Turbo rotational speed sensor/meter

[hhumble]

Does anyone know of a sensor that would be adaptable to or is designed for measuring the rotational speed of a turbo? Any help/suggestions are highly appreciated. as for the motives, for the application I have in mind, a boost guage is just not good enough.

Cheers

 

[Chancy99]

I've been toying with a design for a while for that. Pretty simple actually

You may be familiar with the plastic light-pipes that are used to move LED illumination from a circuit board to a front panel ? Same thing - use two pipes, running along the inside of the intake tract and terminating just in front of the intake wheel.

The other end has a bright LED shining into one pipe, the other has a photodiode. A microcontroller turns on the LED, then counts the pulses coming from the photodiode over a given period. Divide by the number of vanes in the turbine wheel and you get revolutions per period.

The plastic light-pipes are easy to shape - just heat and bend.

Dean.

 

[NickB]

The contrast of the light/dark reflections won't be very great, and you could have a hard time counting them at high speed, as some phototransistors can be pretty slow. Make sure you get a good fast one.

Another option would be to drill a small (1/8&quot hole right through the whole housing, crossing through the outer edge of the compressor inducer. A light source in one end, and a receiver in the other will count the light blocked and passed by the passing blades, and will give higher contrast for the sensor.

I saw once a setup that used a microwave transducer embedded inthe side of the housing at the inducer. The impedance of the transducer would change as the blades passed it as it would cause small eddy currents in the compressor.

 

[bassboy]

Consider a hall effect sensor, good high frequency

 

[NickB]

Hall effect sensors sense the presence of a magnetic field. They work great for crank and cam sensors, as yo can easily embed a small magnet in the flywheel, or elsewhere on the engine. I see that as being VERY difficult to do with an aluminum compressor wheel spinning at 150000+rpm.

 

[carnage1]

besides a mgnetic field moving that fast would be a problem.

 

[NickB]

Yeah, i just figured it would be easier to understand the difficulty in putting a magnet in there, and not worry about trying to explain magnetc flux

 

[Andy330hp]

I thought some OEMS (GM on the buick grand national) had some turbine speed sensors. Also, I know for fact that Garrett senses shaft speed. How else could they develop their compressor and turbine maps? Of course, I have no idea how they do it....

 

[NickB]

The Grand National has a boost pressure gauge, as do many turbocharged cars. It's not a shaft speed indicator.

 

[Andy330hp]

I had remembered something about a turbine speed sensor error code, but it turns out it's for torque converter turbine speed. Oops. Still, there's obviously a way to do it....

 

[Chancy99]

Typical turbine speeds are in the 70k rpm range I believe ? How many blades (thus pulses) on a typical compressor wheel ?

Here's one that is 4uS rise and 6uS fall time.

http://www.pacer.co.uk/db_pdf/VTT1212 & VTT1214.pdf

And 2uS rise 3uS fall.

http://www.pacer.co.uk/db_pdf/VTT1222W & VTT1223W.pdf

I don't think we'll have any trouble getting one fast enough. Hrm - anyone know if light-pipe plastic will transmit IR as well as it does visible light ?

Dean.

 

[zapatatapaz]

Well what if you drill the little hole at the tip of the shaft or put a light cap on it so that every turn of the shaft you`ll get only two signals instead of one per every vane
then with anything capable of sensing 300000 optical signals a minute will do or if you drilled the hole not though the middle but to one side and you put your sensors in line with it you would get a signal per turn.

 

[Chancy99]

I was thinking of that while staring at the turbo inlet the other day. There is a nut that holds the wheel in place. One could paint (high-temp paint each alternating face of the nut black and white.

Then drill a small hole in the intake hose flange, so a piece of brass tubing would fit. Run a pair of fibers through it, pointing at the nut. The other end goes back into the cabin to the unit, where the ends terminate at an IR LED and photodiode/transistor.

Would normal multimode or singlemode fiber work here ? That brass tube will get pretty hot ...

Dean.

