The finished plenum installed on the bike.
The second reason was that it had been done before. RB Racing broke a land speed record with a turbocharged K engine back in the 80’s. So I knew that a mild mannered low to mid boost system was possible.
It should be mentioned that while I did buy an Air/Fuel meter from RSR, that was the extent of their involvement in this project. THIS IS NOT AN RSR TURBO KIT. Please do not call them and ask them to build you a turbo K-75. I did, and they made it clear that they have moved on and have no interest in repeating this project. Also, please don't call me and ask me to build you a turbo K. I built this bike for me and me alone. I did it to go fast and I did it to learn. I'm not starting a business. If you want a turbo K, you're welcome to copy my work, but you've got to build it yourself.
There is a third reason as well. Somewhere along the line, I told somebody of my idea and they told me that I couldn’t do it. This always pisses me off and pretty much ensures that I will do it. The problem is that, now that it's done, I don't remember who it was. That's a shame, because they deserve a nice big burnout done on their driveway. There's nothing I hate worse than naysayers and Monday morning quarterbacks. Except for children who talk in the theatre, who are #1 on my list.
All four pipes are metric sizes, so I bought thick aluminum pipe and machined it to fit. I machined four intake runners, learning enough from the first one that the final three were identical and fit well. The real difficulty was learning to TIG weld with aluminum. I practiced with 16ga sheet and the first intake runner. I had good results until I used a piece of stainless filler rod that some nimrod had stuck in with the aluminum. Good thing I was testing.
The plenum went together with little incident, although I chose to add the injector bosses last. This was a mistake, as it was hard to get the torch into the tight angles formed once the plenum was complete. The end result was a blobby, huge weld on each boss. The welds were airtight, so I was satisfied.
I will eventually rebuild the plenum, as I have learned more about plenum design and believe that I can improve on what I have. For starters, the intake runners need a smaller inner diameter to provide a smooth transition into the throttle bodies. Then I want to add velocity stacks inside the plenum itself. Lastly, the volume of the plenum chamber can affect the way the bike performs, so version 2.0 will have a calculated chamber volume. All of this means that I need to start from scratch with a new plenum.
The finished plenum installed on the bike.
But first I needed a turbo. I called up Majestic Turbo and spoke with the owner, Kevin. We agreed that the Garrett GT-12 was the right size for the job. Initially we’d considered the larger GT-15, but it would have been easy to push this turbo into surge. The GT-12 is better sized for a 750cc engine. It is smaller and so will hopefully spool up faster. Maximum power will be limited, to about 130hp, but that's more than I think I'll need. I drove up that weekend and picked it up. With $650 in the hole, I was now committed to this project.
First I made some clamps to hold the pipe onto the head. This was done by using my mini-mill to drill and bore out three pieces of stainless ¼” plate. The pieces were then ground to shape on a bench grinder, and finally polished.
Exhaust clamps from left to right: Rough cut with plasma cutter, bored and ground to shape, finished.
The next item on my list was the header itself. I initially wanted a true equal length header, but kept running into topology problems. On one side I had the oil pan. Too far the other way and I’d drag the header in corners. And it couldn’t be very long, because it had to mount the turbo close in to the engine if I wanted it to clear the front fender. Eventually I settled on a simple approach.
This pipe sucks.
So I went back to the drawing board. After a lot of work, I came up with a better version. While not equal length, this second header is compact, has a real collector on it and could be straightened when the welds misaligned it. Here is the second version with the collector finished. From the collector the pipe will run straight up to the turbo.
Primary runners in place, haven't started on the collector yet.
Collector finished.
Next I fit the turbo itself. This turned out to be more of a bear than I expected. The turbo’s wastegate actuator sticks out to the rear, and unless it is positioned just so, it will rub against the cam chain cover or the radiator. Neither one is a good thing. The problem is that the actuator has only six positions on the compressor housing, and the only way to get it to clear was to rotate the compressor housing until its outlet was inaccessible. And don’t forget, if I moved the turbo too far forward the fender will touch the compressor housing when the suspension compresses.
