PCI-6071E Noise Problems

[MasonJar]

I am using the latest version of NI-DAQmx with the PCI-6071E and a SCB-100 terminal board. The NI-DAQmx recognises the boards and passes self-test and self-calibrate.

I am setting up 1 channel using a K-type TC and using the built-in cold junction on the SCB-100 set on CH0 and my TC is set on CH1. So far all is well and everything is recognized and performing as it should.

Problem is.....I seem to have a 2 to 5 degree (or worse) "flutter" on the reading of the TC. Part of me thinks that possibly the gain is not setting properly to be able to read the TC as it seems that the readings are "in the mud". The other part of me thinks it could be a noise level but everything I do does not seem to eliminate or even change that "flutter". I have tried dozens of different sampling rates with the NI-DAQmx settings and nothing seems to work

I am using an Omega precision TC calibrator to simulate signals and calibrate the NI-DAQmx task. If I go into the calibration function of the program...the signal is rock solid. I may have a fluctuation of +/- .002 degrees and the calibration performs flawlessly. I have even connected the TC (24 awg wire) and gone into calibration and it reads very accurate and again...very stable. I've tried 3 cal points and have tried as many as 10 cal points and everything works great. As soon as I go back into the NI-DAGmx task and press "Test" the signal goes crazy again. I have measured the incoming signal with the Omega as well as other handheld TC meters and the signal inbound is solid.

I've been pulling my hair out on this one!! Questions....is there a way to manually set the gain on the PCI-6071E?? Amd why can't I get the actual readings to match the stability of the Calibrations??

NI's documents and paperwork seem to say that the "E" series boards are "software selectable" on gain using the NI=DAQmx but one part reads it happens automatically. I haven't been able to find any gain adjustments and it doesn't seem as though the board is adjusting auto.

Anybody run into this? All help greatly appreciated and thanks in advance!!

[LAVA 1.0 Content]

QUOTE (MasonJar @ Oct 16 2008, 08:49 PM)

I've been pulling my hair out on this one!! Questions....is there a way to manually set the gain on the PCI-6071E?? Amd why can't I get the actual readings to match the stability of the Calibrations??

NI's documents and paperwork seem to say that the "E" series boards are "software selectable" on gain using the NI=DAQmx but one part reads it happens automatically. I haven't been able to find any gain adjustments and it doesn't seem as though the board is adjusting auto.

Anybody run into this? All help greatly appreciated and thanks in advance!!

Setting your min and max values as near as possible to each other will instruct the card to use the best gain.

To verify this, read the AI.DevScalingValues property (AI,General Properties, Advanced, Scaling Values), these should change during different gains.

Also try to get your task in MAX running, this should help a lot!

Ton

[MasonJar]

QUOTE (Ton @ Oct 16 2008, 02:11 PM)

Setting your min and max values as near as possible to each other will instruct the card to use the best gain.

To verify this, read the AI.DevScalingValues property (AI,General Properties, Advanced, Scaling Values), these should change during different gains.

Also try to get your task in MAX running, this should help a lot!

Ton

Ok...now you've lost me....I am assuming the min/max you speak of is the settings in CH ai1 of the task. And I hate to sound a bit dumb here...but I have not been able to find the "AI.DevScalingValues property (AI,General Properties, Advanced, Scaling Values)" you speak of....

First off....I am using the NI-DAQmx task in MAX. Unfortunately....I am an old Dasylab/measurement computing fan where the gain is set by a dip switch on the expansion boards. The Min/Max I have set at the full span of a T type TC which is -270 to 400 deg C. I use those settings because I am using most of the span on thermal shock tests.

Next thing is the cold junction is a global virtual channel. I tried to set it up as explained by NI on their site and for some reason the cold junction channel will only work as a voltage input...as NI explains that the cold junction puts out a milivolt signal and 10 mV = 1 deg C.

All this with a linear scale formulates out....but...NI says to name the "scaled units" of the cold junction as Deg C. Because the channel is a virtual voltage channel....MAX/NI-DAQmx will only allow me to scale voltage in the cold junction.

As far as any scaling in the Temp task...there is none available. In the task.....ch1 allows only min/max settings along with scaled units that is named Deg C.

