Multifunction DAQ

cancel
Showing results for 
Search instead for 
Did you mean: 

9213 output is dependent on sampling frequency

Highlighted

Hello;


I have an issue whereby it appears that the temperature reading of my 9213 is dependent on the aggregate sampling frequency of my acquisition task. Specifically, if I sample a single channel at 1kHz (high speed, no CJC, no autozero), the response I get is different than if I sample 3 channels 200Hz (high speed, +CJC, +autozero). I've been trying to tackle this problem on and off for the past few months, but still can't figure out it out. Frankly, I am unsure whether this is a NI hardware issue, a Labview software issue, or an experimental configuration issue.

 

As an example of my problem, the following plot represents identical process heating conditions, except for minor differences in intial temperature. You can see that when the heat source is turned off, the response is very different: (The name convention in the legend is  <number of channels in task> <sampling rate> <if CJC and autozero used> . <aggregate sampling rate, assuming CJC/autozero are channels>)

june16_2016.jpg

What is really strange is that the 200Hz tasks will track back to the same cooling curve after some delay time. (green and orange curves are overlapped)

 

Although my notes aren't as good for this experiment, I see a similar effect when I sample a single channel (no autozero, no CJC) at different frequencies, using identical heating schedules:

singleChannel.jpg

This particular graph shows how the TC reading will effectively 'top out' at a certain temperature, Tmax, then track the cooling profile once the workpiece temperature is below that threshold. This result is from the same experiment, except I used 3 channels sample at 200Hz each

3chan.jpg

In the single channel case, fs =200Hz, Tmax = 350C. Yet in the 3 channel case, fs =200Hz, fs_aggregate = 600Hz, Tmax=800C


System details:
Labview 2013(64bit) + Windows 10

Max 14.0.0f0

NI USB-9213


Experimental Details:

K-type thermocouple, 24AWG, no GND shield, spot welded to workpiece using micro-TIG welder

length of cable: approx 6' + k-type vacuum feedthrough

task: always configured to high-speed. The problem is much much worse if using high resolution mode

workpiece is heated with an electron beam, which may cause charging/voltage floating effect

Configuration: I configure the tasks properties in MAX (continuous sample, TC input range= [0,1300]C, sampling rate, CJC/auto zero/ADC timing mode)

I've attached a copy of my code to look at, as well as the .nce export for the relevant task configuration.

 

Frankly, I am baffled by this issue, and still have to do some other experiments to try and understand why this is happening. Specifically, the TC response at low sampling frequencies seems to be sensitive to when the electron beam/heat source is turned off (see second figure), which makes me think that this may be related to charging of the workpiece. But I have gone to great lengths to ensure the workpiece is well grounded, so I'm not sure what exactly is happening. Similarily, doing these measurements in high resolution mode makes them completely wonky (reads 40C, while workpiece is glowing red), so maybe it is somehow related to the input impedance or the open thermocouple detection current circuit. 

I am intending to do some experiments using a high temperature furnace, and seeing if this problem is related to the electron beam specifically.

Anyways, any assistance would be greatly appreciated and ideas are more than welcome!

Thanks

-Paul R Carriere

Download All
0 Kudos
Message 1 of 11
(2,344 Views)
Highlighted

Hi Paul,

 

It looks like you are sampling much faster than the 9213 is spec'd for. The 9213 supports up to 75 S/s aggregate over all channels. If you sample faster than this, the readings you are going to get are probably going to be erroneous. Try slowing down your sampling and see if you see the same behavior. 

-----------------------------------------------
Brandon Grey
Certified LabVIEW Architect

0 Kudos
Message 2 of 11
(2,301 Views)
Highlighted

Hi,

 

I have an issue simillar to Paul has.

 


Experimental Details:

Labview 2011(32bit)

J-type thermocouple, 24AWG, no GND shield, length of cable: 2m 

1s/s

continuous sample

Samples per channel 10

Using 14 channels

CJC built-in

Sampling mode: High Resolution

AutoZero mode: Every sample

 

I've tried with every the sampling & autoZero mode, available, on High Freq It works worst.

 

TC_Graph.png 

 

 

 I'll try with a lower sample rate, to be sure it is not related to the sampling rate.

 

Any help or comment will be valuable.

Thanks

 

Alfredo B.

0 Kudos
Message 3 of 11
(2,257 Views)
Highlighted

Hi Alfredo;


You graph there is pretty hard to understand, specifically since you have multiple y-axes. Are you applying a know temperature ot the sample, and still getting all those different outputs, or are those potentially different temperatures. What exactly is your problem? 

Right away, I see 1S/s, with 10 samples per channel, which means you have a refresh rate of 10/1 = 10 seconds

 

Let me know;

-Paul 

0 Kudos
Message 4 of 11
(2,237 Views)
Highlighted

Hi Paul, 

 

I'm testing motors on locked rotor, so, I apply voltage to the motor to turning on.

The red line which is around 3.2A (first picture) is the current across the motor's winding, the motor has a thermal protection that's why you can see the currents suddenly dorps to zero and the motor start go be cool.

the equipment is an a laboratory, with a temperature controlled at 25°C.

 

The rest of the curves are the temperatures on differents position inside de motor, some are over the windings (isolates), chasis & bearings.

 

When the motor is excited, It is start to be heated because of the current and across it and the locked rotor.

 

 

The measurements must be like the picture I'm attaching. The Graph was took with the same equipment, with some motor works fine & With some others works like the first one picture.

 

 

Good Graph.bmp

 

Any comment it will help me.

 

Thanks.

0 Kudos
Message 5 of 11
(2,230 Views)
Highlighted

What is the resistance from your rotor to ground? How is you NI9213 isolated (floating or COM connected to GND)? You TCs are electrically connected to your rotor? Is it possible that the current return path for your rotor is not consistent/good. 


Just so I understand Hugo, on the first graph, you expect all the readings to drop when the current is turned off, correct?

 

I think both our problems are related, and somehow related to charging....

0 Kudos
Message 6 of 11
(2,224 Views)
Highlighted
Yes, all the readings should drop slowly as the 2th image.
Currently the COM is connected to GND...

0 Kudos
Message 7 of 11
(2,215 Views)
Highlighted

And the rotor is connected to GND as well? Is it a 'good' ground?

In my case, my COM is NOT connected to GND, since I found that made things worse. I need to re-examing that though. 

0 Kudos
Message 8 of 11
(2,212 Views)
Highlighted
The motor is conected to GND also.
0 Kudos
Message 9 of 11
(2,207 Views)
Highlighted

Just to be clear 75 S/s aggregate over all channels for the NI 9213 seams to mean I can sample at 75 hz while reading all 16 channels. But from what I've read on other posts it means 75 S/s aggregate over all channels means I can only sample at 4.68 hz while reading all 16 channels. Can you clarify please?

0 Kudos
Message 10 of 11
(217 Views)