Airflow sensor

For a project I need to have an airflow sensor to protect a circuit in case
of a fan failure. However this airflow sensor needs to be fitted on a PCB,
and if possible only SMD.

Analog devices has an airflow sensor (TMP-12), but this one is not what I
need. 1) the ambient temperature is fixed (can be fixed in some way to let
the device work with a variable ambient temperature), but 2) the resolution
is bad, max 5deg centigrade deviation. I would need to detect an airflow of
min. 0.5m/s. That is not possible with this device.

After a long search I couldn't find any airflow sensor that can be mounted
on a PCB (and SMD). So my question is if someone knows such sensor or how to
build one with simple components?

An idea that I have is to have 2 identical diodes, one self heating due to a
constant current flow and one for measuring the ambient temperature. When no
airflow around the self heated diode the forward voltage drop will get
lower. With an airflow this voltage will go higher (diode cools down). The
delta voltage with the identical temperature sense diode is a measurement of
the airflow. Am I correct?

If someone has any source of information about this subject it would be
handy

Thanks,
Al

 

It's often better and easier to measure the thing you really care about.
The Temperature of the sensitive device is probably easier to measure
than the secondary airflow.
Anything that has a temperature dependence can be used in a bridge.
Low thermal mass is better.
Take a look at hot-wire anemometer designs. Mine has two thermistors.
The BIG one is about the size of a grain of sand. The LITTLE one needs
a microscope to see it.
You can make a sensitive airflow detector from two incandescent
light bulbs side by side. Bust the glass off one of 'em. But it's
not rugged enough for general use.
mike

--
Return address is VALID.
Bunch of stuff For Sale and Wanted at the link below.
Toshiba & Compaq LiIon Batteries, Test Equipment
Yaesu FTV901R Transverter, 30pS pulser
Tektronix Concept Books, spot welding head...
http://www.geocities.com/SiliconValley/Monitor/4710/

 

Makes sense. Where you actually have to measure airflow is something like an
automobile engine. Not positive, but I think the mass airflow sensor
measures the resistance of a piece of nichrome type wire, with enough
current going through it to self heat.

Tam

 

Yes, what you are suggesting should work. I'll suggest an alternative-
look at the ripple in the fan current if it's a DC brushless fan. You
can tell if it has seized or gone open. I'm not sure if this method
has been patented by someone or not..

Best regards,
Spehro Pefhany

 

without airflow, you are left with adiabatic heating (near enough) so the
mass of your heatsink soaks up the heat (actually the Rtheta just got real
high, so the thermal time constant just got a lot bigger, as did the
steady-state temperature. but the bit your interested in looks adiabatic, ie
you lose bugger all heat until T is way too high). The mass gives you time,
which is why a thermal trip works - if, of course, it responds quickly
enough. But in general heatsinks are big and thermal sensors are small, so
it works well. I have used thermal switches, NTCs, silicon temperature
sensors etc - all can do the job. And you can stick the sensor anywhere you
like, as long as you experimentally determine the corelation between
Tmeasured and Thotspot.

Funny story: during development of a 100kW 3-phase motor controller, I had a
junior engineer playing with ducting around the heatsink fins, to maximise
cooling. He was using cardboard & masking tape for the ducting. Which worked
fine until he forgot to turn on the fan (separate supply for this test as we
only had a mechanical prototype). The unit was 97% efficient, so dumped 3kW
into its uncooled heatsink, which got hot, and set all the cardboard on
fire. Production staff came and got him when 3' flames were seen coming out
of the enclosure, which was unsupervised. Procedural changes followed. And
all of our techs told him to go to hell when he essentially demanded they
rebuild his drive - you fucked it, you fix it. He got a bawling out over the
lack of supervision, and a public one for not being nice to the techs. And
he had to re-built it himself. I betcha he never made that mistake again. We
then wired a relay in parallel with the AC fan, that de-energised the main
contactor if the fan power was off.....and made him stand and watch the
experiment. And we put up a 5' perspex barrier on top of the 3' wall around
the R&D test area (but that was more because one of our 400kW destructive
tests involved rapid disassembly, and production staff got bombarded with
shrapnel

Cheers
Terry

 

dont Papst use a similar trick in their expensive fans with "oh shit"
outputs? I have done similar things in the past.

Cheers
Terry

 

Maybe they use magnet, pickup and retriggerable monostable or
something. I don't know how they work, but the fans are available with
such outputs and AC power.


Best regards,
Spehro Pefhany

 

Or just use the tachometer-equipped fans common in many PCs. Motherboard
fan/temperature monitoring with data available on an I2C buss is
commonplace these days.

 

The best, simplest, most failure-resistant design I've seen was a small
flap attached to a microswitch. The flap was positioned in the airstream.
When there was airflow, the flap kept the switch open. When there was no
airflow, the switch was closed. Mechanical solutions frequently can beat
electronic ones due to simplicity, cost, ruggedness, and ease of
implementation.

