact is large but it becomes unknowable.

[Bry5on]

Correct, not pushing boundaries on timing, but it is a couple degrees hotter than OEM in various places at WOT. It’s basically tuned to CA 91 octane until it gets intermittent/random ping events, then is backed off a few degrees. I’m logging knock and DME timing retard at 100Hz for that tuning.

[3staxontheradio]

digger I think on a second read I see what you are saying about primary and collector/secondary interaction? Is the suggestion that a third harmonic is going to be narrower peak and that as a result the primary and collectors tuning won't overlap as well and if kept at the same target RPMs there would be a torque dip in the middle of the graph?

Also comparing supersprint V1 to CSL diameters as ID:

CSL runner ID
38.5

Supersprint ID
38.8mm stepped to 44.8mm


For a well tuned stock cams engine (VE tuned for that to 363 crank hp) 2 step engine pipemax wants
37.7 ID transitioning to 40.91 ID

If you add 10% VE for mild cams it wants:
38.8mm ID transitioning to 42.7mm ID

[siamkazi]

Absolutely not. The euro headers are miles off. I can measure them out but they are waay too short and clearly living on a smaller harmonic than is ideal. It's not even close.

So far Kromer Kraft are the only off the shelf headers that seem to plausibly tuned well but NLA. I don't think supersprint can even really claim their V1 headers are well tuned when the runner lengths are literally all over the place. Not really a thing to tune runner lengths and then just use a 5" spread of different lengths. Which of the many lengths is considered the tuned one?

My takeaway so far is that S54 off the shelf headers have mostly been skipped over for off the shelf products. I'm sure people like achilles motorsport have been making much better custom headers but building headers from Jbends is probably too expensive for an off the shelf product compared to CNC bending used for other products. 3d scanning makes packaging precision way easier than it was 20 years ago. OEMs, even for the euro cars have a bunch of competing constraints that we don't have and CNC bending tube comes with a bunch of limitations for bend placement.

I'm pretty confident we can do better and also produce equal length secondaries at the same time so the car sounds better.

well, FWIW, count me in for a set 👍, this is really cool to watch unfold

[digger]

This is good to know for the next set of testing. When you say primary and collector start to interact can you explain what is happening there? Is there a reason not to just dyno test and then shift the length if its off instead of jumping to a 5th harmonic? Also hitting equal length dead on is a pretty different concept than having that much spread. That irks me especially that the price point of the V1s. I also really doubt there is a good reason for it other than CNC bending limitations.

I did a quick dig into Bry5on logged data to look at EGTs. Since EGT has a big effect on how fast pulses move I wanted to get a better sense for how close pipemax is and try inputting manual EGTs. This is what I got for EGTs at the end of a WOT pull. Then for sanity I did a rough division for how many RPM moving the length changes. The EGT spread here (690 to 730) would equate to moving the peak power about 150 rpm. It would be good to get more logged data from people but it seems like a 700 degree EGT at peak power (7900 as configured) is pretty close. Also I don't think this tune (bry5on mullet tune?) is pushing the limits very hard on timing advance at WOT, that's just my understanding from how the tune was built so bry5on feel free to correct me. That also probably applies to anyone else who doesn't running a fully custom dyno tune which I think is the majority of people. EGT is a tricky factor because its super transient and also different setups, different VE, and different ignition timing will have a direct effect. It doesn't seem right to just handwave EGTs since the impact is large but it becomes unknowable.

with the open header there is an infinite change in area once you reach the end of the pipe so the length is clearly defined. When you add a collector pipe the area change is much smaller than infinite so what seems to happen is the effective length of the primary becomes longer (or it "acts" longer and the "suction" wave arrives not perfectly aligned with the valve overlap phase)

[3staxontheradio]

I'm struggling to integrate that into my understanding. Are there other places I can read about this? Has this come up in other threads on Don Terrill's speed talk?

I'm trying to step through and rationalize what is actually specifically happening. This is where I get with it:

Zoomies/open primaries act like a clear discreet length and strong rarefaction wave is generated at a single point because the pressure differential at the the exit is constant state of atmospheric pressure.

