Noisy and Unexpected Readings

[Gurdas]

My Problem: Diff and NRSE connections show bad readings and RSE shows better (but still not desirable).

The setup: I have an LVDT connected to a DRO (digital readout). The DRO has an AO which I connect to my PCI-6034 thru a CB-68LP. The DRO output is a 2 wire system. One wire carries the signal while the other is soldered to its body and ground. The DRO is powered by line AC.

DRO output is in mV (± 50mV).

The results: When the DRO is set to give 0mV on its ouput this is what I observe in MAX:
a) RSE mode: signal varies between ±3mV (instead of a stable 0 mV).
b) Diff and NRSE: signal varies ±70mV (instead of a stable 0 mV).

Other points:
1) It has been observed that when I use a multimeter on the wires coming from DRO outp
ut (or touch with my fingers) the reading on the DRO display gets affected. I do not know the output impedance of the DRO.

2)I used bias resistors (50K) for Diff and NRSE. There was an improvement but still far worse than RSE.

3) The 5V output on the CB-68LP when re-fed to any of the AI does not show as exactly 5V but shows some mV varying differences.

I am using LV-PDS 6.1 and NI-DAQ 6.9.2 on a Windows 2000 professional machine with Celeron 1.6 GHz and 256MB DDR RAM.

Thanks in anticipation!

- Gurdas
gurdas@qagetech.com

[RIO-Todd]

Gurdas,

I've tried to provide some suggestions below.

1.) Does this seem to happen on every channel or have you just tried it on one channel?
--The SCB-68 has a temperature sensor that can be routed through channel 0. Make sure that if you have your signal connected to channel 0 that you have the switches of the SCB-68 configured as to have the temperature sensor turned off. The SCB-68 user manual provides you with the information of how to set the switches of the SCB-68.

SCB-68 68-Pin Shielded Connector Block User Manual
http://digital.ni.com/manuals.nsf/webAdvsearch/74C86ADEF0E4813F86256C84007CB3AB?OpenDocument&vid=niwc&node=132100_US

If you have only tried one channel then I would suggest to try other channels and see if you see the same thing.

2. How are you connecting the differential signals?
--When configured for differential, you will need to connect the positive terminal of the input to CH X and the negative terminal of the input to CH X+8. For example, if you want to connect differentially to channel 1, you would connect the positive terminal of the input to CH 1 and the negative terminal of the input to CH 9 (1+8). You are probably doing this but I just wanted to make sure.

3. What happens if you internally reference the 5 V in Measurement and Automation Explorer (MAX)?
--If you run a test panel in in MAX (right-click on the 6034 within devices and interfaces and select Test Panels) you can select which channel(s) you want to look at. You can also use special channel strings to reference AI Ground and the 5 V reference. Type in "_Ref5V" for the channel in the test panel and see if you see a solid 5 V signal. The document below tells of other channel strings you can use.

Internal analog input channels for self-diagnostics or the monitoring of analog output voltage
http://digital.ni.com/public.nsf/websearch/6C77B1884F5BBEE8862569E3005BE30E?OpenDocument

If you don't see a solid 5 V signal, then there might be a problem with the hardware. You could try self-calibrating the device from LabVIEW (see link below).

How Do I Calibrate My E Series Device in LabVIEW?
http://digital.ni.com/public.nsf/websearch/51A15DD0CE9491B586256A1D00631940?OpenDocument

I hope one of these suggestions help.

Regards,

Todd D

[Gurdas]

Thanks Todd!

I am using CB-68LP and NOT SCB-68.
I will try your suggestions and let you know.

Previously, I connected the 5V reference signal to AI ch0 and checked results in the test panel. What I got was a slightly fluctuating (some mV) signal close to 4.95V. What could this mean?

I'll try the self-calibration VI and the _Ref5V methods you've suggested.

What really spooks me is that I get far better results in RSE than Diff. or NRSE!
How/When does this happen?

Hope I can solve this!

Best wishes,
Gurdas
gurdas@qagetech.com

[RIO-Todd]

Gurdas,

I also find it odd that RSE is more accurate and less noisy than differential. Differential measurements should reduce noise and thus be more accurate.

Could you measure the source impedance of your signal with a DMM? Usually it is a good rule of thumb to make the bias resistors at least 10 times the impedance of the source. One thing I would suggest is to try a larger value for the bias resistors. Maybe go up to 1 Mohm and make sure that you've connected bias resistors between CH X and AIGND and between CH X+8 and AIGND. Using two bias resistors will assure the signal path to ground is balanced.

I hope this helps.

Regards,

Todd

[Gurdas]

Todd,

I tried the 21290 channel setting in DAQ and I think we've zeroed on the problem.

The customer has two computers. On one of them (new and better grounded) the internal 5VDC was shown as a solid 5VDC with minor fluctuations once every few seconds
On the computer where the said system is installed, the 5VDC is displayed as a noisy signal with ±3mV fluctuations.

We would be working on the grounding of this PC tomorrow. Shall let you know how it goes.

However, I am still baffled by the poorer Diff. readings vis-a-vis RSE.

Thanks once again.

- Gurdas

[Thijs]

Hi,

Not new in this subject, But I am as well very curious why I get so much noise (a perfect 50Hz noise from probably the lights in my office) when I connect an external 5V DC power supply to my SCB-68 connection box to my laptop using a 6062-E DAQ card.

Using RSE on ACH1(+5V) and AIGND (0V) I have no noise, a perfect straight line.

Using Diff on ACH1(+5V) and ACH9 (0V), and using 68kOhm bias resistors between ACH1 and AIGND and ACH9 and AIGND, I have noise in my signal (from a 50Hz source).

Also strange: if I disconnect my laptop from its external power supply, so running on batteries, the noise is reduced, but still apparent.

A simple solution would be to use always RSE, for all my sensors. However, this is contrarely to what NI recommends, and I have some cases in which I want to measure a DIFF signal, for example, I want to connect a K-type thermocouple. When I connect this to ACH1 and ACH9 (reserving ACH0 for the onboard CJC sensor), I get large fluctuations (seems like white noise).

How to overcome this problem? I cannot select RSE for my thermocouple using DAQMAX. Is point-by-point averaging the only solution?

Any ideas on this -apparently existing- topic are highly appreciated!

Thijs

[Laura F.]

Hi Thijs,

If your external 5V power supply is ground-referenced, when measuring this in differential mode, you should not use bias resistors.  There is a great section about this in the Field Wiring and Noise Considerations for Analog Signals Tutorial.  You can also see information about why the measurements are expected to be less noisy when using a laptop battery in this KB. 

When you connect your thermocouple, what is it touching?  Is it just measuring air temperature?  You can implement a low-pass filter on your thermocouple measurements to get rid of the white noise. 

Hope this helps,

Laura