Constant Current Hot-Wire Anemometer

Thread Starter

lfgrdwill

Joined Dec 9, 2011
27
Hey smart circuit guys,

I am a mechanical engineering grad student who needs to build a constant current hot-wire anemometer to measure turbulent air flows. Nowadays the commercial equipment one buys is always constant temperature, which uses a Wheatstone bridge and an amplifier in a control loop to increase the current as the flow speed increases. The constant current circuit circuit is more simple.

Does anybody have any suggestions on how to get started?

Thank you...

Will
 

joeyd999

Joined Jun 6, 2011
5,234
Hey smart circuit guys,

I am a mechanical engineering grad student who needs to build a constant current hot-wire anemometer to measure turbulent air flows. Nowadays the commercial equipment one buys is always constant temperature, which uses a Wheatstone bridge and an amplifier in a control loop to increase the current as the flow speed increases. The constant current circuit circuit is more simple.

Does anybody have any suggestions on how to get started?

Thank you...

Will
Usually, constant temperature mode is used as it can be compensated for changing ambient temperatures, which effect the power dissipation of the sensor (in addition to the moving air, which is what you are trying to measure). If you tell me how you propose to compensate for ambient temperature changes, I could probably help you develop some ideas.
 

Thread Starter

lfgrdwill

Joined Dec 9, 2011
27
I do not think varying temperature is that much of a concern, as most testing will be done at room temperature during wind tunnel testing. I have found an article on correlations that is worth looking into for major temperature differences.

One of my hot-wire books says modern constant current hot-wire anemometers (if there is such a thing as modern) use a Wheatstone bridge like the constant temperature type, but it did not occur why that is to my professor right off the bat.

Unfortunately, i took 2 basic EE classes plus a transformer class, but the entire subject has never been one of my strong points.

I can investigate further if needed (will continue to do so anyways).

Thanks
 

joeyd999

Joined Jun 6, 2011
5,234
Does your prof expect a particular kind of solution (i.e. Wheatstone bridge), or is the final implementation up to you?

Are you going to calibrate the anemometer against a known airflow, or do you expect that you'll be able to use physical constants to predict response vs. airfow?

What do you expect to use for the heated filiment? Operating a what temperature?

What range/accuracy/resolution are you planning for?

What is the maximum thermal response time required?
 

wayneh

Joined Sep 9, 2010
17,496
Nowadays the commercial equipment one buys is always constant temperature...
Pardon my ignorance, but in this technique is temperature determined by a property (resistance) of the wire itself? So the current varies to maintain a constant ∆V across the wire?
 

thatoneguy

Joined Feb 19, 2009
6,359
Pardon my ignorance, but in this technique is temperature determined by a property (resistance) of the wire itself? So the current varies to maintain a constant ∆V across the wire?
Couldn't the current be held constant, and the ΔV indicate windspeed?

Have a "zero" switch/input to indicate ambient temperature, then track voltage change.

That would need a rather high constant current through the wire, if it needed to track high velocity air.
 

wayneh

Joined Sep 9, 2010
17,496
Couldn't the current be held constant, and the ΔV indicate windspeed?
Maybe, as long as the temperature of the wire at zero windspeed doesn't turn it into a fuse. Depending on the upper range of the speed you want to measure, it might need a fairly high current.

But either method isn't terribly hard to implement. One just needs to understand the basic thermodynamics of the system.
 

Adjuster

Joined Dec 26, 2010
2,148
Why don't you want to use the standard constant resistance / bridge method? Doesn't it neatly "roll up" resistance measurement and current control into one process? Apart from reducing the risk of the element overheating, this method also reduces the need for precise circuit components. Only the bridge elements themselves, and whatever monitors the drive level need be accurate.

The apparently "simpler" constant-current system must deliver a regulated drive current, and the indication obtained may have a more complicated relationship to airflow. The usual system runs the element at constant temperature, so that its temperature coefficient need not be precisely known, nor need it be linear over a wide temperature range. The constant-current method may lead to wide temperature range, so the temperature coefficient will affect the calibration directly.

Somebody else has referred to the fact that temperature compensation (using a dummy element in still air?) is understood for the usual method, not necessarily for your proposal. There may also be thermodynamic issues such varying radiation losses, particularly if the element gets really hot at low air speeds, but as an electrical engineer these aspects are a bit outside my field.

So, why try to re-invent the wheel? Is there some complicating factor, such as extremely long cables to access the sensor, making it difficult to achieve a stable response with a distant control system?
 

