Lucifer's Hammer Part 8
Posted: Jan 04, 2007 2:07 AM
Lucifer's Hammer Series Index: Part 1 • Part 2 • Part 3 • Part 4 • Part 5 • Part 6 • Part 7 • Part 8 • Part 9 • Part 10 • Part 11 • Part 12 • Part 13 • Part 14 • Photos
Since somebody PM'ed me about how one can get Volumetric Efficiency (Ve) numbers, I'll throw out some more formulas.
An item worth noting is the essential part played by getting the actual airflow volumes for the head via a flowbench. Without knowing that value, any calculations are pure guesswork, though the turbo models can make pretty good inferences, given sufficient other data elements.
Head flows and Ve calculations:
Ve = (VAF x ES x C)/(D x rpm)
where
Ve = volumetric efficiency, or the amount of intake charge entering the cylinder on the intake stroke. The percentage is relative to the maximum cylinder plus combustion chamber volume.
Values above 100% indicate the intake charge is compressed above normal atmospheric pressure.
VAF = volumetric airflow rate.
ES = engine stroke (2 or 4).
C = conversion factor--cubic inches to cubic feet (use 1728)
D = displacement in cubic inches.
rpm = engine speed where the volumetric flow was measured, or for which the Ve is being determined. In this case, 5700 is what has shown up on the dyno as the top of my HP curve.
Knowing the head flowed 277 cfm @ 28" of water on a Superflow 600 flowbench, we get (277 x 6912) divided by (243 x 5700). 1,914,624 / 1,385,100 = Ve. 1.382 = Ve. Which checks pretty closely with the not2fast model displayed value of 139%.
For comparative purposes, IIRC, T.C.D. had an M30B34 head flowbenched some time in the summer of 2005. His post stated this cyl head had received a "Stage 3" porting job from a shop in TN that is used by Turner Motorsports. (possibly Memphis Motorworks?) This is the head that was on Todd's 745i that produced 331rwhp/316rwtq. Presumably this head flows pretty well. At .4" of lift it flowed 169 cfm in the intake valve and 134 cfm on the exhaust valve. If so, then Ve = (169 x6912)/(209.3 x 6700 ). [3430 cc = 209.3 cubic inches; I'm guessing at the redline here.] 1,168,128/1,402,310 = .833 Ve. This is very respectable. I believe the B35 flows a lot better--in the 200 to 210 cfm range. If that's the case, the Ve would probably go north of 100%.
Back to the story.
Mid-December, 2004. Paul, sounding immensely relieved, sez we're at engine mock-up. So it's off to Misery for a look-see. Best part of the trip across western Kansas was watching a blood-red Ford GT make a valiant attempt at breaking the sound barrier.Thass' nice. Should attract the KSP and leave me alone.
12/17/04 Need to figure on doing a UUC Shortshifter when we start putting things back together; easier to do with the car on the lift and the tranny out. The rubber shift lever boot , etc. is all dried out, cracked so all that stuff needs replacement as well.
12/19/04 At Paul's shop. The motor resideth on the engine stand in all its earthly glory.
By no means complete, but it feels like a major milestone has been reached, and completion is now a question of time. What follows is a listing, incomplete, and in no particular order, of items noted that are going to need attention.
~Exhaust headers being problematical as to proper alignment at the ports. Nothing weird, but some of the flange-to-head surfaces may need a final grinding to bring the mating surfaces to dead flat. Will need to enlarge the holes in the mounting flanges, as there is a little lateral pressure on the exhaust studs. This is a recipe for shearing the studs when the motor is at operating temperature.
~Playing around with the tubing alignment at the merge collectors helped a lot in getting the tubing joints to line up correctly. I want the header R&R to be a slam dunk. If the motor won't install with the headers in place, we will have a real chore on our hands with the motor in the bay.
~Except for die grinding, cleanup and Glyptal coating of the water pump flange, the short block is just about done.
~Block will get one more coat of paint; a few chipped and scratched areas from assembly.
~Crank is in place; oil pump going in this coming week, along with the piston assemblies. The Carillo rods are artistry in steel.
~Knock sensor installed; doesn't appear that there will be any problem in accessing it past the headers. The knock sensor lives in the former block drain location, so when it comes time to change or drain coolant, pull the knock sensor. Potential leakage controlled with a crush washer under the sensor. The location should give us good sensitivity to detonation noises and vibration.
~Plenum-to-turbine clearance measured at a bit under 1/4 inch. Plenum marked for a half-moon cut about 4 inches across. Paul will shape the concave filler plate and do the turbine heatshield design and fabrication. Paul's powdercoater is confident he can do the ///M tricolor after the welding is done on the plenum.
~A number of parts will need to be repowdercoated. Pieces went down in a couple of different lots, and the colors used weren't identical. The workmanship is good, but when pieces from different batches are put side-by-side, the color variation is obvious. Bare aluminum pieces -- throttle bodies, cam tray to name two -- will get a clear ceramic coating to control potential corrosion and allow for ease of removing engine crud.
~Head gasket is waiting for tooling to be completed at SCE. SCE thinks that will be around Jan 10, with delivery of gasket thereafter.
~Left a list of needed Aeromotive fuel system parts with Paul. Aeromotive is located in KC; he will talk with them directly and see if we can get any better pricing than by going through Summit Racing. Going to be enough $$ that it's worth trying.
~Paul is going to call Marren Fuel Injection to obtain a length of their ribbed fuel rail aluminum extrusion. Setting up the raw stock for proper injector alignment, attachment points, etc. is going to take some time on the mill. Paul's looking forward to this; a chance to get a bit creative . . .The Bridgeport Artisan. . . .
~The mild steel turbo flange will go back with me to Colo. Mark has a friend who can pantograph the part in a CNC mill, replicating the piece in 321 stainless. Once done, it can be TIGed onto the 304 stanless downpipe (which is still in Colo).