The solution was to modify the actuator’s mount to allow it to sit in the correct position when the housing was properly clocked. It was also necessary to trim some excess metal off of the bottom of the radiator, but in the end, it all fit.
The problem was that the AFM has a square passage, and pipes are round. My solution was to create an adaptor plate out of ½” stainless plate. With a lot of die grinder work, I was able to make a hole that transitioned smoothly from round to square I then welded on a stub of 2” stainless pipe which will connect to a silicone 90 degree bend from Turbo Hoses.
This is the stock SAAB intercooler.
For the first attempt, I followed conventional wisdom and made an end tank that tapered smoothly from one end to the other. This is supposed to encourage the air to move through all the tubes of the intercooler. But when tested, 90% of the airflow was coming out of the two tubes farthest from the inlet. So, back to the drawing board I went.
Version 1, currently gathering dust in a corner of the garage.
For version two, I designed a squarish end tank with two internal baffles to divide the flow among all the tubes. I tested this tank by blowing compressed air into the inlet, and found equal flow across all the tubes.
The top tank of this intercooler was more art than engineering. The outlet pipe had to be mounted at a pretty sharp angle, and so I decided to make a tall end tank with a curve in it to help the airflow make it around the corner. It took a lot of trial and error, but the results seem good.
It turns out that it's not easy to find fittings that have metric threads on one end and -4 JIC connections on the other end, especially if you need 90 degree connections and they have to be clockable. After a lot of websearching I found what I needed and found a supply house in Austin that would order them for me. But they weren't cheap. I wound up paying about $70 for four fittings. Two of these went on the turbo's coolant connections, the other two went on the oil pump and water pump.
The water pump has a drain plug on it, and this is where one of the fittings was installed. This drain is located behind the impeller, so it makes an excellent water return. To feed water, I simply bored a hole in the pump cover and installed a modified -4 JIC bulkhead fitting. This was one of the easier modifications.
So now coolant was taken care of, but I still had to deal with oil feed to and from the turbo. This turned out to be much more complicated. Most turbos have simple threaded fittings for oil feed and return. Due to its small size, the GT-12's center section is flat on the bottom. There are two holes, a 1/8" oil feed hole, and a 1/2" oil return hole. There are also two tapped holes for screws so you can mount your fitting. What fitting? The one that you get to make.
On the left is a storebought bulkhead fitting. On the right is the fitting that I made on my mini-lathe and mini-mill. On the back of each oil hole is a groove for an o-ring. I also machined and welded up a piece of aluminum pipe with a 90 degree angle, which was press fit into the half inch hole so that I could connect an oil drain hose.
From this fitting, the oil drain hose was about two inches long. Look at the front of any K engine and you'll see that there's a round recessed area on the front of the cam chain cover. I have no idea why BMW put it there, but it was the perfect size for my purposes. I bored it out, tapped it and machined up a hose fitting that screwed into place. The best part is that the oil return is invisible unless you're being run over by the bike. Turbo oil returns into the cam chain area and then flows back to the sump.
The oil return fitting installed, like it was meant to go there. The flange on the right is where the header connects to the turbine. This photo was taken at the 1000 mile teardown, which is why the stainless steel header has turned brown and gotten a good coating of carbon.
Finally I had to make gaskets to seal various parts together. For the turbine's connections, I bought some copper sheet and cut it into the shapes that I needed. Next I annealed the copper. To do this, I used a propane torch to heat the copper until it turned a purplish dark grey. This is as hot as it needs to get. Any hotter and the copper will build up carbon deposits that you'll have to scrape away, ruining your gasket in the process.
Once your copper is at the right temperature, let it cool down. Some people believe that you need to quench the copper in cold water. This is true for some metals, but it is not true for copper. Just heat it, then let it cool and you've annealed it. You'll know if you've done it right, because it will be much more flexible than it was before.
