Air Flow Measurement

Best I can suggest is a very small Pitot tube. I think gliders use a sort of reverse Pitot tube.

Are you sure you'll only get temperature changes of 20^oC: have you looked at Ranque Hilsch vortex tubes. The people trying to figure out exactly how these work must have tackled this problem.

Does it matter if you affect the temperature in the device? I'm thinking about a flat, flush mounted "heatsink" which is being heated by a well regulated source; a temperature sensor in the same device should give you a good indication of air speed. You'd need to do a lot of calibration, and, you might need a reference temperature sensor next to or just "in front" of it.

Good luck! Please let us know how you get on.

drobertson:

If the cylinder could be mounted so that it's free to rotate, it should tend to rotate with the air. The force required to restrain it will then be proportional to rotational velocity. As with any other approach, calibration is key.

DickL

With the current setup the tube is fixed. I may be able to modify the configuration to allow some movement. I could possibly incorporate a strain guage to measure the torque on the outside tube. The Pitot tube is also a good idea, but it would neeed to be carefully incorporated in the design to reduce the turbulence it would create.

Thanks for the ideas. This is a tough one and your help is getting me thinking.

-Doug

If any intrusion is possible then why not use a micro bead thermistor?

The temperature sensing head is only 1 mm diameter, so its effect on air flow disturbance should be negligible.

Failing that perhaps a simple hot wire anemometer sensor could be used, as the hot wire is once again tiny in size.

Just thinking out aloud here hope it helps?

John.

You could try using a constant temperature anemometer with another temperature gauge to compensate for the low frequency overall temperature change. The 20kV ionizing discharge may play havoc with that, though. Or, you could use ink drops to visualize the flow along the walls. Basically in this case you would just add a grid of ink droplets along the wall of the cylinder and see how they are skewed by the flow.

drobertson:

You might also consider using a VERY small hole in the cylinder wall coupled by tubing to a pressure sensor. Pressure coupled from a hole perpendicular to the wall may be dominated by Bernoulli, while one tangent to the wall may provide positive pressure. A little experimentation and, again, calibration would be called for.

Best regards.

DickL

Good idea DickL

However, will a 20C shift in temp. give errors in the pressure reading due to the changes in air density. I susjest introducing a thermocouple into the system. Temp reading taken with pressure reading to increase accuracy.

RPM will be a function of pressure and temp.

Regards

Jon

We will have a pretty good system for reading temp. I can use that for making adjustments to the RPM measurements.

This is going to be a bear to calibrate. :-(

-Doug

Better still how about using a microbridge air flow sensor, taking the microbridge out of the sensor and mounting it flush to the inside of the tube...

This would give you a temperature compensated mass flow measurement of air flow, which shouldn't need re-calibrating, assuming the air flow is reasonably laminar when flowing across the tiny sensor...

I've used hundreds of the Honeywell ones before now and they are remarkably accurate even when exposed to temperature and pressure changes. Even when hacksawed open to remove the micro-bridge sensor?!

John.

Interesting. That one would fit my "Simple for me to implement" standard. I am lazy and hate getting into overly complicated designs if I don't need to.

Do you have a link for sourcing these and the specs I could look at?

Thanks

Doug

Farnell stock them as standard parts worldwide...

here's a link (hopefully) _

http://uk.farnell.com/jsp/search/browse.jsp?N=500004+1002015&Ntk=gensearch_001&Ntt=airflow+sensor&Ntx=

Good luck - John.

Hello,

If you have an open view from the end, I would suggest introducing a light particle such as a small piece or snip of paper and then use a high speed camera, or even a strobe light might do to check the advance of the paper snip per 1/1000 second or so. Hopefully the rotation in that time is less than 1 rotation. Even if it is more you would still know its speed if you have a ball park figure anyway.

Without open view a glass insert and the camera would still do.

Hope that helps.