Hello everyone,

I am currently working/struggling to get a thermal mass flow sensor (IST FS7) to behave like intended, regarding the response time.
The sensors datasheet states that is has response time of t66% = 200 ms going from a low air velocity to a higher air velocity.

I have build the circuit the manufacturer provided, but this was described as rather suggestive.
With testing this circuit the response time from low to high air velocity seems to be like the datasheet stated.

But for my application (which I can elaborate on if interested) I want to measure the flow from a high to a low velocity. With a high response time. From high to low velocities it can take about 5 seconds to reach its final voltage!

I am a mechanical engineer and have learned a lot about circuit in this project. But if anyone with more knowledge than me about circuit and sensors could nudge me in the right direction it would be really appreciated

Thank you!

 

Welcome to AAC.

According to the datasheeet, the response time of the sensor is very low. The circuit doesn't appear to have anything that would create such an extreme slew.

In order to eliminate the most trivial possibilities, is it possible that your apparatus is actually taking 5 seconds to settle on a flow rate? I ask because troubleshooting the sensor-circuit part of the system would be a useless exercise if it is completely operational.

Is there any test, independent of your apparatus, that can verify the high slew is in the sensor-circuit subsystem and not somewhere else?

 

Thanks for the quick reply Ya'akov!
Also thanks for looking at the circuit.

As to my understanding, a t66% of 200 m/s is relatively quick for air flow sensors (although not a lot of literature is available on this characteristic of flow sensors). If you can disprove this please do so

My test setup do contain a few parts which could possibly result in false findings.
I will simplify my test setup. I will give a burst of air on the sensor and plot the voltage reading.
From here we can see if the increase in voltage is much more steep than the decrease.
I will send this to you in a few hours once I am done.

Greats Wybe

 

Hi Wybe,
What voltage do measure at the Vout, with no air flow ?

E

I assume Vcc =10v?

 

Here is what happens when I blow on the sensor for a really short time...

sorry for the lack of time label. Take every time step as 0.01 second (10ms)

 

Here is what happens when I blow on the sensor for a really short time...

sorry for the lack of time label. Take every time step as 0.01 second (10ms)

Aside from sensor, circuit, and mechanical fixes it seems like you might be able to characterize that decay and extrapolate. Of course the efficacy of that approach would certainly be heavily dependent on what and why you are measuring.

It also strikes me that reducing the constant temperature might reduce the slew. At the very least if you get exactly the same curve with less heat, I think that would point somewhere else than the sensor. This is predicated on the idea that less heat would require less convectional cooling to respond.

I have to say, that‘s surmise, and I could have it wrong.

 

Hi Wy,
By blowing into and stopping, you are not flushing out of the measuring cavity the residual blown air, so it will be a slow decay response.

What voltage Vout range can you get by setting the 500R pot over its full range?
E

 

Hi Wybe,
What voltage do measure at the Vout, with no air flow ?

E

I assume Vcc =10v?

4.25V at Vout.

However this is the response when I blow on it quickly. I find it pretty interesting how slow the Vout returns to its base value. What is your opinion? @ericgibbs @Ya'akov

 

Hi Wy,
By blowing into and stopping, you are not flushing out of the measuring cavity the residual blown air, so it will be a slow decay response.

What voltage Vout range can you get by setting the 500R pot over its full range?
E

I can't test the full range of the the 500ohm pot because lower it too much would exceed the voltage the sensor is allowed to get.

 

Hi Wy,
I understand.
As you know, the mass flow measurement is done using a PT1500 RTD and heating element, so the signal response maybe slow, especially when the blown air is stopped.

I have run a simulation, very rough tests of the circuit, modelling the thermal inertia of the mass is difficult.

Ignore the below swing, it is due to the simulated air pulse.

E

 

Hi Wy,
I understand.
As you know, the mass flow measurement is done using a PT1500 RTD and heating element, so the signal response maybe slow, especially when the blown air is stopped.

I have run a simulation, very rough tests of the circuit, modelling the thermal inertia of the mass is difficult.

Ignore the below swing, it is due to the simulated air pulse.

E
View attachment 299516

WOW this is so cool!!!

 

hi Wybe,
How do you plan to display the air flow rate?
We could test your circuit in simulation.
This sim is revised to show a slower decay rate on the Vout fall, suitable for sim testing.
E

 

hi Wybe,
How do you plan to display the air flow rate?
We could test your circuit in simulation.
This sim is revised to show a slower decay rate on the Vout fall, suitable for sim testing.
E
View attachment 299539

Sorry for the late response I did not get a notification. What exactly do you mean by 'display'? If you mean the conversion of voltage to airspeed I can give you the equation.

 

hi,
How will the user/operator read the MAF value?
E

 

Hello,

I am working on a fast response airflow measuring system.

However I have discovered the output of the LM358 op-amp I use takes a relatively long time to go from a high to lower voltage (this voltage changes is induced by a change in airflow).

I found an article by TI about op amp not being able to go to true GND by nature. This article also provided a way to work around this, would this resolve my problem? Do you have other suggestions?

Picture 1: Red = op amp, Blue = sensor output. First pulse is lowering airspeed, second pulse is increasing airspeed


Picture 2: The circuit


Thanks for helping!!!

 

hi W,
Why do think that the LM358 will not give an output close to 0V is relevant to your application, when your zero flow point is referenced to +3Vout and the MAF signal is a positive increase in Vout.

E

 

hi W,
This LTS sim shows the TLV and the LM358.
I have used a linear flow versus MAF sense in order to show the operation of the two OPA types.
The difference is minor in their responses.

Consider your concern regarding the LM358 Vout not getting close to zero volts.

If Vout approached zero volts, the sensor heater would not be powered.?

As the App notes say Pt1200 is a nominal 1280R @20C and the heater resistance is approx 51R

For calibration, Vout is set for +3V
E

 

hi W,
This LTS sim shows the TLV and the LM358.
I have used a linear flow versus MAF sense in order to show the operation of the two OPA types.
The difference is minor in their responses.

Consider your concern regarding the LM358 Vout not getting close to zero volts.

If Vout approached zero volts, the sensor heater would not be powered.?

As the App notes say Pt1200 is a nominal 1280R @20C and the heater resistance is approx 51R

For calibration, Vout is set for +3V
E
View attachment 299791

Ah thank you so much, I was almost going to buy another op-amp because I thought this was the problem. However, in the datasheets of both op-amps I couldn't find a significant difference. Your simulations support this so that is amazing!

 

Hi Wy,
With regard to the slow decay of Vout when you stop blowing, if you consider how the sensor is activated, it is what I would expect.
When blowing, the PT1500 senses the drop in temperature and drives the heater hotter until the PT1500 reaches equilibrium
When you stop blowing, the heater has no cooling air flow, and so it slowly cools down to ambient.

E

Added Power dissipation plot of the Heater and PT1500