~Diverter valve will get located and TIGed in once the IC feed piping is in the car.
~Intercooler piping will get a ceramic coating to match the headers once the final fitting and locationsare set. (After the IC plumbing was finished and coated, the pieces were powdercoated white.)
~Head metal work has been completed; ceramic coating finished at Calico, using their CT-2 compound.
Head has been flowbenched. numbers, taken at 28" of water:
With runners No runners
Intake Exhaust Intake Exhaust
@ 70% of max lift of 10.50 mm (7.34 mm): 232 213 241 213
@ measured max lift of 10.50 mm: 260 253 277 253
These numbers are from the flowbench worksheet, and represent the per-hole averages. Variance was ~1%, so what's here is a close approximation. I have a copy of the actual flowbench sheet somewhere in the banker's box of documentaton on the motor.
Based on these values, Paul thinks about 450hp NA is achievable with proper cam setup and timing. Not bad . . .around 1.85 hp/ cubic inch, or 113 hp/liter.
Using the formula, square root of (atmo + boost), expressed in bar, or ( 1.0 +1.02) = 1.422 x estimated NA output, the estimate goes to 639 flywheel hp. This is substantially above the 600 fwhp target I established with EB when we made the decision to go forced induction. Paul thinks we can go upwards towards 18 psi boost, given internal component strength, and if the head gasket stays together. He is adamant that 700+ is attainable with proper tuning, cam profiles, and the Electromotive TEC-3r. I am a bit skeptical, and certainly don't want to risk the motor to get to these levels.
Sidebar: The GT-35 turbo was already in hand and the exhaust system fabricated when the flowbench testing was done, so we were locked into those system elements. What we didn't know at this point was the effect the Ve values would have on power delivery. We had been using a good, but far less sophisticated, turbo sizing program and had made guesses as to Ve and BSFC. As it turned out, the older program hinted that a T66 or T72 might be the correct size turbo; this model didn't contain any of the GT series compressor maps. We were certain the Ve would be somewhere above 110%. On that basis, the GT-35 was probably going to be close to its limits, but OK. Not dissing anyone here, but Ray Hall's advice, as good as it was, came up a bit short in the final analysis. The events that followed -- better design software, the flowbench results, the exhaust system efficiencies -- resulted in revising the turbo sizing. It's called R&D . . .
~About the only shortcoming in the head is a case can be made for some narrowing of the valve seats, but under my conditions of engineering for reliability and longevity, we are going to leave them alone. The valve faces protrude slightly, maybe a millimeter, into the combustion chambers, thus the seats have adequate material to support them (good thing).
The next few paragraphs are from a thread regarding head work that gos posted in the summer of 2005. I edited this to smooth things out and cut down a bit on the verbiage. My respects to Chris Graff and Todd D.
KH:
"Good stuff, Todd. [refers to TCD's work on the M30B34 head mentioned earlier.] FWIW, I looked at the bench numbers for the head on Lucifer's Hammer. Keep in mind I am running the 38.5 mm intake and 32.5 mm exhaust valves from the Euro S38B38 motor. 5-angle grind, bowls cleaned up, valves unshrouded. The measured lift on this head is .413," or 10.49 mm. Stock specs call for 11 mm. At 28" of water, I am getting 277 cfm on the intakes, and 253 cfm on the exhaust.
The variation cylinder-to-cylinder highest-to-lowest is less than 1%. This may be due to operator error, as the intake port volumes are virtually identical. #5 is off by around 1 cc, or about 8/10 of 1%. #5 is right in the middle on flow volumes,BTW. The exhaust port volumes are identical within 1/10 cc. Chris Graff might be able to weigh in here, but I suspect that boundary layers and laminar flow at the sides of the tract play a significant part in limiting flow rates, as well as overall runner length. The tracts were left relatively rough, not polished smooth, to induce turbulence. I'm aware that this level of tuning begins to move into the "black arts" and how much is "just enough" is what makes some guys geniuses."
From Chris Graff:
"In looking at what might (or might not) be feasible with taking a ///M motor (or an M30) to higher output levels, I think you're correct in identifying the valvetrain as the limiting factor...it's not the valvetrain exactly; on an M30 it's primarily the intake port tract. There are significant losses associated with the relatively long runners and the path the air takes to the valve. The valve itself is pretty large, the curtain area of the intake valve (47 mm) is roughly equivalent to that of the S50 (2 x 33 mm valves).
The S38 however, not only has larger intake valves (37 mm or 38.5 mm), but it has a much different intake port design in the head, and different runners, which really help. The S38's intake tracts are more or less cylindrical, and do not have a great deal of surrounding material that can be removed. BMW engineered this thing pretty close to the wire. On the B35 head, the line of sight and intake port cross-sectional area is much better than on the B34 and older heads. Also the combustion chamber shape is much different, and allows for significantly more scavenging.
You'll notice the design of the M50/S50 family is derived from the S38, but with changes for higher production (volume) manufacturing. But, the head is much flatter (lower). In fact, an M50/S50 head is VERY wide. Much wider than the S38 You ever have trouble changing plugs on an S38 or S14 as opposed to the M50/S50 ???) They don't leave a whole lot of room between the cams on an S38.
The reason they needed to do that on the M50/S50 was the push to lower the height of the engine, overall, by angling the intake valve and the initial intake tract over, which in effect lowers the intake runners, which is then able to lower the overall height of the engine. Notice also, that in the E34, the M50 sits at a 30 degree angle versus the comparable M20 which is canted over at 20 degrees."
