Hi,

I am now doing experiments to measure the radio carrier frequency. I now set up wireless nodes to transmit packets and I have a spectrum analyzer to capture their frequency.
1. The transmitted packets are carried by 433MHz and modulated as 2FSK. Thus there should be two frequency to represent binary 1 and 0.
2. I collect the frequency data from the spectrum analyzer every 1 second, and the sweep time on the spectrum analyzer is 89 ms.
The data I got for 10 minutes looks like these:


Why there are always three lines when these secondary lines should be outliers? I changed the resolution settings of my spectrum analyzer and I believe they should not be caused by the instruments.

I am pretty new to these kind of experiments, could anybody give me some hints and instructions?

Thanks in advance.

 

Are you using an attenuated probe on your spectrum analyser? Excessive amplitude on the input can cause spurious off freq. readings

 

Are you using an attenuated probe on your spectrum analyser? Excessive amplitude on the input can cause spurious off freq. readings

Thanks Kermit,

No, I am not using attenuated probe, but I have a 10dB internal attenuation in spectrum analyzer. Then the amplitude of the peak is about -30 dB.

Could that problem be caused by coding? I am using interleaved coding and maybe there will be somethings transmitted like headers and checksum?

 

You are still theoretically only transmitting two freqs. Head r data will be 1's and 0's just like any other data. If it is not an issue with input overdrive, then the spurious readings may be due to transients which are being captured during a sampling cycle.
The period of a single sinewave at 433 MHZ is 2 nano-seconds. Did you say your sample period is 89 milliseconds?

 

You are still theoretically only transmitting two freqs. Head r data will be 1's and 0's just like any other data. If it is not an issue with input overdrive, then the spurious readings may be due to transients which are being captured during a sampling cycle.
The period of a single sinewave at 433 MHZ is 2 nano-seconds. Did you say your sample period is 89 milliseconds?

The sweep time on the spectrum analyzer is 89 milliseconds and I collect the frequency data from the spectrum analyzer to my computer every 1 second. Is that fast enough to capture the correct frequency?

As you can see from the figures, the spurious readings look like a straight line depending on some laws so it might not be a random error from the reading? How do you think about it?

 

It is hard to get good readings at high frequency even with the best equipment. You are proceeding correctly, in that you are getting a very large number of samples. From this data you should exclude all readings that are not part of the largest, best defined set. The majority of your scope plottings show a proper reading. Taking this average as a correct result is the right way to handle this situation.

 

I'm not exactly sure what you are trying to do. If you want to see the two carriers bounce, set time base on scope to the baud rate.

 

I'm not exactly sure what you are trying to do. If you want to see the two carriers bounce, set time base on scope to the baud rate.

Thanks BR,

What I am trying to do is to collect carrier frequency data across time and to see whether the carrier frequency changes through time. But now from the data I got I see there are outliers around the actual carrier. I am not sure the reason why they are existing, whether from the crystal circuit generating the carrier frequency or the communication or my collecting method; and how can I deal with them, whether I could eliminate them or just ignoring them.

 

Hi,

I am now doing experiments to measure the radio carrier frequency. I now set up wireless nodes to transmit packets and I have a spectrum analyzer to capture their frequency.
1. The transmitted packets are carried by 433MHz and modulated as 2FSK. Thus there should be two frequency to represent binary 1 and 0.
2. I collect the frequency data from the spectrum analyzer every 1 second, and the sweep time on the spectrum analyzer is 89 ms.
The data I got for 10 minutes looks like these:

View attachment 97531 View attachment 97532
Why there are always three lines when these secondary lines should be outliers? I changed the resolution settings of my spectrum analyzer and I believe they should not be caused by the instruments.

I am pretty new to these kind of experiments, could anybody give me some hints and instructions?

Thanks in advance.

Those are sidebands caused by the switching process.

 

I'm not sure what you mean by outliers. The shifting carrier is the result of a mixing process. Some mixers are clean but most are dirty. The spectrum analyzer is great for spotting this.

How many rf nodes are active? And I don't understand your graph labels? Surely you don't mean high tone and low tone, do you?

Are you trying to see the carriers shift on the spectrum analyzer?

I still don't know what you are trying to do. If you want to SEE the carriers shift................turn your baud rate down to 4 or 5 baud. USE only one rf node. The carrier shift is only a few thousand hertz at best.

 

Thanks BR,

I am now measuring one single rf node, and using a spectrum analyzer to capture its carrier frequency when transmitting. The frequency is got from the peak amplitude position of the signal. I am using FSK modulation and the baud rate is 27.8kbps.

My graphs show the frequency data I captured from the spectrum analyzer during 10 minutes, the upper graph shows the high tone above the central carrier and the lower one shows the low tone. The shift between high and low tone is about 100 kHz.

"Outliers" means the points out of the mainstream of the frequency value. I am expecting two clean line of frequency showing the high tone and low tone without , while I cannot get that ideal graphs. Are these "outliers" existing on the "sideband"? Or it may be caused by the fact that my spectrum analyzer is not as fast as the baud rate and as a result I was capturing the value during switching?

 

Those are sidebands caused by the switching process.

Thanks KL7AJ,

Here is the plot showing the frequency domain of the carrier frequency. So the sidebands are the width of band near the two peak value? And I was capturing them because my spectrum analyzer is not fast enough?

 

Turn your baud rate down to 9600, and see if there's a difference.