Multifunction DAQ

cancel
Showing results for 
Search instead for 
Did you mean: 

volt divider noise

I had some really handy 10/1, 1meg voltage dividers built into BNC couplers. They have BNC male to attach to my NI-9215 USB sample and hold AI modules, and BNC female to attach the signal.
 
Bad Graphic, sorry
 
BNC M<-- |
                |
      {100K |  1meg}
  ---{\/\/\---------\/\/\}-----<BNC F
  |   |                    |
 _  _                   _
  -  -                    - Shell Common
 
I get the attenuation of a DC signal I expect, but the ripple on the signal increases by a factor of about 10 to about 10mv. I did not expect this. What am I doing wrong?

Message Edited by Jim@SONGS on 07-05-2007 11:42 AM

0 Kudos
Message 1 of 9
(3,684 Views)
New information. The ripple only occurs with a USB AI device. I tested using a daqcard-6036E and have no problem.
0 Kudos
Message 2 of 9
(3,657 Views)

Hi,

Do you have screenshots of the ripple you're seeing? I'm a little unclear, is this occuring on a purely DC signal or something changing (square wave)? If you run the acquired signal through an FFT, does it seem to be at a certain frequency (powerlines, other sources of noise...)?

Also, you mentioned that the ripple increases by a factor of 10, is that with respect to without using the attenuator or something else?

As for what is causing this, I can think of a couple possible sources though at this stage I would consider them guesses.

Impedance mismatches - the 9215 has and input impedance of 200 kohm between ai-'s (they share a 100kohm ground reference) do you see this with only one signal connected?

External Noise - are there any large noise sources near bye?

Hope this helps, please post back with some more info and we'll see what's up.

Cheers,

Andrew S

0 Kudos
Message 3 of 9
(3,647 Views)

Andrew

The quick answer is the ripple appears on top of a pure DC test signal from the daqcard AO. I was calibrating the new dividers when I recognized the noise exists. I have no FFT, but it is sinusoidal with a couple of frequencies. The predominant looks like 250Hz doing a simple pk-pk delta time.  It also contaminates a field sensor signal I tested in the field. If I look at the signal w/o the divider the ripple is >2mv, with the divider about 10-15mv. Same signal input, or even with the input cable shorted to get a 0VDC input.  As I said in my previous post, it is only significant when I use the USB-9215, not when I use the daqcard-6036E AI. I will try to get a screen shot next week.

0 Kudos
Message 4 of 9
(3,639 Views)

Hi Jim,

It seems like even the ~2 mV ripple you say you are seeing without using the voltage divider is a problem in and of itself.  This seems much too high of a ripple.  Could you tell me, what is the input voltage range that you are using on your device? The spec on the 9215 for peak-to-peak input noise is 7 LSBs peak-to-peak.  So, if you were using the max input range of -10 to +10 V, I would still not expect more than a few tenths of a mV of peak-to-peak input noise on the signal.

So I guess the question is, when you say the ripple is >2 mV, do you mean it is just slightly less, or an order of magnitude less?

It would be helpful if you posted an image of what the signal looks like.  If you could run an FFT on the signal (if you have LabVIEW there is an Express VI to do an FFT so you could make a really quick program to do this) that would help a lot for us to know exactly what frequency the noise is occuring at.

I still cannot say for sure what could be the difference between the 9215 and the 6036E that could account for this.  I noticed that the input impedance is 100x greater on the 6036E, but it still seems that it is large enough for both devices that it shouldn't make a difference.  I will look into the specs for the two devices some more and see if I can find anything else that might give us a clue as to why this could be happening.

0 Kudos
Message 5 of 9
(3,603 Views)

Here is a pdf with divider and without divider. At low sample rates the noise is not as bad. I need to sample at 100K as shown. You can see in the pdf fft there is not a tonal frequency just higher baseline.

0 Kudos
Message 6 of 9
(3,590 Views)

Hi Jim,

Thanks for posting that image.  After seeing it, I have to admit you have stumped me as well as several of my colleagues!  I would like to try to reproduce the problem you are seeing.  Could you tell me, what terminal block are you using with the E series card?  Also, would you happen to know the part number or the manufacturer of your probes?  If possible I would like to see if I can find any specs on the probe.

Sorry I can't be more of a help at this point.

Smiley Indifferent

0 Kudos
Message 7 of 9
(3,556 Views)

Hi Elizabeth

I had sent the info to Support@ni.com and I will repeat here.  E-series term block is a CB-68LPR. I am not sure to what "probes" you refer. My signal source is a rectangular 9VDC transistor battery, slightly worn out. All else are simple BNC cables and my voltage divider BNC coupler.

Jim

0 Kudos
Message 8 of 9
(3,511 Views)

Hi Jim,

I did speak to the engineer Michael who was helping you with this issue on the phone.  We are working on seeing if we can recreate this problem.  Either Michael or I will let you know what we find out.

Thanks for your patience Smiley Happy

0 Kudos
Message 9 of 9
(3,478 Views)