My initial plan was to have a short pipe that exited the turbo and dumped the exhaust below and to the front of the engine. The theory was that the turbo would muffle the engine enough that I wouldn't need a muffler. And, I could build this dump pipe out of the mandrel bends left over from making the header. I quickly bashed this part together and rolled the bike out of the garage. With the fuel pump disconnected, I cranked the engine until the oil pressure light went out, then another 30 seconds to make sure.
I reconnected the fuel pump. Fire extinguisher at the ready, I pressed the starter button and was rewarded with an engine that cranked strongly, but would do little more than pop. Adding some throttle got the bike to run, but it was horrible. It would miss almost every other stroke, would barely respond to throttle, and kept dying on me. What the heck?
At times like this, I've learned that it's best to take a step away from the problem and then approach it again. A cigarette smoked, a walk around the bike and what's this connector dangling under the gastank? It wouldn't happen to go to the AFM, would it? Yes, I believe it would. Connector plugged in, key on, ignition on, starter button and vroom! It runs! It's running!
I let the bike idle for five minutes to break in the turbo bearings, shut down and looked for leaks. No problems found, I geared up and rode my turbo bike for the first time.
I didn't know what to expect, so I was very cautious. My first few passes up and down the street were at not much above idle. With each pass I'd give it more throttle. At about the 6th pass I rolled on and heard that distinctive whine of a turbo getting onto boost. A few more passes and that whine was peaking and disappearing as the turbine blades broke the speed of sound. And boost was coming on.
I continued to be cautious for a while, so I'll skip that and get to the good stuff, my first all gears WOT run from a standing start.
The bike launched like a normal K-75. It was only once the engine reached 3000RPM that things got interesting. The turbo went from off boost to fully on boost in about half a second, and the bike SURGED ahead. Second, third and fourth gears were a blur, and then I was in 5th, doing 115, approaching redline and still not running out of power. And this was only with 5PSI of boost. At a guess, my 0-60 time was 4 seconds.
In traffic, the bike really shines. When passing cars on highways, I used to wonder,"Can I get around this guy before the passing zone ends?" and,"Should I downshift to fourth or third to pass this guy?" Now instead I ask myself,"Should I pass three cars in this pazzing zone, or should I drop to fourth gear and pass all three cars and the semi that's slowing them down?"
I had to learn to ride corners all over again. On a bike, as you enter a corner, you always roll on the throttle to stabilize the chassis and power you through the turn. Now, though, halfway through the turn the turbine would be spooled up and I'd get a disconcerting increase in power. It wasn't hard to learn how to compensate for this with delicate movements of the throttle, but I can understand why bikes aren't sold with turbos. For the average rider, it's too much to deal with. I still have concerns about how I'll handle this effect once I'm running higher levels of boost.
The first problem was a thin coating of oil on the back of the front fender and oil dripping off of the turbo's oil fitting. This problem was traced to the O-rings. I'd used cheap ones from a Harbor Freight O-ring assortment, and they had dried and hardened in the heat from the turbo. I replaced them with Viton o-rings, and the leak is now almost gone. There's still an oily feel to the fitting after a long ride, but that's it. Next time it's apart, I'm going to put some Hylomar HPF on the joint to help seal things up.
Next, I didn't really like the dump pipe. It had not been as quiet as I hoped. It wasn't Harley loud, but it wasn't BMW quiet either. And the K just doesn't have a pleasing exhaust sound. Instead it sounds like three chainsaws revving in unison. I had extended the dump pipe to go back into the muffler, and that had silenced the bike. But because the tubing was so small, it had also restricted the bike. I could feel it in increased turbo lag and a later spool-up RPM.
With some larger pipe I made a new exhaust. This one looks better and flows better. The results were the same level of quiet with reduced turbo lag. It's still not as quick to boost as the unmuffled system, but that's to be expected.
The next modification was to add some sensors. Westach sold me a 3" triple EGT gauge, RSR sold me an Air/Fuel meter that has been irreplacable in tuning the engine properly, and a boost gauge finished up the set. Initially I had them mounted on the handlebars. Later on I built a housing for them that bolts onto the gastank, which keeps the instruments out of sight but still easily viewed when I want to see them. I found early on that these instruments were too distracting and were interfering with my ability to ride safely. Now that they're on the tank, I can look down and see them, but my eyes are not constantly drawn to them.