More from Chris:
"Well, actually, at full flow rate, the flow through the intake tract is in the turbulent boundary layer Reynolds number regime. The reason why the walls are left just very slightly rough is (in theory) to keep the turbulent boundary layer really close to the wall, in order to keep overall flow losses through the tract to a minimum, not to INDUCE a turbulent boundary layer per se (you already have one). Boundary layers are governed by Reynolds number; boundary layer behavior is another ball game altogether, and I'd have to dig out my aero texts to find out what those are.
Granted we're not talking much here, but every bit counts. As for PERFECTLY smooth walls, there is some debate on that, and basically without doing a comprehensive CFD to find out the root causes of why something is better than the other at one point or another (say right by the valve as opposed to around the intake plenum), it's difficult to say. The problem is coupled by the flow in the intake tract around the intake valve. It gets very complex REALLY quickly. But as a basic rule, you're limited in flow at any point by the air going sonic. You're choking the flow at that point. And locating that point exactly, and knowing what to do exactly to alleviate that problem is very intensive; that requires a lot of analysis and geometric information. The black arts are for those engine people with enough experience who have found out what needs to be done purely from trying stuff out for years on particular engines.
The massive computing power in today's computers has enabled engineers at manufacturers to be able to gain a lot of insight and improve their products. I mean, take a look at, for example, the massive power outputs of today's modern engines, which are cleaner, more efficient, and more reliable than engines of 20 years ago. The US E28 M5 came with 256 hp and 243 ft-lb of torque in 3.5 liters. Today we have 333 bhp and 262 ft-lb out of 3.25 liters and 10 x better emissions, with slightly better fuel economy. And I won't even start to compare the Japanese and American cars which have seen even BETTER gains (which makes sense since they started at a lower level)."
1/12/05 More issues with SCE. They are telling Paul it will be at least another two weeks before we see anything.
This is now turning into a major worry.
Paul is running into some problems in shaping the turbo heat shield, and is evaluating what the alternatives might be. I suggested he contact John Craig at Limit Engineering in Lake Havasu and see if John has any answers.
1/22/05 Paul has contacted VAC in Philly to order a set of their adjustable cam sprockets.
Given the problems we are having with SCE, we looked at the specs on the S38B38 Euro M5 head gasket. Bore is 94.6 m, so this looks promising if SCE continues to delay. Paul will check with his BMW sources as to availability; Maximillian can get them, price is around $173. Everything is coming together, but can't do a whole lot more until we have the head gasket in hand.
1/25/05 KH has a conversation with the troops at Bimmerhaus. Advised that Paul is very close to having the motor done, absent the gasket running sore. Mark wants to start work on the fuel system in the next couple of days; Aeromotive fuel system components came through from Summit, so we are more or less good to go.
Head gasket. Screw SCE.
S38B38 gasket ordered through Paul's sources. Paul will contact SCE and cancel the trick gasket order and have them send the template gasket back to us. S38B38 gasket will be here next week. Ditto cam sprockets from VAC. Turbo heatshield is going to be fabbed from large-diameter stainless steel tubing, mandrel-bent to needed radius, then sectioned and TIG-welded. Any fine tuning will get done on an English wheel. Once parts are here, Paul thinks 3-4 days to fully assemble the motor.
2/7/05 Paul has been working on setting up the cam timing with the adjustable sprockets. He is considering reworking the locating pins and slots to enable easier adjustment of the cams with the engine installed; present setup is awkward and time-consuming.
He needs info on the dimensions and weight of the Aeromotive fuel pressure regulator in order to determine how it will interface with the fuel rail. Given it's size and weight, it's going to need to be mounted on a separate bracket rather than on the rail proper.
2/9/05 The essentials on the Aeromotive fuel system are installed, but much work needs to be done replacing the feed and return lines as well as reworking on the fuel pickup and return into the tank. The fuel system is proving to be a far more complex undertaking than we had envisioned.
We have hit a potential snag with the intake cam. Seems the Schrick 272 cam has a smaller base circle diameter: 1.250"; the stock 248 exhaust cam is 1.314" This means we need to get a set of .060" lash caps for the intake valves. Available from VAC or Ferrea (the custom valve guys). Another delay.
2/18/05 Lash caps are in from Ferrea, however they are .070," not .060." This means the caps will get blanchard ground to the correct height. Machining will be quicker than returning the parts to Ferrea and getting a new set. Hzzzt.
More discussion regarding the cams. Paul is still working on the adjustable sprocket installation/setting method.
He thinks the cam combination, 272 intake and 248 exhaust will work, but limits the motor's power. Going to the S38B36 264 degree cam will reduce the overlap, but the solution doesn't lie simply in increased duration. What we most likely need is a longer "dwell" period at or near full lift -- with the cam lobe having more of a "flat top" profile. Ideally, the intake opening ramp needs to give a rapid valve opening, transition to the extended full-open segment, then roll off to a relatively gradual closing. The exhaust cam would be a reverse on the ramps, but would also have an extended full-open period.
Overlap could be reduced to perhaps 6-8 degrees.
Paul has some contacts in the cam grinding world and is looking into what he can find out about doing experimental grinds, costs, delivery, etc.
He thinks the profiles such as what he's contemplating do exist, having been done for small block drag motors. What might happen is a prospective profile is on someone's CAD/CNC disc. This particular profile could be suitably modified to the needed S38 dimensions. Paul has done a fair amount of working with experimental cam profiles related to his drag engine building, so I am definitely all ears.
He thinks the Schrick 272 is probably going to give us 40-45 fwhp over stock. I didn't ask if this was NA or FI, however. A more efficient longer max lift profile might double that number.
Nothing comes easy; in order to validate what profile(s) and duration(s) might work best, it will be necessary to grind several cams then dyno accordingly.