Thread Starter

lfgrdwill

Joined Dec 9, 2011
27
Thank you for all the responses so far.

Velocities planned are up to about 50 m/s.

The constant current method is cheaper to implement without having to pay for a feedback amplifier. This circuit must be small and portable and i cannot route cables to a distant control system. For now it is not an option.

I am just trying to pick up whether the flow is turbulent or laminar, not necessarily measure the turbulence. One paper i found said the amplitude attenuation is about 0.5 at 300 Hz. I have no idea what resistances or circuit setup this guy used, other than it being constant current.

That dummy gauge idea for temperature compensation in still air didn't occur to me until now. I took a course that covered a lot of strain gauge set-ups so i am familiar with that.

Generally the probe resistance is about 5 ohms or so at room temperature, and increases by an overheat ratio (multiplying factor) of 1.6-1.8 at a wire temperature of 250 degrees. The wire is made of tungsten, 5 micrometers in diameter.

Just an fyi...convection heat transfer dominates radiation and conduction to the probe wire supports once a bit of velocity is flowing over the wire.

I am waiting for some more information sources at our library so until then i am just making do with the reading material i have.
 

Adjuster

Joined Dec 26, 2010
2,148
The constant current method is cheaper to implement without having to pay for a feedback amplifier. This circuit must be small and portable and i cannot route cables to a distant control system. For now it is not an option.
Perhaps you are overestimating the likely size and cost of an amplifier made using modern components, although this may not be so cheap if your only option is a professional packaged system. If this is the case, bear in mind that you may need to build the non-standard constant current driver for yourself.

In comparison, a constant current driver and monitor circuit is not much less complex than a control amplifier: a constant-current driver of any accuracy effectively is a control amplifier, except that it acts in response to its own output current, rather than by sensing a bridge output. Choosing open-loop operation does not of course remove the need for cabling between the sensor and its driver, so I fail to understand the objection to routing cables to a control system.

It is true that by driving the sensor directly rather than via a bridge, you will reduce the drive level required considerably, which will save power and may allow lower rated components to be used. If only a qualitative indication of turbulent vs. laminar flow is required, the constant current method may be more acceptable, but you will need to take seriously the risk of burning out the sensor with low or nil airflow.

At the least, I think the circuit should have an effective over-voltage limit, set not too far above the expected working level. Otherwise, if the flow stops the constant-current drive may send the sensor into a sort of thermal runaway, as its rising resistance will increase its input power.
 

Thread Starter

lfgrdwill

Joined Dec 9, 2011
27
What would be the best way to implement an over voltage limit? The circuit must have this.

I am glad you brought up the constant current source operation. Does anybody have recommendations on where to get a product like this. Would it be hard to build my own constant current driver?

Does a Wheatstone bridge eat up more current because of more resistors?
 

Adjuster

Joined Dec 26, 2010
2,148
You might start by trying to find out whether there are any pieces of instrumentation available to do the constant-current drive for you, but beware of the fact that some power supplies with apparently constant-current outputs have very little bandwidth if used as current sources.

A current controller could be made using an op-amp and a power transistor / FET. How hard this would be would depend a bit on the resources available, including your own experience. Not trivial, I would guess. How much current would be required?

The simplest way to limit the output might be to limit the voltage headroom of the driver device, but that might not be accurate enough. A closed-loop method monitoring the output voltage could in principle be far more accurate, but would require care to get a stable result.

The Wheatstone bridge inherently uses more power than the sensor within it. I would guess twice as much voltage, because the fixed resistance in series with the sensor is likely to be of about equal value. The current may not necessarily be twice as much, for the other side of the bridge might be made with higher resistances. Overall, this probably at least doubles the power input required. Another issue might be that at high airflows the bridge circuit would have to be driven harder to keep in balance, but I don't know if that would be significant.
 

joeyd999

Joined Jun 6, 2011
5,234
Velocities planned are up to about 50 m/s.
111 MPH! Unless this is a typo, you are out of my league. I think the thermal dissipation from your small wire will hit an asymptote at air speeds well below this.

My best guess is you will not see a significant reduction in temperature (in constant current mode) or a significant increase in power dissipation (in constant temperature mode) past the asymtote which should occur well below these velocities.

Also, trying to run constant current will probably fry your wire at low airflows. At constant temp, your going to need lots of power to keep your wire hot (warm?).

