PXI interference 25.5kHz

[pschmal]

Hello,

I have a custom hardware setup with a PXI to assist with hardware control and data acquisition. The point of the data acquisition is to read in a signal from the hardware, do a FFT, and analyze 1kHz - 256kHz. The signals are small and near the noise floor of the ADC.

I got a continuous signal at 25.5kHz in all of my measurements. I used a spectrum analyzer to figure out where this signal was coming from. I ended up taking everything apart until I was down to only the PXI. The PXI itself was generating the signal and putting it all over the place (Even the casing).

What generates this signal on the PXI? Can I turn it off? It appears the instant the PXI is powered on.

The signal is quite strong, about -70dBm at the DAQ port. You can even find the signal on the ground of the SH68 connectors.

I've tested on two different PXI systems in two different places and its the same result.

System: PXI-8101e OR PXI-8135 (appears for both, seems to be stronger on PXI-8135)

PXI - 6115 (DAQ)

PXI - 6723 (AO)

PXIe-1071 (Chassis)

[johnsold]

I do not have access to a PXI chassis but that sounds like it could be the switching power supply frequency.  Many are at higher frequencies but that is within the range of common switchers.

Do you see any harmonics? Switcher interference usually occurs as narrow spikes, often with some ringing.

Lynn

[pschmal]

There is a harmonic at about 50kHz. At 100kHz I have intereference from the environment (something pulsing like crazy) so I didn't really look to closely.

Is there a way to shield the modules from this power supply switching? If its the case, then we cannot turn it off.

[JÞB]

Turn off the lights.  Does it go away?

(Electronic ballasts in some florescent system can cause similar problems) 

[johnsold]

You probably cannot shield it. You "may" be able to sample it on a second channel and try to subtract it out but getting perfect amplitude, phase, and harmonic matching on a signal like that will be nearly impossible.

Can you amplify your desired signal?

Lynn

[pschmal]

---

@JÞB wrote:

Turn off the lights.  Does it go away?

(Electronic ballasts in some florescent system can cause similar problems) 

---

No, some of the tests were outside. I tested for environmental noises by making a big loop. I was catching some 10kHz noise on the loop that wasn't in the measurement, so if the 25.5kHz was from the environment it would show up there I would think. If the lights are corrupting the mains, its kind of disappointing this makes it through to the PXI, as none of my other measurement equipment receives that signal.

To be frank, it is disappointing that a signal measurement device has this signal corruption. It is unusable except for the most basic measurements, and people aren't going to sink this kind of money to get -50dBm noise floors. At least I've moved up a frequent flier class trying to troubleshoot this between the two locations.

[johnsold]

A few years ago I sent a $13000 oscilloscope back to a major manufacturer because the noise floor from the internal power supply with certain, not particularly rare, lead configurations was about so large as to make many of my routine measurements quite difficult.

My suspicion is that many manufacturers assign power supplies and packaging to engineers who have no experience with the kinds of measurements the equipment is designed to make and that their priorities are high efficiency and low cost, not measurement performance.

Lynn

[pschmal]

Indeed, I wish it could be better.

I'm putting a balun in my hardware and I will use differential signal lines to try and remove the noise in post processing. I'll see how it goes.