Welcome to EDAboard.com

Welcome to our site! EDAboard.com is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.

Transistor flow measuremnt?


Member level 5
Sep 30, 2016
Reaction score
Trophy points
Activity points
I'm looking for a way to measure the amount of gas flowing so that it is small and cheap.
Gradually I ended up at MAF or Thermo Anemometer.
According to my experience, the heated platinum wire has a problem with impurities on its surface, plus I would have to solve how to assign the wire to the electrodes.
That's why I was interested in the solution with heated transistors
Any as this
her id schematics with my notes

How it works in my opinion
Let's leave aside the basic principle of depletion of Q1 and comparison of Q1 = Q2

If Q1 = Q2 and the system is balanced (gas does not flow)
Output 7 LM393 F is Hi
Q7 is close
Q3 is close
Q6 is Open
C1 is fuil charged

If Q1 != Q2 the system is not balanced (gas does flow and cools Q1)
Output 7 LM393 F is Lo
Q7 is Open
Q3 is Open and mae additional cuurent in Q1
Q6 is Close
C1 is disharage
On C1 is saw
And system meke 800us current puls to Q1 and min 200us pause between pulses
the number of pulses is then a function of the flow rate with the maximum frequency being 1000Hz

If so
I can only take Q1 and Q2 with R2 + R1 a Q3 for generating current pulses take some small cheap MCU with integrated comparator
for exmple STM32L011x3 in TSSOP14 and Q1 Q2 connet to internl comparator, and Q3 drive using PWM.
Am I much out?

If not have
several questions

Technically I would need to measure the gas flowing 1/4" pipe speed 0-15m/s
Physically brass T Pipe Fitting Q1 STO23 in center of flow, Q2 either in the measuring inlet or in direct contact with the brass fitting thanks to the use of the MCU, an NTP thermistor could also be placed in the measuring input and the gas temperature could be measured.

1. The 2N4401 is exceptional for this type of use, that it is used so long and I have to use it in SMD MMBT4401 or does it matter and can I use BC817-40 for example?
2. Location Q2 . Do you see how it is more convenient to use it so that it is not drained by gas but rather isolated from the body of the fitting, or is it more suitable to have contact with the body of the fitting and take the temperature of the cells of the fitting as a reference?
3. The use of a processor brings much greater variability in the control and calibration of results. If you have any ideas, I'll welcome him


  • Fig_1_Eng.gif
    32.7 KB · Views: 20


Advanced Member level 5
Mar 4, 2008
Reaction score
Trophy points
Activity points
I expect that a heated-transistor sensor would
have the same problem as a hot-wire MAFS
regarding contamination. That is, it changes
the surface and the surface's thermal resistance
to the air stream.

But in automotive applications, tuners know
to use some QD cleaner (oil-free electronics
contact cleaner) to "reset" any such cal drift
by removing the cause.

If your airflow of interest is similar to engine
airflow, then an automotive MAFS is certain
to be your least expensive, most reliable
option. Not necessarily the nicest interface,
many are a nonlinear frequency vs airflow.
But the cal-curve is pretty well known and
I have been told by one testing house, that
they found group tolerances in the 5% range.

A MAFS meant for a SMART car or motorcycle
would push the low end down quite a bit
relative to the V8s I was playing with. But
the low-flow fidelity has its own problems
such as MAFS body thermal conduction,
picking up or shedding heat from / to the
engine bay or whatever your environment
may be.

There are ball-in-tube flow meters for
welding gas and medical applications
which you might find suitable, add your
own instrumentation to read the ball's
position in the column. Inductive pickup
perhaps, calibration would be your

For ultimate smallness, Doppler flowmeter
can measure through the tube wall of
"whatever it is. That might actually be a
not-that-bad design project, you need
an oscillator, couple of acoustic transducers
and a mixer, from which you'd pick off and
measure the beat frequency (cal, again,
is on you although you might find some
references on Doppler shift vs tube diameter
and velocity).

Now if you want true mass, in a lot of cases
you will also need temperature and maybe
even ambient pressure. Here again automotive
parts are a good play, if you have some sort
of "smarts" attached to cal-map the individual
inputs to real values.


Member level 5
Sep 30, 2016
Reaction score
Trophy points
Activity points
Thank you, I know the principle of ultrasonic anemometers,
As long as I developed one, but that would be a different story.
It is not suitable for this purpose in terms of both price and size. I need to measure the flow in a tube below 1/2 ", the price can fit in about $ 10

I will return to the idea of a thermo anometer with a cooled transistor.

Guiding principle
We have two identical transistors
one Q2 is connected as a diode (C and B connected) and is not cooled by a flowing medium ie. the PN junction temperature is the same as the ambient temperature
second Q1 it is cooled by a flowing medium and by means of a stream Ic it is heated to a temperature approx. 50C higher than the temperature Q2 . If it is not cooled, it should have a temperature 50C higher and should apply

Q2 UBE + U R13 = UBE1 if Q1 is 50C warmer Q2

Q2 UBE is 100mV lower becoause PN has a 50C lower temperature and the temperature coefficient is about 2mV / K
Q1 UBE is is higher because the current Ic is about 100 times higher than Ic Q2

Practically, at zero medium speed, the current at which the current must be found must be used with the potentiometer 10k (R8)
will be Q1UBE = Q2UBE + U onR13. This should occur when the temperature of Q1 is 50C higher than the temperature of Q2 (environment).
The flowing medium Q1 cools and thus rises Q1UBE and the energy supplied for heating Q1 is proportional to the cooling rate and thus the speed of the medium that cools Q1

It is so?

If so I can use something like this


The connection of Q1 and Q2 is practically the same, I only added Q3 allowing to turn off the heating of the transistors, ie low enrgy standby mode
I added a thermistor RT 1 to measure the medium temperature, maybe it fits when calibrating the pluis informs about the medium temperature
Transistor Q4 for generating pulses for heating Q1 is also the same, only it is not switched by the selection comparator, but by the MCU using PWM
R3/R6 used to measure the supply voltage and refine the temperature calculation.

VQ1P a VQ2N is compare on internal comparotor in STM32L011, If they are unbalanced
MCu genrrate on PWM short puleses which heat Q1.
The question will be how to choose control, whether to use PID or just count pulses and wait for a response in the form of balance?
In any case, the magnitude of the current through the potentiometer R10 5k should be such that the power dissipation on Q1 you at about 1000 pulses per second should cover the heat losses at full flow, ie the maximum speed of the medium.

In the afternoon I'll find some NPNs and try to warm up and see what she did.


Super Moderator
Staff member
Apr 1, 2011
Reaction score
Trophy points
Minneapolis, Minnesota, USA
Activity points
I have a thermistor purchased at Radio Shack for two dollars. Its resistance at room temperature is 10k. When I pinch it between my fingers, it drops within seconds to 7k. That's pretty sensitive range of response.

The heated transistor system appears ingenious, and it will require proper adjustment and calibration. However I can't help thinking that 2 thermistors can also be made to work. (One as a reference against which the other measures.)

As with a transistor, you can try heating a thermistor slightly to watch how gas flow affects it.

LaTeX Commands Quick-Menu:

Similar threads

Part and Inventory Search

Welcome to EDABoard.com