This approach, while I agree with the method, has the downside of getting stupid expensive, even if the cam grinding shop is willing to give Paul a break on the raw billets, writing the CAD/CNC application files and the machine time. Sign over the intellectual property as well and the number is still frightening. The asymmetric ramp concept is not new, but applying it to my motor may be something original; we aren't aware of this being tried on a BMW DOHC motor anywhere.
In any case, for the time being, we'll go with 272 intake and 248 exhaust and manage overlap and exhaust timing via the adjustable sprockets.
2/25/05 Paul added O2 sensor bungs in the collector tubes, these in addition to the one for the main O2 sensor ahead of the cats. These bungs are for placement of additional O2 sensors for exxhaust gas analytics. FWIW, the S38 has individual bungs located just outboard of the exhaust ports on the US exhaust manifold. These are used for fine-tuning on individual cylinders -- a feature which the TEC-3r has. The idea here is to determine if an individual hole runs too rich or too lean, then adjust injector pulse width accordingly. Trying to rework the six individual tubes in my case has access problems; Paul thinks we can get a reasonably good idea by looking at the exhaust in groups of three. Not perfect, but certainly not a bad idea.
Probably going to leave the template head gasket at SCE for the time being. Paul says they may need it for reference purposes in making the wedge-ring design originally discussed. While the S38B38 gasket will be OK, down the road we may yet need the SCE design. Time will tell.
Paul is getting antsy about the capacity of the stock radiator to handle the motor's heat; thinks it will be less-than-marginal. This is due to several factors:
One, a motor is a heat pump, and more hp = more heat.
Two, the placement of the intercooler takes up a fair amount of the frontal area exposed to incoming air. The air which as passed across (not through) the IC has picked up a fair number of BTUs. Thus the radiator is that much less effective in reducing coolant temperatures.
Three, the size and placement of the IC meant removing the stock auxiliary fan, which is a pretty effective unit. So when moving at highway speeds, we get ram-air cooling; in traffic or at low speeds, forced cooling air is much reduced or absent.
Four, the radiator in there is the original unit, so about 17 years old and parts corrode and get brittle.
So I need to find an auxiliary cooling fan that is small enough to fit and can move effective air volume, and see what's out there for a replacement radiator.
Jumping ahead a few months, we did find a 7" diameter SPAL fan that would fit. It moves about 400 cfm, vs. the couple of thousand cfm from the stock fan. Some help in traffic, but not a whole lot. Note to self: when in traffic, the A/C gets turned off. The car is very much prone to push towards overheating when stuck in stop-and-go in combination with summer temperatures.
This past fall (2006) I did have a conversation with the good people at Ron Davis Radiators in Phoenix.
A number of findings.
One, they are familiar with the radiators in E28s. They have patterns in stock, and can fabricate a dimensioinally identical drop-in for an M5.
Two, the RDR replacement is a bar-and-plate design, with all-aluminum construction. Hose fitting points are machined out of billet stock.
Three, while dimensionally identical, the replacement would have considerably more cooling area due to the bar-and-plate design.
Four, depending on how much clearance there is between the back of the radiator and needed clearance with the fan, the body of the radiator could be made deeper front-to-back, thus adding additional cooling capacity. Measure with the present unit in place by adding cardboard shims. Pull the radiator with the shims attached. Send to RDR.
Five, RDR estimates six to eight weeks turnaround, given their current book of business. Best part, price, whether stock dimensions or upsized, will be less than the numbers in realoem.com. R&R the radiator is a couple of hours job.
Lucifer's Hammer is off the road for the winter, sitting in the garage minus insurance. Dave Stackhouse at Bimmerhaus has access to another M5 and has a spare M5 radiator. Will do the measurements and send the spare down to RDR this coming February. Should work out just about right for the coming year.
3/24/05 Paul delivers the motor to Bimmerhaus. Several small items, e.g., turbo heatshield, yet to come; being powdercoated. Council of war with Paul, Mark and Dave. We want to get this farquar done by May 1 to complete the interior, suspension, drivetrain work, break the thing in and work out the kinks before going to 5erFest in Greenville at the end of May. Mark is a bit worried about his time availability; he is penciled in for the TEC-3r installation and tuning, but has a number of added responsibilities on his plate at the shop. Doing the TEC-3 install is one job where he doesn't need a bunch of distractions.
3/25/05 Clutch MC, slave and high-pressure hose all replaced. Car goes to the detailers tomorrow or Thursday for a thorough cleaning of the engine bay.
KH contacts Tony Werth at Premier Coatings in Denver regarding their capability on doing a high-temperature coating on the exhaust system. Tony sez they can do a zirconium porcelain treatment, both interior and exterior. Exterior finish can only bne done in satin black, however. "Is that a problem?" "Is Paris Hilton virgin?"
The only constraint may be the physical size of the pieces going into the oven at Premier. Turnaround no more than a week or so.
Paul, having dropped off the motor, also leaves me with three 36-exposure rolls of 35 mm photos. The film, which he hadn't gotten developed, covers the head, it's assembly and flow benching, machining on a number of pieces, and the final assembly of the intake components. So that evening, I stop by King Soopers on the way home to have the film developed, a couple of sets of prints made, and get a digital disc burned. Paul was certain the pictures would be very good. He used slow ASA speed Kodacolor negative film in his very trick Canon SLR. The idea here was to get pictures which would be used in a planned sales brochure for Imagineering (the name of Paul's business).
It's late, but Customer Cervix at KS tells me to come back in an hour or so for the photos. I do so. They hand me three strips of exposed film. The witless krunt opened the film cartridges before placing them in the developing machine.
The film is totally lost. Then they want to charge me for the developing. Faaaaack.
Next morning I call Paul with the wonderful news. He is terminally pissed to put it mildly. Still hasn't let me forget it, either.
The absence of these photos has been a real downer, as it gutted a lot of interesting material for this narrative. In and of itself, it didn't cover anything someone who has built an engine hasn't seen before, but it leaves a real gap in the record.