If I am wrong, I am sure someone here will correct me :)
 

Adjuster

Joined Dec 26, 2010
2,148
111 MPH! Unless this is a typo, you are out of my league. I think the thermal dissipation from your small wire will hit an asymptote at air speeds well below this.

My best guess is you will not see a significant reduction in temperature (in constant current mode) or a significant increase in power dissipation (in constant temperature mode) past the asymtote which should occur well below these velocities.

Also, trying to run constant current will probably fry your wire at low airflows. At constant temp, your going to need lots of power to keep your wire hot (warm?).

If I am wrong, I am sure someone here will correct me :)
Yes, the more I think of it, the more I'm afraid that to keep the sensor safe, the driver may need to be a control circuit in all but name.

Whatever you do, you would be well advised to find out as quickly as possible what drive current will be needed to get a useful response at the expected airflow and what the sensor voltage will be, also what current is safe in still air and what the voltage would be in that case. From this information, you should be able to work out if it would be possible to protect the sensor by having a voltage limit, or even whether a straight current limit would be OK without a voltage limit.

Perhaps a simple current limit will do, in which case you can laugh at us cautious old f@rts, but personally I'd not want to bet on it.
Good luck!
 

Thread Starter

lfgrdwill

Joined Dec 9, 2011
27
So i calculated out some details for this hot-wire project.

The current required through the hot-wire itself, which is actually a 0.0005 mm diameter tungsten wire probe, is about 100 mA.

Nearly all of the setups for one of these constant current devices were mostly used in the early 1900s before feedback technology was developed.

The tiny wire resistor is put into one of the arms of a Wheatstone bridge connected to a constant power source, so that when the temperature of the wire fluctuates due to transient air flow over the wire, the bridge becomes unbalanced and the voltage fluctuations across the bridge can be measured.

One of my articles says that measuring bridge voltage offers higher resolution than just measuring the voltage fluctuations across the wire gage setup in series with the constant current source and no bridge. Any idea why this is? Is it true? They say it is because much of the wire voltage is subtracted, with no further explanation.

Also, it would be nice to have some type of current limiter to prevent wire burnout for when the air flow is interrupted and there is insufficient flow over the wire removing heat through forced convection. I think one way to do this is by monitoring the voltage across the resistor. When the current x resistance is too high (the resistance will be highest at the hottest temperature when the air flow is lowest) the current should bypassed or turned off. I am working on finding out a max safe value of current at no flow.

Lastly, are there any good sources of thin tungsten wire in this diameter range. My professor suggested investigating lightbulbs.

Thank you again.
 

Thread Starter

lfgrdwill

Joined Dec 9, 2011
27
Hot-wire resistance is about 6.3 ohms at 250 deg C which is probably the highest temp i would like to go. 3.5 ohms at 20 deg C.
 

wayneh

Joined Sep 9, 2010
17,496
Also, it would be nice to have some type of current limiter to prevent wire burnout for when the air flow is interrupted and there is insufficient flow over the wire removing heat through forced convection. I think one way to do this is by monitoring the voltage across the resistor. When the current x resistance is too high (the resistance will be highest at the hottest temperature when the air flow is lowest) the current should bypassed or turned off. I am working on finding out a max safe value of current at no flow.
Just thinking out loud here, but what if you just had a voltage limit on top of a constant-current supply? As the constant-current circuit keeps upping its voltage as the wire gets hot, trying to maintain current, when it runs into a voltage limit, that should prevent it from maintaining full current when it's too hot.
 

Adjuster

Joined Dec 26, 2010
2,148
Just thinking out loud here, but what if you just had a voltage limit on top of a constant-current supply? As the constant-current circuit keeps upping its voltage as the wire gets hot, trying to maintain current, when it runs into a voltage limit, that should prevent it from maintaining full current when it's too hot.
That was the sort of thing I was trying to get at in post #12. Whether it would be sufficient is another matter, but at least it looks worth trying.

Data on the maximum safe current in different airspeed conditions, and the expected sensor voltages in normal and max temperature operation might clarify whether this would actually be safe enough.
 

MrChips

Joined Oct 2, 2009
30,702
It is common practice for such applications to use a Wheatstone bridge driven from a constant current source. The top two arms will be two identical sensors, one is the actual sensor in the air stream, the other is at the same location but protected from the air flow.

The bottom two resistors include a variable resistor to balance the bridge.
The signal from the center taps feed into a differential amplifier. I can post a schematic if you wish.
 
Top