First Experiments with a Turbocharger
 
This pages shows some pictures of my first experiments with a turbocharger turbine. I didn't build the turbine myself. It is from Thomas Baumgart who was so kind to lent it to me. I only installed the measurement equipment. You can have a look at Thomas' page to see the details of construction.

On the right you can see the turbine together with some accesoires. The compressor case is in the middle, the combustion chamber at the top left side with the throttle valve in front of it. The transparent/grey flex tube leads to the pressure regulator which will be attached to the LPG tank later. Behind the regulator is the ignition box on the left. The white cable leads to the (kind of) spark plug on the right side of the gas valve. To the right of the ignition box you can see the oil pump and the oil filter connected with the black tube.
The oil pump is connected to the oil inlet port at the top of the turbocharger with a thin transparent tube. The yellow tubes go to the two pressure gages (left for compressor air pressure, right for oil pressure). Unfortunatelly, I only got those small ones with a 0-7bar (100psi) range, so the resolution is not very good as the interesting range is from 0-2bar (30psi). The black box on the left is a power supply for the RPM sensor and the oil pump. The thin spiral wire from the back of the turbine to the multimeter is the temperature probe, a thermocouple in a stainless steel tube.

(pictures can be enlarged by clicking on it)

This picture shows the exhaust of the turbine with the thermocouple mounting. I simply bent a small piece of thin stainless steel sheet and drilled a hole in it to use it as clamp for the probe. The diagonal angle has no special meaning, I just wanted to make sure that the distance from the tip to the clamp is large enough so that I actually measure exhaust gas temperatue instead of case/clamp temperature. The tip should be located at aprox. one quarter of the diameter, behind the middle of the turbine blades.
The RPM sensor is based on an optoelectronic barrier made out of a discrete infrared LED and a photo transistor. The LED is mounted behind a hole drilled into the compressor inlet at a position where the rotor blade tips interrupt the beam while passing the hole. The rotor has a total number of twelve blades but every second has an offset so only six pass the LED window. Because revolution speed is measured in rounds per minute (60s) six impulses per revolution is ideal resulting in a frequency output of RPM divided by 10.

The receiving photosensitive transistor is mounted in a hole at the bottom of the inlet. I've choosen this position (and not the opposite side of the LED) so that the electric starter I use will not disturb the beam when I couple it to the shaft.

Because of the relatively long distance between LED and receiver an amplifying circuit was required to get a signal with sufficient amplitude to trigger the multimeter (I'm lucky, my multimeter has a build in frequency counter). Details on the circuit can be found here.

The circuit could also be used for magnetic/inductive pickups which are more robust for higher temperatures and have less problems with dirt. I didn't use them because they require some magnetic material as wheel. The aluminia compressor blades gave only a negligible signal lost in the noise. And I couldn't use the steel nut in front of the compressor for the pickup because I use the nut for coupling to the starter.

The starter was made of a small DC motor (about 2" dia) and a 10mm button die from my toolbox. I simply bored the square hole in the back to a round one and made an adaptor bolt (on the lathe) to fit on the shaft of the motor. Unfortunatelly, the first attempts were unsuccessful. I underestimated the power needed to spin up the engine (or overestimated the motor's power), especially when the oil is cold.

So I dumped the small motor and suddenly realised that I had a bigger
one laying around in my workshop all the time: the spindle motor from my CNC-mill. It has 600W output power and 24,000 RPM max. I made a new adaptor bolt (cyanacrylate glue sticks really good when the gap tolerance is tight, I had difficulties ripping the old one off).
Ok, enough theory, for now. Let's feel some heat, smoke and noise...

I preheated the oil tank with a hair dryer. After connecting the regulator to the LPG tank and applying power to the oil pump I pushed the die of the starter on the compressor nut and turned on the motor. My homemade RPM sensor worked quite well and showed me 0.7kHz meaning 7000rpm. After turning on ignition and opening the gas valve a little bit, the engine lit up immediatly and the EGT rose to 200-300°C. I advanced the throttle and speed regulator further to about 10,000rmp and 500°C EGT.

It took over 5 minutes to warm up and for the oil to get liquid enough to go to higher speeds. The small DC motor was a little bit more comfortable at this point because I could set the power supply to a fixed current limit and let the motor spin up by itself (constant torque, variable speed). With the spindle motor I had to turn the speed poti step by step, open the throttle while pressing the shaft against the compressor, difficult to do at the same time with only two hands.

