How much noise expected in analog output?

[Kevin Archie]

I'm using a PCI-MIO-16E-4 with LabView to generate buffered analog output, and at the BNC-2090 output (DAC0/1OUT) I'm finding ~180 Hz hum of about 20 mV peak-to-peak. It doesn't seem to matter whether I'm writing a waveform or not; the noise magnitude is always about the same.

Is there something I can do to eliminate or reduce this, or is it typical so I'm stuck with it? I futzed with the BMC-2090 shield ground jumper (W1) but none of the three documented settings made any difference.

Any suggestions would be greatly appreciated.

thanks,
Kevin

[<Mike Ross> mross@smpcorp.com]

No particular noise is "typical." No noise at all is atypical. 20mV is low enough that it could be radiated noise for nearby equipment. I doubt you card would have that noise in the lab situation where they test for noise. Which essentially means something about you estup is responsible for the noise.

Since the frequency is a multiple of 60Hz it is probably environmental noise, radiated ot picked up off of the ground. Really and truly, 20mV is not a lot. I have acquisitions where I must be very careful in grounding ALL the associated equipment or I see huge blurts of noise from a motor drive - on the order of 3 Volts until I drove an eight foot gound rod through the floor to get an isolated safety ground for my DAQ PC and scope. I still see the noise, but it is now lower than 0.8V so I don't get false triggering of TTL circuitry from it. This is AI not AO like you are discussing.

If you check the specification of your card there is a table on "System noise." I am not saavy enough to understand it, but I see that the noise is dependant on the output range. I suspect that the range is not a user selectable item, but in theory if you see more noise than the spec sheet says, then it is part of your setup not the card.

Here is part you really don't want to hear: Solving a noise problem by text messaging is like solving a mental illness over the phone, maybe worse. I can tell you all sorts of practical, common sense things that are not specifically helpful. I can tell you from personal experience that you really need to follow all the good practices you can, if you really want to have as high a SN ratio as possible.

Can you set up a single point ground? Can you eliminate all ground loops (shielding only grounded on one end.) I had a stray strand of shielding that brushed against a case that made terrible noise once.

There is a good book called ,"Grounding and Shielding Techniques," by Ralph Morrison from Wiley Inter-Science. There are papers on this in in the tech resources of ni.com. In particular the paper on "Field wiring" is good. Keithley has a good book they will send you for free on low level measurements that is very informative. You have to call them on the phone to get it.

[Kevin Archie]

Mike, thanks for your response. We're doing electrophysiology (small signals) so we're pretty careful about grounding etc. The next instrument down in the rack (an AxoClamp 2B) generates outputs that I'm measuring with apparent noise level << 1 mV, and a few spaces down there's an AC amplifier with input signals around 10 uV and no obvious n*60 Hz hum. From the behavior of other instruments in the rack, I don't think we have a general problem with grounding or radiation. (Though I haven't tried moving the BNC-2090 to a different location in the rack. Maybe I'll do that next.) I've pretty much reduced it to: I'm doing something wrong (grounding seems the likely culprit) with the DAQ hardware, or that level of noise is unavoidable from this hardw
are.

And of course you're completely right that solving noise problems can't really be done remotely. Just hoping maybe this specific situation rang a bell with someone.

thanks,
- Kevin

[&lt;Mike Ross&gt; mross@smpcorp.com]

Well, The AxoClamp is a $9.500 piece of specialize dequipment so it ought to be better that the PCI-6040E which is sort of on the cut rate side of all purpose equipment. Your AC amp with no hum is also probably more specialized and expensive also.

It is quite possible that your DAQ card isn't going to function as a function generator in the way you need it to. You might want to discuss this with NI salepeople since neither you nor I can make anything of the noise spec for your DAQ card. I presume NI has something that is capable for your needs, but a really good function generator and amp combo might be what you need.

Just to see, put a scope on the pins of the cable that carry the output signal and see what noise is
present there. You can probably open up the 2090 and figure this out without too much heartache.


You really might want to get that Keithly book, it is VERY good info on low level signal work.

[Alan Armstead]

The output accuracy of the PCI-MIO-16E-4 is just over 8mV, so 20mV peak-to-peak is above expected. I have a couple suggestions. First, you can run the online E-Series diagnostic on your board at www.ni.com/support/selftest to make sure your board still meets specs. Second, depending on the frequency you want to output, you can build a very simple lowpass filter on your BNC-2090 to get rid of your 180 Hz noise. There is detailed instructions to do this in the product manual for the BNC-2090: http://www.ni.com/pdf/manuals/321183a.pdf
I hope this helps.

-Alan

[&lt;Mike Ross&gt; mross@smpcorp.com]

Kevin,

Just out of curiousity, I put my scope on a lead acid battery with a light load. Set it to record peaks and saw lots of hash, 250mV. Probe on skin gets 500mV of noise. Your job must be very interesting if you are sweating 20mV.

Mike

[Kevin Archie]

Hi Alan -

Thanks for the pointer. The board passed the specs just fine. Building a filter is a possibility; I'll just have to decide whether I want to do this that much.

thanks,
Kevin