NI9411 wrong frequency measurement of flow sensors / 24 impulse signals

[JohannesR]

Hello Ni Forum,

I already searched for similar problems, but haven't found anything helpful. So I'm posting our problem here. I'm apologizing in advance for bad english, it's not my mother tongue

We're encountering some strange behaviour of our NI9411 Module. We're using it in a Compact Rio with some other modules (2x Analog In).

Our goal is to measure the frequency of three flow sensors (impulse high = 24V, 0-22kHz) among with a torque sensor (TTL 5V, 30-60kHz) and a speed sensor (TTL 5V, 0-51,2kHz).

They are wired as follows:

DI0a = torque sensor
DI1a = flow sensor 1
DI2a = speed sensor channel A
DI3a = speed sensor channel B
DI4a = flow sensor 2
DI5a = flow sensor 3

Correspondig ports b are left unconnected, as it is said in the data sheet. All ground ports of sensors are connected to COM port of 9411. The module was connected to a 24V power supply.

Now the torque and speed sensor measurement works fine. But measuring the signals of the flow sensors seems to be difficult.

If an external frequency generator (impulse high about 9V, f=1,000kHz) is connected to port DI1a instead of the sensor signal, LabView shows exactly 1,000kHz. DI5a showed an avarage of 10 Hz, with a range from 10 to 60 Hz. DI4a equaled constant zero. The flickering up to 60Hz of DI5a was gone after disconnecting the 24V supply of the module. This changed nothing on the phenomenon described below.

If the signal high impulse voltage is raised, the measured signal frequency goes up to 1300 Hz @ 24V signal high impulse voltage. The signal additionally raises, if an oscilloscope is connected to DI1a, to 2180 Hz (range 2110 to 2240Hz). However, there is no change of frequency if the osci is connected to DI5a. With the 24V supply connected, DI5a showed a frequency up to 600Hz @ 24V signal high impulse voltage on DI1a, also rising when connecting the osci on DI1a. As mentioned, there was no frequency measured on DI5a after disconnecting the 24V power supply.

The signals of the generator showed some peaks on each edge (rising and falling) with values for example of 11V above high impuls voltage of 9V (20V together). That may causes that wrong frequency measurement.

The signals of the flow sensors do not have such peaks, the LabView-frequency nevertheless does not match the real frequency, as shown below (all values in Hz):

LabView		ideal values
Flow Sensor 1	Flow Sensor 2	Flow Sensor 1	Flow Sensor 2
4600	12200	740,6	6760,0
6400	13500	1481,3	7500,7
9700	16400	2962,6	8982,0
12900	19500	4443,8	10463,2

We're kind of helpless now. How can we get the signals of the flow sensors measured the right way? Are we doing something essentially wrong? Thanks in advance for offered help. I hope, I've given any necessary information. Any missing details will of course be delivered.

Puls-Measure-VI is attached beneath.

Regards,

Johannes Rost

[johnsold]

I have not used the NI 9411 so my comments are based on reading the manual.

For single-ended signals the Input high range is 2 to 24 V while the input low range is 0 to 0.8 V. The input impedance is 8400 ohms. Note that these ranges are not symmetrical and that the input impedance is relatively low, requiring ~4 mA at 24 V. I wonder if your flow sensor does not have sufficient drive to cleanly force the signal into those input ranges of the 9411?

Lynn

[JohannesR]

Dear Lynn,

thank you for your reply. The impedance was an issue that our local technical distributor already suggested to check. According to the datasheet of the flow sensor (see attachement) it can drive a current of 25mA. The signal generator we used is a test generator that can be used instead of the flow sensors to check the any evaluation unit the sensor is connected with. So it should generate signals that are equal to those the sensor would have generated. And even with this the frequencies measured with LabVIEW are not as expected.

We are in contact with the sensor manufacturer to exclude mistakes in connection on the sensor side. But the behavior with the signal generator makes us think that is has to do something with the characteristics of the NI9411 that we don't know yet.

[johnsold]

Johannes,

I do not see anything obvious in that data which looks like a problem. The sensor output should be able to drive the NI 9411 inputs.

If you have an oscilloscope, check both the waveforms and the high and low voltages.  Often some subtle problem can be spotted quickly with an oscilloscope.  Since your LabVIEW measured values are higher than expected, a spurious transistion or low slew rate are other possibilities, also more easily seen on an oscilloscope.

I just looked at the image of your VI.  It appears that you are counting every transition of the digital input. That should give you a count which is twice the number of pulses received.  Of course that would be the same on both the test generator and the sensor.

Lynn

[JohannesR]

Dear Lynn,

we checked the signals with the oscilloscope and could make out some very small peaks in the signal of the testbox, but no peak-distorted signals of the sensors. Both were measured wrong. We as well detected the phenomenon of the frequency rise of the LabVIEW measured signal while connecting the oscilloscope as described in the first post.

Something we learned from the manufacturer VSE: the low-voltage is dependent from the supply voltage and reaches 1-1,5V at voltages greater than 20V. So what does the NI 9411 if the low voltage is between the the low tolerance (0-0,8V) and the high-tolerance (2-24V) ? Does the signal detection work on edges or status or on some combination of both and gets a bit confused if the low-voltage is not or not always (the signal is overlaid by a bit of noise) within the low-tolerance?

Next step for us is to connect a pull down resistor on the signal channels of the sensors as was suggested by VSE. Maybe that keeps the low-level under 0,8V. I will report if this was the magic thing that made the sensors work with the digital module.

Anyhow thank you very much for your help so far.

Johannes

[johnsold]

Johannes,

When the specification is 0.8 V and 2.0 V it usually implies that the internal behavior is like TTL.  Original TTL had specifications like that but the "real" threshold was about 1.4 V.  Regardless, the behavior of the NI 9411 will not be guaranteed unledd the voltage swings past the limits. I think it is very likely that the low voltage being too high is your problem. If a pull down resisitor does not move the voltages low enough you may need to add a comparator with a threshold at about 1.7 or 1.8 V.

Lynn