Issues regarding the RF characteristics of your product (CYW43907)

[Eli]

To whom it may concern, I have recently purchased one fo your products - CYW43907 (dual-band Wi-Fi soC), and been testing its performanceon a dedicated kit -CYW943907AEVAL1F (as suggested on Infineon official website). I am writing to address several issues I encountered during testing the product's WLAN 2.4 GHz receiver performance which conflicts with the specifications started within the datasheet you provide.
In your datasheet (page 53) it is stated that the sensitivity (measuredon 8% PER) in 1 Mbps DSSS mode is typically -98.9 dBm (no maximum value provided). To test this, I used a CMW290 Functional Radio Communication Tester as an AP which sent data packets and calculated PER by counting ACKs (acknowledgements) received from the DUT (CYW43907). The DUT was connected to the CMW by RF cable which attenuation was considered during our testing. PER was calculated in various 2.4 GHz Wi-Fi channels as function of the input power. By examining the results, it appears that there are two unexplained phenomenons occuring in every channel. At input power level of -76 dBm there is a PER of 80%, which is 22 dB above the -98 dBm sensitivity stated in the datasheets.
In addition, below -76 dBm there are "spikes" of PER appearing every 1 dB drop in the input power level (I assume this can beas a result of an unstable feedback loop in the AGC).
I would greatly appreciate ifyou can address these issues in detail and suggested methods to fix them.
Best regards,
Eli

[Murali_R]

How are you testing this? Can you describe the RF environment being used here along with the command sequence?

[Eli]

Hi,

Our setup consists of your evaluation board which we connected using RF cables directly to a CMW290. The CMW acts as an AP and creates association with the DUT as defined in 802.11b protocol. Then, data packets (1024 octets)  are sent over the created link by the CMW and acknowledged by the DUT. This way, the CMW calculates PER. Sensitivity testis done by repeating this process over several input power levels and calculating PER for each.

Best regards,

Eli.

[Murali_R · Accepted solution]

If this is on the WICED, then use the test.console app, enable WL in the makefile and enable the large buffer in the console.mk file. Below is the command sequence.

You can set the country and the channel to the needed test.