The next chapter begins with getting the motor into the car, and dealing with doing stuff on the body itself: suspension and brakes, driveline and the interior.
Since somebody PM'ed me about how one can get Volumetric Efficiency (Ve) numbers, I'll throw out some more formulas.
An item worth noting is the essential part played by getting the actual airflow volumes for the head via a flowbench. Without knowing that value, any calculations are pure guesswork, though the turbo models can make pretty good inferences, given sufficient other data elements.
Head flows and Ve calculations:
Ve = (VAF x ES x C)/(D x rpm)
where
Ve = volumetric efficiency, or the amount of intake charge entering the cylinder on the intake stroke. The percentage is relative to the maximum cylinder plus combustion chamber volume.
Values above 100% indicate the intake charge is compressed above normal atmospheric pressure.
VAF = volumetric airflow rate.
ES = engine stroke (2 or 4).
C = conversion factor--cubic inches to cubic feet (use 1728)
D = displacement in cubic inches.
rpm = engine speed where the volumetric flow was measured, or for which the Ve is being determined. In this case, 5700 is what has shown up on the dyno as the top of my HP curve.
Knowing the head flowed 277 cfm @ 28" of water on a Superflow 600 flowbench, we get (277 x 6912) divided by (243 x 5700). 1,914,624 / 1,385,100 = Ve. 1.382 = Ve. Which checks pretty closely with the not2fast model displayed value of 139%.
For comparative purposes, IIRC, T.C.D. had an M30B34 head flowbenched some time in the summer of 2005. His post stated this cyl head had received a "Stage 3" porting job from a shop in TN that is used by Turner Motorsports. (possibly Memphis Motorworks?) This is the head that was on Todd's 745i that produced 331rwhp/316rwtq. Presumably this head flows pretty well. At .4" of lift it flowed 169 cfm in the intake valve and 134 cfm on the exhaust valve. If so, then Ve = (169 x6912)/(209.3 x 6700 ). [3430 cc = 209.3 cubic inches; I'm guessing at the redline here.] 1,168,128/1,402,310 = .833 Ve. This is very respectable. I believe the B35 flows a lot better--in the 200 to 210 cfm range. If that's the case, the Ve would probably go north of 100%.
Back to the story.
Mid-December, 2004. Paul, sounding immensely relieved, sez we're at engine mock-up. So it's off to Misery for a look-see. Best part of the trip across western Kansas was watching a blood-red Ford GT make a valiant attempt at breaking the sound barrier.Thass' nice. Should attract the KSP and leave me alone.
12/17/04 Need to figure on doing a UUC Shortshifter when we start putting things back together; easier to do with the car on the lift and the tranny out. The rubber shift lever boot , etc. is all dried out, cracked so all that stuff needs replacement as well.
12/19/04 At Paul's shop. The motor resideth on the engine stand in all its earthly glory.
By no means complete, but it feels like a major milestone has been reached, and completion is now a question of time. What follows is a listing, incomplete, and in no particular order, of items noted that are going to need attention.~Exhaust headers being problematical as to proper alignment at the ports. Nothing weird, but some of the flange-to-head surfaces may need a final grinding to bring the mating surfaces to dead flat. Will need to enlarge the holes in the mounting flanges, as there is a little lateral pressure on the exhaust studs. This is a recipe for shearing the studs when the motor is at operating temperature.
~Playing around with the tubing alignment at the merge collectors helped a lot in getting the tubing joints to line up correctly. I want the header R&R to be a slam dunk. If the motor won't install with the headers in place, we will have a real chore on our hands with the motor in the bay.
~Except for die grinding, cleanup and Glyptal coating of the water pump flange, the short block is just about done.
~Block will get one more coat of paint; a few chipped and scratched areas from assembly.
~Crank is in place; oil pump going in this coming week, along with the piston assemblies. The Carillo rods are artistry in steel.
~Knock sensor installed; doesn't appear that there will be any problem in accessing it past the headers. The knock sensor lives in the former block drain location, so when it comes time to change or drain coolant, pull the knock sensor. Potential leakage controlled with a crush washer under the sensor. The location should give us good sensitivity to detonation noises and vibration.
~Plenum-to-turbine clearance measured at a bit under 1/4 inch. Plenum marked for a half-moon cut about 4 inches across. Paul will shape the concave filler plate and do the turbine heatshield design and fabrication. Paul's powdercoater is confident he can do the ///M tricolor after the welding is done on the plenum.
~A number of parts will need to be repowdercoated. Pieces went down in a couple of different lots, and the colors used weren't identical. The workmanship is good, but when pieces from different batches are put side-by-side, the color variation is obvious. Bare aluminum pieces -- throttle bodies, cam tray to name two -- will get a clear ceramic coating to control potential corrosion and allow for ease of removing engine crud.
~Head gasket is waiting for tooling to be completed at SCE. SCE thinks that will be around Jan 10, with delivery of gasket thereafter.
~Left a list of needed Aeromotive fuel system parts with Paul. Aeromotive is located in KC; he will talk with them directly and see if we can get any better pricing than by going through Summit Racing. Going to be enough $$ that it's worth trying.
~Paul is going to call Marren Fuel Injection to obtain a length of their ribbed fuel rail aluminum extrusion. Setting up the raw stock for proper injector alignment, attachment points, etc. is going to take some time on the mill. Paul's looking forward to this; a chance to get a bit creative . . .The Bridgeport Artisan. . . .
~The mild steel turbo flange will go back with me to Colo. Mark has a friend who can pantograph the part in a CNC mill, replicating the piece in 321 stainless. Once done, it can be TIGed onto the 304 stanless downpipe (which is still in Colo).
~Diverter valve will get located and TIGed in once the IC feed piping is in the car.