 

Objection! I have seen this solution in the old Ampex VTRs, and it was a frequent
source of failures.
The things would get stuck, damaged by maintenance (you just needed to touch it),
and usually cause severe damage to overheating to the equipment if they did
not work.
Tacho is good.
And proportional too:

CPU core: +1.66 V (min = +1.53 V, max = +1.73 V)
I/O: +3.33 V (min = +3.13 V, max = +3.43 V)
+5V: +4.83 V (min = +4.70 V, max = +5.23 V)
+12V: +12.07 V (min = +11.39 V, max = +12.44 V)
CPU Fan: 4787 RPM (min = 4687 RPM, div = 2)
P/S Fan: 2824 RPM (min = 2700 RPM, div = 2)
CPU Temp: +57.6°C (limit = +63°C, hysteresis = +65°C)
SYS Temp: +38.9°C (limit = +45°C, hysteresis = +48°C)
VIA686a Temp:
+26.3°C (limit = +38°C, hysteresis = +40°C)

Ain't Linux COOL
This was lm-sensors.
JP

 

The temperature data is probably as helpful as the tach data. If a fan
stops, you have time to shut down before the temp rises. But if the
airflow is blocked (and the fan keeps going), you'll be glad you had
overtemp alarms too.

Interesting question: If the airflow is blocked, what happens to the fan
speed? Vacuum cleaners speed up when you block the hose. If the P/S or
CPU fans do so as well to a significant degree, maybe a fan overspeed
alarm could be added to the code.

--
Paul Hovnanian mailto
note to spammers: a Washington State resident
------------------------------------------------------------------
It's easier said than done.
.... and if you don't believe it, try proving that it's easier done than
said, and you'll see that it's easier said that `it's easier done than
said' than it is done, which really proves that it's easier said than
done.

 

I have tried, power supply fan does not seem to speed up if you
cover the outlet..., not 100% though, but the reason is likely
that these motors are synchronous somehow.
In it is a 3 pole stator, with the magnets rotating around it.
There are no brushes, so the 3 stator windings likely are driven by
some small chip (3 phase rotating?).
(I opened one up for repair, some small PCB is in it).
JP

 

If there is no overlap between the blades, he could shine a light through
the fan at a light detector. for a 4 bladed fan, that would give 4 pulses
per revolution. Amplify it and rectify it.

Tam

 

OK, I agree with this. The only thing is that I can't use any mechanical
sensing device such like a DC fan, flap and switch or anything like. The
reason why is that it must be placed on a PCB without any adjustments for
calibrating the sensor, and it must be as small as possible (if possible
even no through-hole device)

I will just put something together with a self heating diode and a measuring
diode and see how it reacts. Will post the results

Thanks everyone for the ideas!
Al

 

A google search for 'Electro fluidic autopilot truned up the
Sport Aviation index at:

http://www.cozybuilders.org/ref_info/sportavi80.html

August 1980 issus

The project uses two thermosistors as an airflow sensor.

Includes the complete schematics and quite a bit of theory...


Bext of luck,


Ricahrd Lamb

 

Sounds interesting! Only the small problem is that I never heard about this
magazine.. and august 1980 is already some time ago, so I cant get it in a
shop anymore

If someone can scan these pages I would be very happy!

EAA Sport Aviation: August 1980

I guess these pages:
p16...p24 : Fine Tuning The Electro-Fluidic Autopilot.. . by Doug Garner

Thanks,
Al

 

Wouldn't be fairly easy to detect a failed fan by the power it draws?

You could put it in a vacuum system and call it a Pirani-style pressure
gauge. The thermal conductivity of a gas depends on the pressure. Other
metals used in the Pirani gauge include tungsten, nickel, iridium, and
platinum.

 

Sorry Al. It's NEVER been on the new stand.

Sport Aviation is (one of) the private magazines published by the
Experimental Aircraft Association.

Chapters world wide - probably one near you.

Or try www.EAA.ORG ?


Richard

 

depends entirely on the fan. cheaper ones, yes. better ones, perhaps not - I
have tested Papst fans that draw the same current when stalled as running. I
have tested other fans that catch fire when stalled.

Cheers
Terry

 

I looked at a little AC drive (50kW or so) in our service dept. once, that
kept tripping on over-temperature fault. The IP56 (ish) drive was in a milk
powder manufacturing plant. The heatsink was an extrusion with 1.6mm slots
spaced about 5mm, and had 1.6mm Al plates glued in place for the fins (we
later changed to mechanical staking as it was better. And Aussie stopped
selling us the weapons-grade Al powder we had used for 10 years to allegedly
improve the thermal conductivity of the epoxy - later tests showed the Al
did bugger all). So the fin-to-fin gap was about 3.4mm. Except this drive -
the entire heatsink was a solid block of baked milk powedr & Al - water
blasting wouldnt get rid of it, we had to replace the whole damned heatsink.
Other than that, the drive worked - yay for thermal protection circuitry.

Cheers
Terry