CC1200: Poor 140 MHz FM/AM RX Sensitivity

AM and FM is two rather different modulation formats.
- Do I understand you correctly if you use RSSI to demodulate AM and CFM to demodulate FM?
- Do you see the same sensitivity for both AM and FM modulation?
- For AM, 25 kHz RX bandwidth sounds narrow. Have you checked the signal bandwidth? Also have you tried if you are able to demodulate the signal using the OOK setting in the chip? It should also work for AM, dependent of the modulation depth.
- For FM, do you need to use CFM or would it be possible to use transparent mode?
- For the hardware solution: To test that the hardware is not limiting sensitivity you should use a signal generator (ideally sending packets but preamble also work) using FSK with the same datarate as your FM signal.

Thank you for responding...  answers to your questions:

1.  Yes, I use the RSSI output to detect Amplitude Modulation and the CFM output to detect Frequency Modulation

2.  I am looking at the RSSI and CFM outputs displayed on a computer in an oscilloscope-like graph of amplitude vs. time.   Both AM and FM signals degrade at approximately the same input signal strength of about -80 dBm with an RX Bandwidth of 20.8 kHz  (BB DECFACT= 20 and ADC DECFACT = 48).  

If the CC1200 had a noise figure of 7 (as I have seen mentioned on other posts) then I would expect a sensitivity level of more like -113 dBm.  That's a huge difference!

3.  We are using the chip to demodulate voice.  For AM actually the occupied bandwidth is more like 6 kHz, so 25 kHz is plenty wide.  We can't use the OOK function because it would distort the voice.

4.  Yes, I am using the Transparent Mode.  I am reading the CFM_RX_DATA register.  With strong signals, it works great.

5.  As mentioned, we are using the chip as an analog voice demodulator.   Our firmware is not setup to enable the digital sections of the chip for the proposed test.  Do you think this is important?

Poor performance could be either software and/or hardware.

Here, it sounds like you have made your own hardware and the first step would be to check this. You don't need to change the firmware, what you can do is to use SmartRF Studio to control the board though a SPI interface.

>Both AM and FM signals degrade at approximately the same input signal strength of about -80 dBm with an RX
What do you mean by degrade?

More info and questions:
- I have programmed the MAX and MIN AGC to provide the highest levels of gain so that AGC isn't an issue
- Are there ANY configuration settings that could possibly impact the sensitivity to such a great degree?
- Is it possible that we need to program some mode for the RSSI and CFM to work with weak signal levels?
- Is it possible that the chip is not capable of providing the desired sensitivity level at 140 MHz for some reason?
- Are there any undocumented debugging hooks or commands that could be used to diagnose this problem?
- Is there a way we can get support faster for this issue?

Thanks!!

Degrade means that the signal is corrupted by noise. Is there a way I can post screenshots?

What value would there be in using SmartRF Studio to control the board? I am able to fully control the CC1200, and the demodulation works great as long as the signal is strong. Thanks!

Is there a way to get the SmartRF Studio and the demo boards to stream RSSI and CFM analog signals so I can measure the sensitivity using them?

Art,

Yes, you should be able to program the MSP430F5438A that is on the TRXEB board to extract the CFM and RSSI data in realtime using the same algorithm you have developed for your own hardware.

We have also create a demo for that some time ago on an experimenters board (due the fact that is had a microphone built in)

e2e.ti.com/.../165695

This example uses the CFM register in realtime on a MSP430F5438A.

Regards,
/TA

TER said "- For the hardware solution: To test that the hardware is not limiting sensitivity you should use a signal generator (ideally sending packets but preamble also work) using FSK with the same datarate as your FM signal."

I did this using transparent mode and a GPIO pin for the RX Output data.

Results were that the sensitivity was approximately the same as seen with AM and FM signals.

I am just looking at GPIO3 using an oscilloscope with the SERIAL_RX mode selected.  Data is perfect with strong signals and degrades between -80 and -90 dBm, just like the AM and FM signals.

The schematic is very similar to the reference designs, except that the input matching LC values are scaled to 140 MHz.  I also tried going into the LNA inputs single ended, with approximately the same sensitivity results.  Sorry, I cannot post the schematic.