When I finally reached the 24,000 max rpm I tried to pull the starter off but the engine didn't accelerate. The speed droped and I didn't react fast enough to close the gas valve, so the mixture got too rich and a flame shot out of the exhaust (sorry, no picture). The temperature display rose to 1000°C (!) for one second but fortunately the temp probe heats up quicker than the turbine wheel, it glowed red but wasn't damaged, hopefully...

Still no success but I knew I was pretty close. I waited for the compressor wheel to come to a halt (took really long now that the oil was thinner so the bearings had less drag) then immediately tried again. This time it took less than 20 seconds to spin up to 24,000 rpm. I opened the throttle valve a little more until EGT rose to 600°C then pulled the starter off and gave the engine a little "kick" with the throttle. The temperature rose to 650-680°C and... (big grin in my face) it did slowly accelerate to over 30,000 rpm. As the oil heated up further reducing bearing drag the exhaust temperature decreased back under 600°C and speed stabilized at 31,000.
The digital multimeter I used to measure temperature and RPM is really worth its money (30 Euro bucks). It has a thermocouple input and a frequency measuring mode. Unfortunatelly, I didn't buy two of it, so I had to switch back and forth between the two modes. I carefully advanced the throttle while watching the EGT display to avoid overheating the turbine. At 40,000rpm the response to throttle movements was getting better and the temperature fell to only 520°C. The air pressure gage was still at zero.

Now I became a little more courageous and opened the throttle faster. Wow! What an amazing noise! It accelerated to over 70,000rmp (pressure gage showed 0.5bar/7psi) and was getting much louder than when I had saw it running when I had visited Thomas three weeks ago. That time, we had no measurement equipment at all. Maybe the engine now runs faster or the "sound effect" was caused by the turbine standing directly on concrete ground which reflected some of the noise. At Thomas' home we had mounted it on a stair handrail 3 feet above the ground.

I didn't dare to go above 75,000rpm because I didn't trust the oil pump (a windscreen wiper one) which achieved only 1 bar (14psi) oil pressure and even less as the oil became hotter. For safty reasons, I set theregulator at the gas bottle to only 2bar, so no turbine runaway could occur. If the compressor pressure rose significantly above 1 bar, the back pressure to the regulator would suppress the gas flow automatically. When I have solved the oil pressure problem, next time I'll try to go to 115,000 rpm max. ... Will have to find some ear protectors though. Unlike others said, I don't expect any problems running the turbine at maximum power on propane. The regulator allows a max setting of 4 bar and freezing is not a problem with a small turbocharger turbine and short run times.

Before shutting the engine down, I made some experiments to find the lowest possible self sustain speed. Below 31,000rmp the EGT rises from somewhere at the 500-550 range to about 650°C at 28,000rpm which seems to be the limit. At lower speeds acceleration becomes difficult and requires higher temperatures, so maybe my starter motor (24,000rpm) is still a little slow. But when the oil is hot, the power is more than sufficient, and a smaller motor which can speed up to 30,000rpm could be used. I consider installing an electric oil heating.

This picture was taken while the engine was running at idle speed (about 40,000rpm). The left gage shows 0.2bar (3psi) air pressure, the right one 0.6bar (8psi) oil pressure.

After the first run, I discovered that my ingition cable had melted because it touched the front of the combustor. I used standard TV antenna coax cable which has a thick insulation. You have to use the 75 ohms (larger diameter) type though. The insulation of the thinner 50 ohms cable (I have lots of it from cheapernet wiring) broke down internally giving no spark at the plug. I used a simpler version of  Thomas' bang box eleminating the complex step up converter and replaced it by a simple voltage doubler fed from mains voltage. I was to lazy rolling my own tesla transformer as Thomas did so I used a small standard 230 to 12V molded transformer which suprisingly worked (secondary winding used as primary).
Another improvement could be a proper inlet duct. Thomas and me discovered that most of the noise comes from the compressor inlet and at higher speeds the engine makes some sipping/fluttering noises which are probably caused by unstable air flow at the inlet due to the sharp edge. A bellmouth inlet might cure this although such a thing is not easy to make.
 

Thanks again to Thomas Baumgart for his turbine and Heiko Bleicher for shooting the photos!