Some how....I get the feeling that the cold junction being measured in mV and then scaled over to degrees C is not working well with the Temp task on ch1. Altough I have tried a manual cold junction and a "built-in" cold junction and neither seems to cooperate. So.....I named the global virtual as CJC...set it up on ch0 as instructed and then made a linear scale (also named CJC) and then instructed the Temp task to use "channel" and named CJC.

Some how...what NI attempts to explain and what I come up with...doesn't seem to be the same.

Any ideas????

[Neville D]

What if you set the CJC manually to approx room temperature 25degC? I remember there was an option in MAX to do that.

I suspect your problems are because you have specified too wide a range of temperatures to be measured with your thermocouple. That sets the gain to make sure the whole mmt range is covered, which means none of the range is covered very well.

Also the cheaper cards like the 6071 or the USB units have a courser gain selection (or the gain doesn't go high enough), so if you fall between the ranges, and the more conservative range is selected the SNR will be degraded. The TC outputs are in the microvolts range as you probably know.

Try setting temp mmt range to say 0-100deg and see if it helps.

In the past I had problems with thermocouples and PCI daq cards, so we switched over to SCXI and the temp mmts were great with the additional pre-amp gains added by the SCXI modules.

Sorry I haven't played with DAQ in a few years but thats what I remember.

N.

[MasonJar]

QUOTE (Neville D @ Oct 17 2008, 01:18 PM)

What if you set the CJC manually to approx room temperature 25degC? I remember there was an option in MAX to do that.

I suspect your problems are because you have specified too wide a range of temperatures to be measured with your thermocouple. That sets the gain to make sure the whole mmt range is covered, which means none of the range is covered very well.

Also the cheaper cards like the 6071 or the USB units have a courser gain selection (or the gain doesn't go high enough), so if you fall between the ranges, and the more conservative range is selected the SNR will be degraded. The TC outputs are in the microvolts range as you probably know.

Try setting temp mmt range to say 0-100deg and see if it helps.

In the past I had problems with thermocouples and PCI daq cards, so we switched over to SCXI and the temp mmts were great with the additional pre-amp gains added by the SCXI modules.

Sorry I haven't played with DAQ in a few years but thats what I remember.

N.

I agree with what you're saying Neville D....

The 6071E does have 10 different gain ranges and the highest setting is 100. In the old MMC/expansion boards...all TC's were set at a gain of 200 which allowas 0 - 50 mV range. That's why I am thinking the TC's may be "down in the mud" with the 6071E.

I have tried mmt ranges of 0 - 100 and even 20 - 30 degrees C to see if it helps....and no good. I have switched the cold junction to manual and entered 25 (ambient in the shop is about 27) and still no good.

By the time the 6071E, the 100 pin cable, and the SCB-100 was paid for....it was not all that cheap. Every flyer, ad, manual, and/or brochure that NI has about the 6071E states that it easily accepts TC's and at a stable reading. Not to mention that things really get confusing when I go into the MAX/NI-DAQmx calibration and using my Omega calibrator I can run the span of the TC and everything is "rock solid" at +/- .002 app. Deg C. I just keep thinking that if the calibration can do it....why can't I???

To date...the only way I can get any kind of a stable reading is by using an averaging module in my setup. Problem is with this....you lose all accuracy of controlling due to delayed response time.

I have seen the manual/analog gain adjust of the SCXI.....I'm just skeptical if that setup will mate with the 6071E??? Or do I own a boat anchor here??

[Gabi1]

if you use instruments with low sampling rate, then this jumpy behavior will not be observed.

even if you are at worse gain condition, on the Tc you should get an accuracy above 0.5deg. if the gain is the problem, you will see a digitation of the possibles temperatures values, and not random behavior as you describe.

my explaination to your problem is grounding issues. it is known that Tc support badly grounding on one of its wire, and this will lead to strong noise at relatively high frequency. to check this, it suffice to sample at 100KHzfor example, and average out the results down to the KHz regime. if the reading becomes more stable, you have defined the problem as well as the solution. but i would go to track the grounding issue, as it will bug you forever. one thing you can try is force ground at reception (on the card), by short circuiting the negative pin. This is not the best solution, rather if your Tc is somewhere grounded, then this will stabilise output. a better solution is to purchase from omega a Tc amplifier, based on batteries (floating). then ground noise become negligible.

