CC2520: Problems with continuous carrier FCC test transmit even after trying solution from related thread

Glen,

Unfortunately I don't have a CC2520 board with me to perform the tests alongside with you, but I have some considerations.

Glen Meloy said:

If I have it send a packet of all 0xA5 data normally, I get a nice bell curve with a bunch of stuff under it as follows, which make sense:

That is indeed expected and your measurement procedure makes sense.

Glen Meloy said:

I assume what this is telling me (because the bell curve is just as wide as it was when sending real data) is that it's still sending modulated data, it's just all zeroes now.

The equipment might be tricking you. My suspicion is that the RBW of the TinySA is too large (your screenshot shows it as Auto) and therefore the resolution of the CW spike is widened.

A narrower RBW will take much longer to complete the scan, so you could also consider reducing the span from 20 to 10 or even 5 MHz.

There are various references online about this subject - one that i really like is the short video below from Alan Wolke (W2AEW) from Tektronix.

https://www.youtube.com/watch?v=Ffhs9Ny03lM

Another aspect is if the detector is set to peak or averaging. The averaging might decrease any measurement of the peak value. This is observed by the lower value when measuring the CW signal alone (less adjacent power) when compared to the modulated signal.

I will think about additional details that might be contributing to this behaviour and report back in case I find anything else.

Hope this helps,

Rafael

Very interesting lesson there - I definitely learned something, but it does not appear that that is my problem.

In the waveforms from the reference thread his test tone from the CC2530 was on a span of 5.78MHz, with 10 divisions shown on the screen, and pretty much the whole waveform fits within one of the divisions, so I'm expecting the tone to be something like 500-600kHz wide.

In his incorrect still-modulated CC2520 waveform it's set to a span of 12.8MHz. So that's about 1.28MHz per division, and the waveform takes up about 5 divisions. So that's about 6.4MHz wide.

I ran new captures with my RBW set to 10kHz and my "unmodulated" one is showing significant energy for about 10MHz. Our nicer Rhode & Schwarz analyzer at the office is showing similar results.

Some more information... I have another 802.15.4 product on my desk that's about to go into FCC test as well. This one uses a Fanstel module containing a Nordic Semiconductor radio. I thought I'd scan the output from that just for reference and to make sure I'm not fooling myself with this cheapie analyzer. Here are the modulated and unmodulated images I captured from that product using the same spectrum analyzer settings. You can see that the unmodulated test tone is very narrow:

Glen,

Thanks for the additional details, which isolate the test equipment from the problem. Indeed the device seems to be transmitting a modulated signal or something quite unexpected.

As you mentioned, you already copied the settings from the SmartRF Studio to your board and I imagine you used the Code Export feature of SmartRF Studio, is that so? The reason is that, when configuring the Tx transmit unmodulated carrier, I see a few additional registers being set other than the aforementioned MDMTEST0 and FRMCTRL. Some are setting the power but there might be other factors at play.

One additional detail that might be able to isolate any possible problems with the running code: are you able to connect SmartRF Studio directly to your board to set and inspect registers? The reason for my question is to see if the settings are being actually set in the device (I have seen some devices to not properly set the values due to some technicality).  

Historically, SmartRF Studio has been the tool used during certification/compliance tests on CC2520 based boards, thus I can't locate any example code that sets the transmitter to operate in CW. Unfortunately the source code for SmartRF studio is not publicly available.

I will try to gather information from colleagues that are more experienced with the CC2520 and report back in case I find any relevant information.

Best regards,

Rafael

Thanks - so far I've still been unsuccessful. It looks like we're going to proceed without this mode. If it becomes critical my only other idea is to buy an eval kit and see if it actually works, and if so I'll capture all of the comms with a logic analyzer and see what it's doing different from what I'm doing.

Excellent - thanks very much. What is the exact kit you got? It looks like the CC2520DK is out of date and the current thing to get would be a SMARTRFTRXEBK  plus a CC2520EMK, right?

Glen,

I got this with a combination of a CC2520EMK (still available at Mouser) with an old SmartRFEB05 board. However, the current option of using a SMARTRFTRXEBK as a baseboard indeed seems to be the only option available for purchase at the moment.

Best regards,

Rafael