2.8 Turbo Engine Build
 

I'd like to share this build. Maybe there's something of interest here for you. This has been mainly a DE car since '95, but it does drive to (and back from) the track over the roads.

It's a 2.8 (70.4 x 92 mm), built on a sandcast aluminum case. I had the case bored for 2.8 in the 1980s.

It uses a 964 oil pump and 993 squirters.

http://forums.pelicanparts.com/uploa...1395346342.jpg

http://forums.pelicanparts.com/uploa...1395346426.jpg

• stock 2.4/2.7 crank, crack tested
• shot-peened Porsche rods, raceware rod bolts

http://forums.pelicanparts.com/uploa...1395346668.jpg

• JE pistons running on Porsche 2.7 cylinders with US Chrome Nikasil plating
• Currently 7.0 C/R
• It still has all 24 Dilavar studs!

http://forums.pelicanparts.com/uploa...1395346937.jpg

• 2.4 E heads, chamfer cut, ported to 39 mm, bored for low buck EFI injectors

http://forums.pelicanparts.com/uploa...1395346989.jpg

The cams are not great for this motor, crane p-306, but I love the results! Intake lift is 12.55mm at the valve, intake advertised duration is 306 degrees.

http://forums.pelicanparts.com/uploa...1395347116.jpg

I was thinking of trying some SC turbo cams I have stored.

http://forums.pelicanparts.com/uploa...1395347535.jpg

Nice build, more details please!

^ yes! I love reading about unique builds.

Thanks for the kind words. This one was low-buck start to finish.

So ANDIAL ported the heads and Pacific FI ported the 911E throttles to 40 or 41mm.
I chopped up some aluminum 3" square and 2-1/4" round tubing, 1/4" thick to form the intake manifolds. Thanks to the Evans Brothers in Elburn IL for CNCing the flanges. Tech-Weld in Elburn zipped them up (with some pre-heat).

http://forums.pelicanparts.com/uploa...1395409090.jpg

For the first iteration in 2001 the headers were Bursch 1-5/8 linked to a couple of small Garretts with internal wastegates. The plan was the intakes could be rotated 180 degrees to work with or without intercooler.

http://forums.pelicanparts.com/uploa...1395409131.jpg

In 2006 after some good advice from Midwest Eurosport, we decided to increase the intake square tube to 4" and link the intakes together. So the upper part of the 3" square tube was milled off, leaving a flat base to mount the 4". With the manifold open, it was possible to taper the inside of the runners, and screw 44 IDA trumpets to the flat base.

http://forums.pelicanparts.com/uploa...1395409160.jpg

Then the 4" tube could be screwed on (with prevailing thread fasteners and Loktite). There are six 2" pipes linking the intake to cooler because the air in both is very turbulent, so more pipes is better just here.

http://forums.pelicanparts.com/uploa...1395409201.jpg

http://forums.pelicanparts.com/uploa...1395409212.jpg

Had some spare time and a lot of acetylene, so I chopped the Bursch headers into 40 pieces with a hacksaw and build some shortys that were suprisingly equal-length. About 13" primaries as I recall. Don't look too hard, if the welding doesn't nauseate, then the China wastegates should do the job. Turbos are now T03e hybrid (imported). Funny thing, I work for Honeywell.

http://forums.pelicanparts.com/uploa...1395409261.jpg

http://forums.pelicanparts.com/uploa...1395409279.jpg

Two different oil drains here. The chain box drains worked, but the pipe or hose was so short there was no room to flex and assembly was difficult. The sump drain still works (now with two AN-10 fittings) but if I don't drive the car for a few weeks the oil builds up in the engine and comes out through the turbos, especially if the front end is raised.

http://forums.pelicanparts.com/uploa...1395409304.jpg

http://forums.pelicanparts.com/uploa...1395409317.jpg

saved

Yep... subscribed as well.... very cool!!!!

I think this is the coolest build I've seen yet....

:cool: got to watch this with great interest...
Kozmo

What turbos do you run for the twin configuration? I see integral wastegates in your early pictures.

