6733 analog ground

[pedrom]

I'm using a differential amplifier (AMP03) to buffer the analog outputs of the 6733 and I have noise problems (40mV pk-pk on any DC level) that I think are related to pick up due to ground loops.  I was wondering if anyone knows what is the difference between chassi ground, analog ground and digital ground on the 6733 (or any PXI modules).  

Thanks

[hoffiz]

Hi,

Unless the card is specified to have isolated outputs, inputs either analog or digital the GNDs are going to be connected to the earth ground.  Your chassis ground (i.e. earth ground) is going to be the same as the analog ground and digital ground. As a reference look at:  How to determine if an Input Signal is Ground-Referenced or Floating and High-Voltage Measurements and Isolation.

One of the best resources for noise is the Developer Zone article: “Field Wiring and Noise Considerations for Analog Signals”

Let me know if it helps

[EBL]

Hi, Pedrom.

Your AMP03 should work fine for you. Here are a few pointers:

1) The analog outputs on the 6733 are all referred directly to AO_GND. You should connect the + input of the AMP03 to your analog output of choice and the - input of the AMP03 to AO_GND. While AO_GND is indeed connected to the chassis, it is by no means at the same AC potential, and there can be a lot of noise between AO_GND and chassis ground. This noise can be induced both in the backplane ground system and between the backplane ground and the front panel. So always go directly to AO_GND.

2) The power supply that feeds the AMP03 should be referred somehow to AO_GND to ensure that the DAC outputs and AO_GND remain within the common-mode input range of the AMP03 at all times. Probably the best thing to do is connect the power supply ground to AO_GND. Keep in mind that the AMP03's load must also have a common-mode voltage compatible with the AMP03. Its ground need not be connected directly to the 6733, as long as there isn't a huge amount of noise between its ground and the 6733.

3) It's possible that the system will work better if the AMP03 power supply's ground is connected directly to the load ground instead of to AO_GND. Only experimentation will tell. In any case, if the load ground is already somehow connected to earth ground (and hence the PXI chassis ground), don't connect it to the PXI system again with another wire or you will cause a ground loop.

If this doesn't help you get rid of the noise, you might post some descriptions or images of the noise you're seeing - that will help us debug the problem.

Hope this helps,
EBL

[pedrom]

Hello and thanks for your replies.  I found that the main problem that was causing the noise was about 1ft of  unshielded and untwisted cable that I was using to bring in the power to the AMP03's from the supply.    The AMP03's work better when the supply ground is connected to load ground  (instead of AOGND) and also 0.1uF decoupling capacitors from +15 to GND and -15 to GND are absolutely necessary to reduce the noise.   

As I said before I was having 40mV pk-pk  noise, which went down to 15-20 mV pk-pk after shielding the supply cable and went further down to 2-4mV pk-pk after putting the decoupling capacitors.  

Does anyone know if this is an acceptable level of noise?  I know it varies with the application but if you could give me some examples it would help me form an idea, since this is the first time I work with electronics.   I will be using the DC levels to control voltage controlled RF oscillators.

Thanks!

Pedro

[JoeFriedchicken]

Hi Pedro,

Yes, given your noisy environment, the 2-4mV noise is expected. Page 2 of the specifications [1] shows how the output accuracy is calculated. The typical offset error is +/-1 mV, which corresponds to 2 mV peak-to-peak. The added noise sounds like it's coming from your surroundings and load.

[1] NI 6731/6733 Specifications
https://www.ni.com/docs/en-US/bundle/ni-6731-6733-specs/resource/371232b.pdf

[EBL]

Hi, Pedro.

2 - 4 mVpp of noise still sounds like a lot, though it does depend on how you're measuring the noise. Are you using an oscilloscope? What's your measurement bandwidth? Using the 20 nV/rt(Hz) noise characterization for the AMP03 and the 80 uVrms (DC to 1 MHz) characterization for the 6733, I really wouldn't expect more than 1 mVpp on a 20 MHz scope. Again, can you characterize the noise? Is it broadband random noise? Is there 60 kHz video monitor noise or power-line (60 Hz, 50 Hz) noise in there?

In any case you -should- be concerned if you're going to drive an RF VCO control voltage, because the noise will cause a considerable amount of modulation, which will show up as phase noise. If you're worried about modulation and you can afford to reduce the control signal bandwidth, I'd recommend lowpass-filtering the control voltage so that you can eliminate much of the broadband noise that will certainly be there. Set the cutoff frequency of the filter as low as you can tolerate.

Make sure there is a lot of ground around the AMP03 (preferably a ground plane) so that the bypass capacitors will have a good ground to refer to. You might consider adding a 10 uF bypass capacitor to each power rail.

Hope this helps,
EBL