~Intercooler piping will get a ceramic coating to match the headers once the final fitting and locationsare set. (After the IC plumbing was finished and coated, the pieces were powdercoated white.)
~Head metal work has been completed; ceramic coating finished at Calico, using their CT-2 compound.
Head has been flowbenched. numbers, taken at 28" of water:
With runners No runners
Intake Exhaust Intake Exhaust
@ 70% of max lift of 10.50 mm (7.34 mm): 232 213 241 213
@ measured max lift of 10.50 mm: 260 253 277 253
These numbers are from the flowbench worksheet, and represent the per-hole averages. Variance was ~1%, so what's here is a close approximation. I have a copy of the actual flowbench sheet somewhere in the banker's box of documentaton on the motor.
Based on these values, Paul thinks about 450hp NA is achievable with proper cam setup and timing. Not bad . . .around 1.85 hp/ cubic inch, or 113 hp/liter.
Using the formula, square root of (atmo + boost), expressed in bar, or ( 1.0 +1.02) = 1.422 x estimated NA output, the estimate goes to 639 flywheel hp. This is substantially above the 600 fwhp target I established with EB when we made the decision to go forced induction. Paul thinks we can go upwards towards 18 psi boost, given internal component strength, and if the head gasket stays together. He is adamant that 700+ is attainable with proper tuning, cam profiles, and the Electromotive TEC-3r. I am a bit skeptical, and certainly don't want to risk the motor to get to these levels.
Sidebar: The GT-35 turbo was already in hand and the exhaust system fabricated when the flowbench testing was done, so we were locked into those system elements. What we didn't know at this point was the effect the Ve values would have on power delivery. We had been using a good, but far less sophisticated, turbo sizing program and had made guesses as to Ve and BSFC. As it turned out, the older program hinted that a T66 or T72 might be the correct size turbo; this model didn't contain any of the GT series compressor maps. We were certain the Ve would be somewhere above 110%. On that basis, the GT-35 was probably going to be close to its limits, but OK. Not dissing anyone here, but Ray Hall's advice, as good as it was, came up a bit short in the final analysis. The events that followed -- better design software, the flowbench results, the exhaust system efficiencies -- resulted in revising the turbo sizing. It's called R&D . . .
~About the only shortcoming in the head is a case can be made for some narrowing of the valve seats, but under my conditions of engineering for reliability and longevity, we are going to leave them alone. The valve faces protrude slightly, maybe a millimeter, into the combustion chambers, thus the seats have adequate material to support them (good thing).
The next few paragraphs are from a thread regarding head work that gos posted in the summer of 2005. I edited this to smooth things out and cut down a bit on the verbiage. My respects to Chris Graff and Todd D.
KH:
"Good stuff, Todd. [refers to TCD's work on the M30B34 head mentioned earlier.] FWIW, I looked at the bench numbers for the head on Lucifer's Hammer. Keep in mind I am running the 38.5 mm intake and 32.5 mm exhaust valves from the Euro S38B38 motor. 5-angle grind, bowls cleaned up, valves unshrouded. The measured lift on this head is .413," or 10.49 mm. Stock specs call for 11 mm. At 28" of water, I am getting 277 cfm on the intakes, and 253 cfm on the exhaust.
The variation cylinder-to-cylinder highest-to-lowest is less than 1%. This may be due to operator error, as the intake port volumes are virtually identical. #5 is off by around 1 cc, or about 8/10 of 1%. #5 is right in the middle on flow volumes,BTW. The exhaust port volumes are identical within 1/10 cc. Chris Graff might be able to weigh in here, but I suspect that boundary layers and laminar flow at the sides of the tract play a significant part in limiting flow rates, as well as overall runner length. The tracts were left relatively rough, not polished smooth, to induce turbulence. I'm aware that this level of tuning begins to move into the "black arts" and how much is "just enough" is what makes some guys geniuses."
From Chris Graff:
"In looking at what might (or might not) be feasible with taking a ///M motor (or an M30) to higher output levels, I think you're correct in identifying the valvetrain as the limiting factor...it's not the valvetrain exactly; on an M30 it's primarily the intake port tract. There are significant losses associated with the relatively long runners and the path the air takes to the valve. The valve itself is pretty large, the curtain area of the intake valve (47 mm) is roughly equivalent to that of the S50 (2 x 33 mm valves).
The S38 however, not only has larger intake valves (37 mm or 38.5 mm), but it has a much different intake port design in the head, and different runners, which really help. The S38's intake tracts are more or less cylindrical, and do not have a great deal of surrounding material that can be removed. BMW engineered this thing pretty close to the wire. On the B35 head, the line of sight and intake port cross-sectional area is much better than on the B34 and older heads. Also the combustion chamber shape is much different, and allows for significantly more scavenging.
You'll notice the design of the M50/S50 family is derived from the S38, but with changes for higher production (volume) manufacturing. But, the head is much flatter (lower). In fact, an M50/S50 head is VERY wide. Much wider than the S38 You ever have trouble changing plugs on an S38 or S14 as opposed to the M50/S50 ???) They don't leave a whole lot of room between the cams on an S38.
The reason they needed to do that on the M50/S50 was the push to lower the height of the engine, overall, by angling the intake valve and the initial intake tract over, which in effect lowers the intake runners, which is then able to lower the overall height of the engine. Notice also, that in the E34, the M50 sits at a 30 degree angle versus the comparable M20 which is canted over at 20 degrees."
More from Chris:
"Well, actually, at full flow rate, the flow through the intake tract is in the turbulent boundary layer Reynolds number regime. The reason why the walls are left just very slightly rough is (in theory) to keep the turbulent boundary layer really close to the wall, in order to keep overall flow losses through the tract to a minimum, not to INDUCE a turbulent boundary layer per se (you already have one). Boundary layers are governed by Reynolds number; boundary layer behavior is another ball game altogether, and I'd have to dig out my aero texts to find out what those are.