 

[frankeng9]

Hi hhumble,

Sensing the speed of a turbocharger could be a very difficult task. My main concern is that the high temperatures involved could fry any sensor in a short period of time. From my limited experience in working with speed pickups used in diesel engines (engine RPM that is), I can offer some possible suggestions.

Do not interface your speed pickup sensor directly at the turbine wheel where the maximum temperatures will be seen. There is no need to count the number of blades on the turbine wheel. On diesel engines, a magnetic pickup (“mag-pickup for short) is mounted on the bell housing of the starter gear and the sensor “counts” the teeth on the flywheel as the engine spins. I think a similar approach would work for your application. Instead of counting blades, you could mount your mag-pickup at the coolest part of the turbo shaft (assuming the temperatures there will not melt the sensor) and trigger the sensor off some single part of the shaft (keyway or whatever). Now you have one digital pulse from the sensor per revolution (MUCH easier, no need to beat yourself up). I have worked mainly on the software side of these applications, but I know the Electrical Engineers feed the pulse signal from the mag-pickup into an electrical hardware circuit called a phase-locked-loop (PLL). The PLL processes the signal and increments a dedicated counter once for every pulse from the mag-pickup. This frees up the ECU or microcontroller from operating in a dedicated mode where it’s trying to run code to count input signals, convert the count to RPM and do something with the value obtained, all while performing the other tasks required to run the application. By setting up the microcontroller to trigger on a DEDICATED interrupt schedule (e.g. every 20 milliseconds the microcontroller stops whatever it’s doing and runs this routine) you can read the elapsed counter value at the start of the interrupt, and since you know the micro is reading this counter EVERY 20 milliseconds (i.e. a standard timebase), you can easily use a simple lookup table to convert counts to RPM. This may be the only practical approach in acquiring pulses at the speeds you’re looking at. My best suggestion would be to find an Electrical person with some experience in PLL’s and see if you could get some assistance with a prototype circuit. What you are trying to do has been done before in other applications (they do measure the shaft speed of turbine engines), but I’m not sure the extreme temperatures will cooperate in your application. Best of luck!

 

[bb61]

We are re-launching a sensor for aerospace applications that has been proven in turbine applications. It should be adaptable to a turbocharger application. It meets the temperature, speed, shock/vibe etc requirements for flight. Let me know if this is of interest

 

[Chancy99]

bb61 -

This definitley sounds interesting. Umm, given that's it's aerospace, how does pricing look ?

Dean.

 

[cirque]

Omori has a turbine speed meter that looks at the blades of the compressor.

http://www.omori-na.com/

 

[torquer]

We test the speed of the turbos on all gas and diesel engine dynamometer test stands. We record this speed with the data acquisition system. Typical is using a magnetic pickup as several above people suggested. This is typically used to pick up 2-PPR off the turbo. There aren't magnets on the shaft, but you can find where you can get some offset steel 'teeth' on each side, typically designed in by the manufacturer.

Visit

http://www.speed-position.invensys.com/pdf_aero_b1-1.asp

to help choosing the correct mag-pickup. You can then take the frequency output ie typical diesel engine turbo operates at about 120kRPM, or with 2-ppr at about 4kHz. For a cheap monitoring device, choose a signal conditioner that takes the 0-4khz and changes it into a 0-5VDC output, then put that into a calibratable panel meter, Newport or someone. For our data acquisition we use Dyne Systems Company's "Companion" system (dynesystems.com), but we're monitoring way more channels of data than just turbo speed.

-Bob

 

[beaudreau]

I am new to the electronics field and this a project I have been looking at also. I believe that the light sensing option would be the best method. My plan was to use the compressor nut painted white on just a small portion and the rest black, most of these nuts are 12 point. I hoped to angle the light in and get enough relection for another sensor to pick it up. If this reflection wasn't enough, it might be tried on on the face of one of the blades. Whatever method is used, I'd prefer that just one pulse per revolution was used.

I'd like feedback on whether any of my plan is feasable and if so, what components might be used for the project.

Thank you,

Roger

 

[Doug020668]

You could still use a hall effect sensor with a stationary magnet located near the compressor blades, as long as the blades were constructed of a suitably magnetically conductive material.