Thanks for the encouragement. I was afraid this build would be too far "out in the weeds" for the community. Just trying to have some fun without too many $$$.

At first the turbos were Garrett T3 with 50 compressor trim. Tried the .48 and the .63 turbine housings. Was trying to be conservative to begin with but everything turned out too small. The turbos, the fuel pump, and the injectors were too small. Was not happy with the performance.

The second iteration was China T04e hybrids with stage III or IV turbines. They are $149 each! Throw 'em out as soon as they get dirty! Injectors are now Summit 83lb/hr (a little too big). There's no boost until 3rd gear, but then it lifts up the front end in 3rd. I probably don't want boost any sooner for two reasons. One reason it would be too much torque for the 915 trans. Also currently not using 2nd gear on the track (RA and Blackhawk).

So loving the T03e compressors but maybe the turbines should be smaller for good spool. I would like to use KKK but the China parts are great because you can keep switching them until you find what size the engine likes.

I love this build! I've been thinking for years about doing something similar on my 914/6 race car. My enbgine is a 2.7L built on a sand cast case with MFI throttle bodies.

There is a gentleman on this board who also built a similar engine, a 2.5L as I recall, for his 914 a number of years ago. His car was featured as a budget build in Excellence back in the late 90s.

Keep the details coming!

Scott

Wow. You make mention this us a budget build but......

stownsen: It's possible to just 'ease into' the build. Initially, I dropped the C/R and switched to EFI. Just kept driving it that way! I know, that's not feasible with your race car. Then it was simple to adapt the turbos to headers later. On the intake side, I know some have pressurized the MFI stacks. Simple, but not sure how well that works. You would have some nice plenums sticking out of the engine cover! It would look a little bit 908.

quattrorunner: You have a good point. I would not want to know how much was spent over the years. But little by little...

that car is off the charts in coolness factor.

Thanks aschen. Whenever I have a choice A or B, I just pick the one that's cooler. Serious.

Nothing is too far out for this board.
Cool build.
Thanks for posting and sharing. I like when people play around without spending too much.
This is the way to learn as well.
Keep posting :-)

Here's the long version on 4 different injection systems that have been in this car. Probably too much information, but maybe something we can use.

So it started life as a 911 E with that wonderful mechanical injection. The sound with the air cleaner off was unbelievable. People kept asking "what you got in that thing".

http://forums.pelicanparts.com/uploa...1395757813.jpg

It was sad to remove the MFI pump when the 2.4 Mag engine was replaced with Aluminum 2.8L/turbo, but as I understand it is costly to re-calibrate the pump for 2.8L and to have it compensate for the turbo. Has anyone done this? Also there are no mounts on the early Aluminum cases for the MFI pump. I hear there is a fix for that however. Worst of all, the custom turbo intake meant no more great noises.
The MFI pump was replaced with an economical EFI system, the ACCEL DFI/Calmap (now Holley / Mr. Gasket).

http://forums.pelicanparts.com/uploa...1395757842.jpg

http://forums.pelicanparts.com/uploa...1395757896.jpg

It worked best when programmed for synchronous batch fire (once or twice per engine rev). That way, there was naturally less fuel flowing at idle, because there was more time between pulses. It has a pure batch fire mode that just fires every 16mS regardless engine RPM, but that mode doesn't have enough flow range to match requirements for both idle and WOT.
Most of the EFI components came from Kinsler. Got some fuel rail stock, a bypass regulator, an electric high pressure pump, and filters. They also have some extensive write-ups how to design a racing fuel system (EFI, Lucas, Hillborn etc). I drilled the injector bores in the heads to fit EFI injectors.

http://forums.pelicanparts.com/uploa...1395757921.jpg

http://forums.pelicanparts.com/uploa...1395757936.jpg

I didn't intend to spend any time on a dyno, but I felt that mass air flow meters (MAF) were accurate enough to meter the fuel without any dyno tuning. I created a fuel map that worked with MAF, not MAP. The ECU wasn't made for that but it was possible to create the fuel map with some math, spreadsheets etc.

http://forums.pelicanparts.com/uploa...1395757968.jpg

The ACCEL system worked about 7 years before going kaput. I recall using an oscilloscope to confirm that the injector drivers were actually inactive. In other words, it was not just due to a problem with a sensor. Overall, I would say the ACCEL system did the job. It just didn't have the features you expect from a modern system. Anyone else (that wants to admit to) using the ACCEL DFI?