Granted we're not talking much here, but every bit counts. As for PERFECTLY smooth walls, there is some debate on that, and basically without doing a comprehensive CFD to find out the root causes of why something is better than the other at one point or another (say right by the valve as opposed to around the intake plenum), it's difficult to say. The problem is coupled by the flow in the intake tract around the intake valve. It gets very complex REALLY quickly. But as a basic rule, you're limited in flow at any point by the air going sonic. You're choking the flow at that point. And locating that point exactly, and knowing what to do exactly to alleviate that problem is very intensive; that requires a lot of analysis and geometric information. The black arts are for those engine people with enough experience who have found out what needs to be done purely from trying stuff out for years on particular engines.
The massive computing power in today's computers has enabled engineers at manufacturers to be able to gain a lot of insight and improve their products. I mean, take a look at, for example, the massive power outputs of today's modern engines, which are cleaner, more efficient, and more reliable than engines of 20 years ago. The US E28 M5 came with 256 hp and 243 ft-lb of torque in 3.5 liters. Today we have 333 bhp and 262 ft-lb out of 3.25 liters and 10 x better emissions, with slightly better fuel economy. And I won't even start to compare the Japanese and American cars which have seen even BETTER gains (which makes sense since they started at a lower level)."
1/12/05 More issues with SCE. They are telling Paul it will be at least another two weeks before we see anything.
This is now turning into a major worry.Paul is running into some problems in shaping the turbo heat shield, and is evaluating what the alternatives might be. I suggested he contact John Craig at Limit Engineering in Lake Havasu and see if John has any answers.
1/22/05 Paul has contacted VAC in Philly to order a set of their adjustable cam sprockets.
Given the problems we are having with SCE, we looked at the specs on the S38B38 Euro M5 head gasket. Bore is 94.6 m, so this looks promising if SCE continues to delay. Paul will check with his BMW sources as to availability; Maximillian can get them, price is around $173. Everything is coming together, but can't do a whole lot more until we have the head gasket in hand.
1/25/05 KH has a conversation with the troops at Bimmerhaus. Advised that Paul is very close to having the motor done, absent the gasket running sore. Mark wants to start work on the fuel system in the next couple of days; Aeromotive fuel system components came through from Summit, so we are more or less good to go.
Head gasket. Screw SCE.
S38B38 gasket ordered through Paul's sources. Paul will contact SCE and cancel the trick gasket order and have them send the template gasket back to us. S38B38 gasket will be here next week. Ditto cam sprockets from VAC. Turbo heatshield is going to be fabbed from large-diameter stainless steel tubing, mandrel-bent to needed radius, then sectioned and TIG-welded. Any fine tuning will get done on an English wheel. Once parts are here, Paul thinks 3-4 days to fully assemble the motor.2/7/05 Paul has been working on setting up the cam timing with the adjustable sprockets. He is considering reworking the locating pins and slots to enable easier adjustment of the cams with the engine installed; present setup is awkward and time-consuming.
He needs info on the dimensions and weight of the Aeromotive fuel pressure regulator in order to determine how it will interface with the fuel rail. Given it's size and weight, it's going to need to be mounted on a separate bracket rather than on the rail proper.
2/9/05 The essentials on the Aeromotive fuel system are installed, but much work needs to be done replacing the feed and return lines as well as reworking on the fuel pickup and return into the tank. The fuel system is proving to be a far more complex undertaking than we had envisioned.
We have hit a potential snag with the intake cam. Seems the Schrick 272 cam has a smaller base circle diameter: 1.250"; the stock 248 exhaust cam is 1.314" This means we need to get a set of .060" lash caps for the intake valves. Available from VAC or Ferrea (the custom valve guys). Another delay.
2/18/05 Lash caps are in from Ferrea, however they are .070," not .060." This means the caps will get blanchard ground to the correct height. Machining will be quicker than returning the parts to Ferrea and getting a new set. Hzzzt.
More discussion regarding the cams. Paul is still working on the adjustable sprocket installation/setting method.
He thinks the cam combination, 272 intake and 248 exhaust will work, but limits the motor's power. Going to the S38B36 264 degree cam will reduce the overlap, but the solution doesn't lie simply in increased duration. What we most likely need is a longer "dwell" period at or near full lift -- with the cam lobe having more of a "flat top" profile. Ideally, the intake opening ramp needs to give a rapid valve opening, transition to the extended full-open segment, then roll off to a relatively gradual closing. The exhaust cam would be a reverse on the ramps, but would also have an extended full-open period.
Overlap could be reduced to perhaps 6-8 degrees.
Paul has some contacts in the cam grinding world and is looking into what he can find out about doing experimental grinds, costs, delivery, etc.
He thinks the profiles such as what he's contemplating do exist, having been done for small block drag motors. What might happen is a prospective profile is on someone's CAD/CNC disc. This particular profile could be suitably modified to the needed S38 dimensions. Paul has done a fair amount of working with experimental cam profiles related to his drag engine building, so I am definitely all ears.
He thinks the Schrick 272 is probably going to give us 40-45 fwhp over stock. I didn't ask if this was NA or FI, however. A more efficient longer max lift profile might double that number.
Nothing comes easy; in order to validate what profile(s) and duration(s) might work best, it will be necessary to grind several cams then dyno accordingly.
This approach, while I agree with the method, has the downside of getting stupid expensive, even if the cam grinding shop is willing to give Paul a break on the raw billets, writing the CAD/CNC application files and the machine time. Sign over the intellectual property as well and the number is still frightening. The asymmetric ramp concept is not new, but applying it to my motor may be something original; we aren't aware of this being tried on a BMW DOHC motor anywhere.