When primaries dump into a merge collector pressure in the collector is variable and the secondary has its own tuned pressure harmonics that interact and can cancel out or diminish wave strength?

I'm struggling with a mechanism for "acting longer" aside from something happening where at some RPMs collector harmonics cancel out or diminish a percentage of primary tuning waves reducing the frequency effective wave frequency/ acting longer?

[heinzboehmer]

The problem becomes support material removal for complex geometry like this. You’d end up with a lot left in between the walls, which could get heavy fast. A weld on set of shells with some locating bosses would probably function better but at greater cost.

Why not thicken up the walls slightly and let the infill do the work?

Could design the walls and infill into the model itself for full control over the final part.

[Bry5on]

Why not thicken up the walls slightly and let the infill do the work?

Could design the walls and infill into the model itself for full control over the final part.

Weight and thermal conductivity would both take a hit but you could in theory. You’d end up with non-uniform thermal conductivity but I guess that could be acceptable. I dunno, doesn’t seem like the right choice to me.

[maupineda]

Absolutely not. The euro headers are miles off. I can measure them out but they are waay too short and clearly living on a smaller harmonic than is ideal. It's not even close.

So far Kromer Kraft are the only off the shelf headers that seem to plausibly tuned well but NLA. I don't think supersprint can even really claim their V1 headers are well tuned when the runner lengths are literally all over the place. Not really a thing to tune runner lengths and then just use a 5" spread of different lengths. Which of the many lengths is considered the tuned one?

My takeaway so far is that S54 off the shelf headers have mostly been skipped over for off the shelf products. I'm sure people like achilles motorsport have been making much better custom headers but building headers from Jbends is probably too expensive for an off the shelf product compared to CNC bending used for other products. 3d scanning makes packaging precision way easier than it was 20 years ago. OEMs, even for the euro cars have a bunch of competing constraints that we don't have and CNC bending tube comes with a bunch of limitations for bend placement.

I'm pretty confident we can do better and also produce equal length secondaries at the same time so the car sounds better.

My response was facetious, but also pragmatic, as in euro headers are a good compromise (in every technical way), and up to not long ago, affordable.

[3staxontheradio]

Yeah they certainly got more expensive after that Kassel video came out. I have Megan racing headers that came with the car, at some point I will scan and measure one out and get a really good direct comparison to the euro headers. I think those are likely the best bang for your buck low cost option. Just looking at them they seem to be copies of the euro headers with some minor differences. Can't tell what the runner ID is from pictures though

[digger]

I'm struggling to integrate that into my understanding. Are there other places I can read about this? Has this come up in other threads on Don Terrill's speed talk?

I'm trying to step through and rationalize what is actually specifically happening. This is where I get with it:

Zoomies/open primaries act like a clear discreet length and strong rarefaction wave is generated at a single point because the pressure differential at the the exit is constant state of atmospheric pressure.

When primaries dump into a merge collector pressure in the collector is variable and the secondary has its own tuned pressure harmonics that interact and can cancel out or diminish wave strength?

that is what seems to be happening but i don't know if it is technically the correct explanation plus you have other cylinders connected to the collector pipe now as well. Remember there is a end of pipe at the end of collector so one large area change plus the smaller area change at the end of primaries so logically it seems unlikely that the system will behave exactly the same as if just the primaries are present.

Also imagine if the collector pipe was the same size as the primary would the engine ever see the primary length? or imagine if you retain zoomies but added a big step after the "primary" to simulate the collector but it didnt connect to each cylinder (stayed as zoomie)

i will try and get some plots in the next day or so of the resulting pressure waves to illustrate what i am talking about and that the shorter primary seems to generate a suction wave that aligns better with overlap around the peak hp rpm.

[digger]

Here are some plots
1) zoomie exh: 540mm zoomie 40ID mm step to 43ID mm (3rd harmonic primary)
2) Collector exh 1: 540mm primary 40ID mm step to 43ID mm (3rd harmonic primary) 1,800mm collector 2.5" OD (2nd harmonic)
3) Collector exh 2: 290mm primary 40ID mm step to 43ID mm (5th harmonic primary) 1,800mm collector 2.5" OD (2nd harmonic)


Power and torque



Exh Pressure