60Hz Noise

[cochenob]

In a previous post, I mentioned that some IQ measurements I was making were suspiciously noisy.  It turns out, if I acquire data long enough (say, 1 second) that I and Q had what looked like some 60Hz modulation on them.

I verified this by connecting the upconverter directly to the downconverter, and using the 5660 demo panel, narrowed in on my band of interest, and did infact see some sidebands 60Hz away from my carrier, about 45dB down.

Not knowing whether it was the upconverter, downconverter, or both that was the culprit...I drove the downconverter with an independent source with similar results. 

While the laboratory space may not be the most benign environments, I was surprised that that amount of 60Hz noise was making it's way into the measurements.  My test set up is rather simple.  I simply connected the two devices with a small, sheilded SMA cable, so I don't think that it's getting in through my 'channel' or DUT, but rather it's inside the chassis itself.    Or perhaps its coming in through the power lines.  I may post this over in the PXI hardware forum...however I was curious as to whether or not this type of behavior was "typical" in the RF hardware.  There don't seem to be any specs relating to the susceptability of the oscillators to this type of interference.

---
Brandon

[Andy_Hinde]

Hi Brandon,
This is not typical of the RF products from National Instruments. Can you please provide some detailed information on how to reproduce this? This includes what type of signals you are generating, what connections you are making, what settings on the demo panel, etc....

Also, if you have any screenshots those would be great.

Regards,
Andy Hinde
National Instruments

[cochenob]

Good to hear this isn't typical.  I didn't think it was!

As mentioned, the set up is simple.  I'm just connecting the upconverter to the downconverter with a small shielded SMA cable.  I'm generating a single tone at 10MHz using the GenerateSingleTone.vi or something similar to that (see IQPhase.jpg attached).  In IQPhase.jpg there is a simple program I made similar to one of the "Getting Started with I and Q" example VI's.  Plotted is I, Q, Magnitude, and Phase (in degrees, calculated from I and Q).  I took 100ms of data...and you can clearly see the modulation.

The second attachment (spectrum.jpg) is a screen shot of the 5660 Demo Panel.  Centered at 10Mhz...a Span of 1kHz...and RBW of a few Hz.  The Delta Peak function shows the side bands 60Hz away.

The chassis is being used on a large optical table, which has been sometimes known to act as a big capacitor, picking up all sorts of EMI.  However, repositioning the chassis (isolating it from the table, picking it up in the air, etc.) did not solve the problem.  It seems like maybe it's getting inside the chassis some other way...though as you mention above, it seems like awfully strange behavior.

I appreciate your insight!

[cochenob]

Just as a follow up...I've taken the chassis into another room and another outlet.  I also realized that the 5660 demo panel has a feature already to look at I and Q vs time.  I used the demo program to re-examine I and Q vs time just to elminate any potential differences between my quick and dirty program, and the LV demo (which should be "correct" in the case that my sources of error were in my own code).  Unfortunately, the fluctuations are still quite noticable.  

Another detail...I have locked the downconverter to the upconverter externally via the 10Mhz synch.  Is it possible that there is some noise getting in there?  The specs seem to suggest that the two modules can be locked quite well.  I'm confident that I'm doing the external locking correctly.  In the 5660 demo panel...I go to Timing...then choose "External" in the reference field.  The device then takes a few seconds to lock (if I'm looking at I vs. Q....I'll see the rotating vector eventually come to "rest".  I say "rest" with quotations, since it still dances around a little bit because of this noise I'm seeing).

Any ideas why these measurements are so noisy?

[Andy_Hinde]

Hi Brandon,
Can you tell me what type of PXI chassis and controller you are using?

Thanks,
Andy

[cochenob]

PXI-1042  8-Slot Chassis

PXI-8105 Controller

[cochenob]

Another little nugget...

I notice that when turned off (but still plugged into the wall socket), the chassis emits a subtle, but certainly audible "hum" or "buzz" around the back panel.  It sounds like maybe there's a short or bad connection somewhere in the power supply paths inside.  This sound is not present in a second PXI chassis I have...and furthermore, goes away when I unplug the chassis from the wall.

I think it might be best to send the thing back for NI to take a look at it.

Could you point me in the right direction in terms of who to contact?

[JerryAL]

Hi Brandon

Can you place the NI 5660 in your second chassis to make sure the issue is the chassis?

Jerry

[Andy_Hinde]

Hi Brandon,
With this additional info, it sounds like it is the chassis. If you can place the PXI-5660 in a different chassis of the same model and the problem is not visible, then you should RMA the problematic chassis.

I would suggest calling 1-800-IEEE-488 (1-800-433-3488) and asking to speak with a RMA coordinator. Typically NI has customers work with our Application Engineers to troubleshoot the problem before an RMA but since you are completing that process now you can explain this to the RMA coordinator to speed up the proces..

Thanks,
Andy

[cochenob]

I have a second 5660 in the other chassis...so I swapped in the upconverter I was using in the "suspect" chassis, and got the same result (lots of noise on I and Q).

This would seem to suggest that the problem might not be the chassis...however...just as a precaution...I'm going to try and power everything from an isolated supply just to see what happens.

Another possibility is that the locking between the up and down converters is the source of this noise (see details in an above message).  The specs however seem to suggest that I should be able to do this accurately...so maybe this isn't the culprit.  Unfortunately...I can't seem to find any VI's in the 5660 library that give any feedback as to the "goodness" of the lock.  Is there anyway I can do this?

I've looked at the 10MHz output reference of the upconverter on a spec analyzer (not the 5660...but an HP device).  There didn't seem to be any noise there.  I've also looked at the output of the upconverter to see how much 60Hz (or other) noise was on the signal.  There was a single peak 60Hz away from whatever tone I decided to output...but it was down about 60dB or so.

What still baffles me is that when I lock the devices and look at I and Q using the 5660 demo panel...I see a significant amount of fluctuation (either from 60Hz noise or a bad lock or both).  However...when I run the pre-canned modulation/demodulation routines...these noises aren't showing up (or at least don't appear to be).  I'm surprised that the software method of carrier acquisition is that far superior that physically locking the two oscillators together manually like I'm doing.

What am I missing here?

---
Brandon