In any case, for the time being, we'll go with 272 intake and 248 exhaust and manage overlap and exhaust timing via the adjustable sprockets.
2/25/05 Paul added O2 sensor bungs in the collector tubes, these in addition to the one for the main O2 sensor ahead of the cats. These bungs are for placement of additional O2 sensors for exxhaust gas analytics. FWIW, the S38 has individual bungs located just outboard of the exhaust ports on the US exhaust manifold. These are used for fine-tuning on individual cylinders -- a feature which the TEC-3r has. The idea here is to determine if an individual hole runs too rich or too lean, then adjust injector pulse width accordingly. Trying to rework the six individual tubes in my case has access problems; Paul thinks we can get a reasonably good idea by looking at the exhaust in groups of three. Not perfect, but certainly not a bad idea.
Probably going to leave the template head gasket at SCE for the time being. Paul says they may need it for reference purposes in making the wedge-ring design originally discussed. While the S38B38 gasket will be OK, down the road we may yet need the SCE design. Time will tell.
Paul is getting antsy about the capacity of the stock radiator to handle the motor's heat; thinks it will be less-than-marginal. This is due to several factors:
One, a motor is a heat pump, and more hp = more heat.
Two, the placement of the intercooler takes up a fair amount of the frontal area exposed to incoming air. The air which as passed across (not through) the IC has picked up a fair number of BTUs. Thus the radiator is that much less effective in reducing coolant temperatures.
Three, the size and placement of the IC meant removing the stock auxiliary fan, which is a pretty effective unit. So when moving at highway speeds, we get ram-air cooling; in traffic or at low speeds, forced cooling air is much reduced or absent.
Four, the radiator in there is the original unit, so about 17 years old and parts corrode and get brittle.
So I need to find an auxiliary cooling fan that is small enough to fit and can move effective air volume, and see what's out there for a replacement radiator.
Jumping ahead a few months, we did find a 7" diameter SPAL fan that would fit. It moves about 400 cfm, vs. the couple of thousand cfm from the stock fan. Some help in traffic, but not a whole lot. Note to self: when in traffic, the A/C gets turned off. The car is very much prone to push towards overheating when stuck in stop-and-go in combination with summer temperatures.
This past fall (2006) I did have a conversation with the good people at Ron Davis Radiators in Phoenix.
A number of findings.
One, they are familiar with the radiators in E28s. They have patterns in stock, and can fabricate a dimensioinally identical drop-in for an M5.
Two, the RDR replacement is a bar-and-plate design, with all-aluminum construction. Hose fitting points are machined out of billet stock.
Three, while dimensionally identical, the replacement would have considerably more cooling area due to the bar-and-plate design.
Four, depending on how much clearance there is between the back of the radiator and needed clearance with the fan, the body of the radiator could be made deeper front-to-back, thus adding additional cooling capacity. Measure with the present unit in place by adding cardboard shims. Pull the radiator with the shims attached. Send to RDR.
Five, RDR estimates six to eight weeks turnaround, given their current book of business. Best part, price, whether stock dimensions or upsized, will be less than the numbers in realoem.com. R&R the radiator is a couple of hours job.
Lucifer's Hammer is off the road for the winter, sitting in the garage minus insurance. Dave Stackhouse at Bimmerhaus has access to another M5 and has a spare M5 radiator. Will do the measurements and send the spare down to RDR this coming February. Should work out just about right for the coming year.
3/24/05 Paul delivers the motor to Bimmerhaus. Several small items, e.g., turbo heatshield, yet to come; being powdercoated. Council of war with Paul, Mark and Dave. We want to get this farquar done by May 1 to complete the interior, suspension, drivetrain work, break the thing in and work out the kinks before going to 5erFest in Greenville at the end of May. Mark is a bit worried about his time availability; he is penciled in for the TEC-3r installation and tuning, but has a number of added responsibilities on his plate at the shop. Doing the TEC-3 install is one job where he doesn't need a bunch of distractions.
3/25/05 Clutch MC, slave and high-pressure hose all replaced. Car goes to the detailers tomorrow or Thursday for a thorough cleaning of the engine bay.
KH contacts Tony Werth at Premier Coatings in Denver regarding their capability on doing a high-temperature coating on the exhaust system. Tony sez they can do a zirconium porcelain treatment, both interior and exterior. Exterior finish can only bne done in satin black, however. "Is that a problem?" "Is Paris Hilton virgin?"
The only constraint may be the physical size of the pieces going into the oven at Premier. Turnaround no more than a week or so.Paul, having dropped off the motor, also leaves me with three 36-exposure rolls of 35 mm photos. The film, which he hadn't gotten developed, covers the head, it's assembly and flow benching, machining on a number of pieces, and the final assembly of the intake components. So that evening, I stop by King Soopers on the way home to have the film developed, a couple of sets of prints made, and get a digital disc burned. Paul was certain the pictures would be very good. He used slow ASA speed Kodacolor negative film in his very trick Canon SLR. The idea here was to get pictures which would be used in a planned sales brochure for Imagineering (the name of Paul's business).
It's late, but Customer Cervix at KS tells me to come back in an hour or so for the photos. I do so. They hand me three strips of exposed film. The witless krunt opened the film cartridges before placing them in the developing machine.
The film is totally lost. Then they want to charge me for the developing. Faaaaack.
Next morning I call Paul with the wonderful news. He is terminally pissed to put it mildly. Still hasn't let me forget it, either.
The absence of these photos has been a real downer, as it gutted a lot of interesting material for this narrative. In and of itself, it didn't cover anything someone who has built an engine hasn't seen before, but it leaves a real gap in the record.
The next chapter begins with getting the motor into the car, and dealing with doing stuff on the body itself: suspension and brakes, driveline and the interior.