The bike runs rich across most of its power band. More importantly, I have a problem with the bike leaning out dangerously if I attempt maintain a constant speed above 85mph. WOT acceleration all the way to redline is just fine, but steady state cruise is a no go. As you can imagine, this is a bit annoying. Sometimes you want to cruise at 110.
The problem is the K's antiquated EFI system. It's an analog system using a flapper valve. This flapper is not very adjustable, and is affected by accleration, braking and cornering forces. I can see the bike go lean under heavy braking, and leaning to the left or right also has a small effect. The computer also does not have an oxygen sensor, so it cannot see the mistakes that it's making.
I can spend a whole lot of time messing with it and maybe making it right. Or I can just replace it with something that was designed in the last 20 years.
MegaSquirt also needs an air temperature sensor in the inlet tract. The stock flapper valve has a sensor built in, but the whole flapper system was going to go anyway. I got a sensor off of a Nissan sedan in a junkyard, and was able to mount it to the stock plenum with some Yamabond and a couple of pop rivets.
I decided that now was the time to remove the plenum and replace it with something better. And there isn't much better than the stock BMW plenum. It's got a lot of volume, internal velocity stacks, it fits well and it doesn't look half bad. So I've reinstalled it, and made an adaptor to connect its metric sized inlet to the 2" tubing that I'm using for my intake system.
One nice bonus is that with this setup in place, I can reinstall the stock airbox. I still need to work out a way to connect it to the turbo's inlet, but I've got a few ideas there. I'm looking forward to ditching the POS K&N filter and going back to a paper filter. Because of space constraints, I may have to reverse the flow in the airbox, and have intake air entering the top and exiting the bottom. It should work OK, since the filter can be mounted upside down in the airbox. I'll have to open up the airbox more often to remove dead bugs, but that's not a huge concern. Water getting into the airbox when it rains is a potential problem, so I'll have to design well.
The Megasquirt ECU fits nicely into the same place as the stock ECU. A spare piece of lexan was the right thickness to work as a mounting base, so I transferred the holes for the stock computer's rubber mounts over to the lexan. It's not necessary to use Lexan, I just happened to have a piece lying around. I kind of like the clear look, so I may build the whole box out of the stuff.
Right now the bike is almost totally disassembled. Early on in the tuning process I discovered that I was getting far less engine vacuum at idle than I should be seeing. Normal would be about 40KPA absolute. I was seeing 70KPA absolute. After a whole lot of diagnosis, testing, guessing, retesting, etc, I finally found the problem where I'd hoped it wouldn't be. The piston rings were worn and leaking badly. A leakdown test showed 20% leakage past the rings.
I've got the first piston out, and the signs of blowby are obvious. There's discoloration on the sides of the piston down as far as the oil control rings. There is also carbon buildup behind the second compression rings.
I've also got some slight wear on the big end bearing for the connecting rod. It's no cause for alarm. The wear is so minor that it's probably just from engine startup. However, it's a good time to replace bearings, so I'll be getting all new shells for the main and big end bearings.
At the same time, I'm thinking that I'll replace the pistons. I have it on good authority that the stock pistons will break off their ring lands somewhere around 12PSI of boost. There's also the problem of detonation. Right now with the stock 10.5:1 compression, the engine will detonate under 5PSI of boost with 89 octane fuel. As boost rises higher, this problem will only get worse. So it seems like a really good time to install some low compression forged pistons. I'm thinking of a CR between 8.5:1 and 9:1, but haven't decided. Advice is welcome.
Finally, the valves will get some attention. The intake valves look and seal like new, but the exhaust valves aren't so hot. The one I pulled out looked like its seat had been worked over with a knurler. I don't think I can safely recut them without removing too much material, so I plan on new exhaust valves and seats. Given the condition of the intake valves, I may settle for just lapping them slightly and replacing their oil seals.
To see images of the rebuild, click here