[Neville D]

QUOTE (MasonJar @ Oct 17 2008, 02:07 PM)

The 6071E does have 10 different gain ranges and the highest setting is 100. In the old MMC/expansion boards...all TC's were set at a gain of 200 which allowas 0 - 50 mV range. That's why I am thinking the TC's may be "down in the mud" with the 6071E.

Yes, therein lies the problem. The SCXI adds a pre-amp stage gain of atleast 500 to 1000, that can be then multiplied by the DAQ card's gain. The SCXI channels are all multiplexed into a single channel on the DAQ card so essentially you need both. But this is a very expensive option. Depending on #channels it can work out affordable if you have a lot of different channels with different (and widely differing) gain requirements.

You can use an SCXI chassis + card with any NI DAQ card that has AI channels as far as I can remember. You will need a chassis, the relevant SCXI card, cable, and breakout board.

QUOTE (MasonJar @ Oct 17 2008, 02:07 PM)

I have tried mmt ranges of 0 - 100 and even 20 - 30 degrees C to see if it helps....and no good. I have switched the cold junction to manual and entered 25 (ambient in the shop is about 27) and still no good.

Yes, it probably won't make a difference, the problem is the gain is too low.

QUOTE (MasonJar @ Oct 17 2008, 02:07 PM)

By the time the 6071E, the 100 pin cable, and the SCB-100 was paid for....it was not all that cheap. Every flyer, ad, manual, and/or brochure that NI has about the 6071E states that it easily accepts TC's and at a stable reading. Not to mention that things really get confusing when I go into the MAX/NI-DAQmx calibration and using my Omega calibrator I can run the span of the TC and everything is "rock solid" at +/- .002 app. Deg C. I just keep thinking that if the calibration can do it....why can't I???

True, maybe a call to NI support is in order? Two other options are:

1 If you have the odd TC channel here or there, get an Omega Temperature reader (about $130 if I remember right), connect the TC to those, and these usually have a 0-5V out or 0-10V output proportional to the temperature and read those in with the DAQ. Again depends how many channels you have. This also gives you a nice digital panel display of the temperature and have all kinds of fancy alarms changing the color of the temperature etc. (programming these is a major pain, but thats another issue). I have used these in the past and have worked very well.

2 Check with NI support and get a more expensive M-Series or S-Series (I'm not sure which will work better for you) with larger gain, maybe you can swap your E-Series for one of those if 30days are not up? Pick one that uses similar cable/breakout board and you can re-use all of that.

N.

[MasonJar]

QUOTE (Gabi1 @ Oct 17 2008, 04:24 PM)

if you use instruments with low sampling rate, then this jumpy behavior will not be observed.

even if you are at worse gain condition, on the Tc you should get an accuracy above 0.5deg. if the gain is the problem, you will see a digitation of the possibles temperatures values, and not random behavior as you describe.

my explaination to your problem is grounding issues. it is known that Tc support badly grounding on one of its wire, and this will lead to strong noise at relatively high frequency. to check this, it suffice to sample at 100KHzfor example, and average out the results down to the KHz regime. if the reading becomes more stable, you have defined the problem as well as the solution. but i would go to track the grounding issue, as it will bug you forever. one thing you can try is force ground at reception (on the card), by short circuiting the negative pin. This is not the best solution, rather if your Tc is somewhere grounded, then this will stabilise output. a better solution is to purchase from omega a Tc amplifier, based on batteries (floating). then ground noise become negligible.

I've stepped the sampling rate down to 20 hz and still a good 3 to 5 degree wobble. Checked all grounding issues that I can see. I've tried the filter "RC" circuit that NI tells of....and still no good. I've checked the TC's with other DAQ equipment on hand and everything seems to be fine. Checked the pc for any "inbound" noise as in 60 hz coming through power supply, etc. and all is well. I've thought about the amplifiers...but not ready to accept that answer as yet. To me....it's a band-aid and not a solution.