So for the third iteration on the fuel injection I made a custom ECU!

It was another case of more time than money.

I wasn't after any particular feature except simplicity and to not require a dyno tune.

I felt that a MAF sensor could determine fuel requirements over a wide operating range, including boost. No other sensors were used! Not temperature, pressure, or RPM. I felt if there were any "flat spots" that could be addressed if they showed up.

The system actually worked well for a number of years for both road and track use. I never really got a wideband on it but the mixture smelled and sounded right in most cases. Plugs looked OK. One exception was cold startup could be sluggish, so I added a cold start button that would open the injectors as long as the button was pressed.

Details on the home-made ECU follow. Caution, hard-core Electrical Engineering content ahead!

http://forums.pelicanparts.com/uploa...1395938404.jpg

So the ECU circuit board is on the right in the photo above, the "Removable Target Board". The left side of the photo is the part that is used to burn a new program into the target. So this thing is tiny!

And cheap. It was $22 shipped. That includes a software development environment with simulator that targets the entire product range. I bought 4 of them.

The MSP430's are often used in bathroom scales and handheld instruments such as DVM's, but they have also been launched in micro-satellites (below). So cheap doesn't mean they are not capable.

There are a wide range of variants with many options available including wireless communications. I selected a variant that had an Analog / Digital converter to interface the MAF sensor and built in timers that had a Pulse-Width-Modulation (PWM) mode to control the fuel injectors.

http://forums.pelicanparts.com/uploa...1395938554.jpg

The only hardware required outside of the MSP430 board was a large FET to drive the fuel injectors. I selected a FET that was designed for this task. Specifically, it had a beefy body diode that was rated to repeatedly absorb the voltage spikes created every time the injector circuit opened.

http://forums.pelicanparts.com/uploa...1395938665.jpg

So the basics of the required C program were to read the voltage on the MAF, and multiply by a constant to get the correct pulse width on the injectors. This task it would repeat as quickly as possible.

There was one catch, however. The output of the MAF sensor is does not correspond to mass air flow linearly. Fortunately, both Pro-M MAF units I received came with a flow bench data curve that is the test results for that particular unit.

So it was the task of the C program to read the MAF value and using the table below, estimate (by piecewise linear interpolation) the instantaneous fuel requirements.

const unsigned int air_mass_table[17] = {0,17,63,126,205,306,439,612,834,1111,1445,1836,22 82,2782,3336,3952,4647};

const unsigned int air_fuel_ratio = 12; // air to fuel ration here ( const unsigned int max_fuel_pph = 498; // sum of rated flow for 6 injectors
const unsigned int fuel_trim = 0x00;
const unsigned int time_to_open = 0x00a0; // injector opening time


There was another fascinating issue to solve. When using large injectors in batch fire, they would be fired infrequently such as every 100mS, because you don't want the time the injectors are open to fall much below 2mS. The 100mS cycle time would cause a somewhat lumpy idle. As the MAF reading rose above idle, I kept the injector open time fixed at 2mS, but reduced the time the injectors were closed to get the correct PW and smooth the running. But at high RPM, it worked the other way around. It was necessary to keep the injector closed time at 2mS to maintain accuracy, and vary the time spent open. A fun lesson in one of the most basic functions of an ECU.

http://forums.pelicanparts.com/uploa...1395938707.jpg

So that's all there is to it. Far from a great system by modern standards, but probably one of the most simple we've seen.

This is why I love this forum. I've been following this thread with great interest since it was started. It just keeps getting more interesting and educational. And as a bonus I don't have to feel so bad about my welding skills :D. I'm just finishing the last of my exhaust fabrication. I'll post pix son as my ugly welds are dressed.
Super cool project. Thanks for posting.

I know how to use a wrench and some JB weld.
You amaze me.