QUOTE (Neville D @ Oct 17 2008, 04:40 PM)

Yes, therein lies the problem. The SCXI adds a pre-amp stage gain of atleast 500 to 1000, that can be then multiplied by the DAQ card's gain. The SCXI channels are all multiplexed into a single channel on the DAQ card so essentially you need both. But this is a

very expensive

option. Depending on #channels it can work out affordable if you have a lot of different channels with different (and widely differing) gain requirements.

You can use an SCXI chassis + card with any NI DAQ card that has AI channels as far as I can remember. You will need a chassis, the relevant SCXI card, cable, and breakout board.

Yes, it probably won't make a difference, the problem is the gain is too low.

True, maybe a call to NI support is in order? Two other options are:

1 If you have the odd TC channel here or there, get an Omega Temperature reader (about $130 if I remember right), connect the TC to those, and these usually have a 0-5V out or 0-10V output proportional to the temperature and read those in with the DAQ. Again depends how many channels you have. This also gives you a nice digital panel display of the temperature and have all kinds of fancy alarms changing the color of the temperature etc. (programming these is a major pain, but thats another issue). I have used these in the past and have worked very well.

2 Check with NI support and get a more expensive M-Series or S-Series (I'm not sure which will work better for you) with larger gain, maybe you can swap your E-Series for one of those if 30days are not up? Pick one that uses similar cable/breakout board and you can re-use all of that.

N.

I think the gain is too low.....and been putting off the call to NI.....but maybe it is time. And the SCXI might be the answer. I sometimes think NI was advertising the "E" series boards as being TC compatible....but with what accessories??

As far as those panel displays.....have you ever worked with NI Dasylab?? It is super easy to learn and user friendly. Version 8 or 9 is now compatible with the NI-DAQmx. That's why I am trying the NI boards.

[Neville D]

QUOTE (MasonJar @ Oct 17 2008, 04:46 PM)

I think the gain is too low.....and been putting off the call to NI.....but maybe it is time. And the SCXI might be the answer. I sometimes think NI was advertising the "E" series boards as being TC compatible....but with what accessories??

As far as those panel displays.....have you ever worked with NI Dasylab?? It is super easy to learn and user friendly. Version 8 or 9 is now compatible with the NI-DAQmx. That's why I am trying the NI boards.

I checked the NI catalog... it doesn't mention anything about TC mmt with E-series; it does mention signal range of measurement 50mV to 10V for the E-6071, with only 12bits resolution, absolute accuracy of the range of only 15mv!! No wonder your TC's are "jumpy".

However, another solution is to get the SCC hardware and the SCC-TC series hardware to add-on to your E-Series card. Again, the chassis (SCC carrier) and modules are all separate giving you lots of flexibility for cost that is less than SCXI, but probably an additional $1000.

Or you could get M-Series Hardware.

Basically, all recommendations hinge on the #of TC channs. What are your requirements?

N.

PS I have never used Dasylab.

[MasonJar]

QUOTE (Neville D @ Oct 20 2008, 11:33 AM)

I checked the NI catalog... it doesn't mention anything about TC mmt with E-series; it does mention signal range of measurement 50mV to 10V for the E-6071, with only 12bits resolution, absolute accuracy of the range of only 15mv!! No wonder your TC's are "jumpy".

However, another solution is to get the SCC hardware and the SCC-TC series hardware to add-on to your E-Series card. Again, the chassis (SCC carrier) and modules are all separate giving you lots of flexibility for cost that is less than SCXI, but probably an additional $1000.

Or you could get M-Series Hardware.

Basically, all recommendations hinge on the #of TC channs. What are your requirements?

N.

PS I have never used Dasylab.

Some how that doesn't surprise me...

I am going to check into the SCC as well as SCXI and also try to pick at NI as to why the calibration mode of the 6071E is so smooth and stable yet it seems as though that same stability cannot be used in the "real world",

I am involved in industrial/aerospace component testing and the amount of TC's can vary from 1 or 2 to many with control outputs.

Sooo.....back to the drawing board....for now....and thanks for all the info.