ESP32-S3 Wi-Fi 2.4GHz receiver test with Wi-Fi tester failed with packet drop issue

[chandramauli]

Hello team,

We are trying to get the certification from the Wi-Fi tester which is required for the production of the product.

Setup details used for this test is as below.
- DUT : ESP32-S3-WROOM-1 Wi-Fi Device
- Remote tester: Wi-Fi Tester from Rhode&Schwarz
- DUT Frequency for Wi-Fi : 2.4GHz receiver test with Wi-Fi tester
- DUT mode : AP-Mode
- AP channel : Channel - 6

Test scenarios:
- Test is executed without creating any noise signal in the channel.
- Test is executed with creating frequency based noise at 805MHz, with vertical and horizontal polarization.

Expected results:
- DUT should not drop the packet more then 10% at 10 V/m noise.

Actual result
- DUT drops the packet more then 10% at 7 V/m noise.
- DUT drops the packet upto 90% at 10 V/m noise.

Please help to answer the queries raised as below.
1. Is the module "ESP32-S3-WROOM-1-N16R8" certified with the Wi-Fi tester device? -
- If yes please help to share the test execution report.
2. Is the module "ESP32-S3-WROOM-1-N16R8" validated for the Wi-Fi receiver test using any Wi-Fi tester instrument?
3. Can the Wi-Fi receiver test passed through "ESP32-S3-WROOM-1U-N16R8" which has a dedicated antenna?
- If yes, please suggest/share the antenna details required for this.
- Please suggest the location/direction of the antenna for placement.

Please let me know if any other details required.

Regards,
Chandramauli

[chandramauli]

Hello Team,

Can anyone please help to resolve the above mentioned query to pass the certification for the product? This certification blocks the production as of now.

Regards,
Chandramauli

[eriksl]

I am (also) seeing particularly bad WiFi performance on the ESP32-S3. The device is associated at 802.11N 65 Mbps, but the throughput is about 1/50'th of that. My ESP8266's with similar code perform better.

The documentation warns about interference of the USB PHY towards the Wifi PHY, causing "performance loss". Can you try this with the USB PHY turned off (see menuconfig option)? Maybe that's the problem. For myself it's not very practical to turn off USB, so I guess I'll have to live with it.

[axellin]

The documentation warns about interference of the USB PHY towards the Wifi PHY, causing "performance loss". Can you try this with the USB PHY turned off (see menuconfig option)? Maybe that's the problem. For myself it's not very practical to turn off USB, so I guess I'll have to live with it.

Could you share the link for above document?

[eriksl]

Actually I found it only mentioned in the menuconfig.

If you start menuconfig, go to
- Component config
- PHY

And then just select "Keep the USB PHY enabled when initializing WiFi" and then press "?", you'll see this text:

On some ESP targets, the USB PHY can interfere with WiFi thus lowering WiFi performance. As a result, on
those affected ESP targets, the ESP PHY library's initialization will automatically disable the USB PHY to
get best WiFi performance. This option controls whether or not the ESP PHY library will keep the USB PHY
enabled on initialization.

Note: This option can be disabled to increase WiFi performance. However, disabling this option will also
mean that the USB PHY cannot be used while WiFi is enabled.

Please note this option is active for IDF_TARGET_ESP32C3 and IDF_TARGET_ESP32S3, so the S3 is definitely affected according to this text.

I am not at all pleased about this of course.

[ESP_Sprite]

It's a known issue, but iirc the effect is minor, from memory only -1dB and only on a few channels. I doubt it's responsible for the packet drops seen here.

[eriksl]

Ok, thanks!

How can we take this issue forward and find out what's going on? Is there any low-level information we can share that would help?

[ESP_Sprite]

Sorry, I'm not a RF guy, but I'd look at the stability of your power supply (EMC spikes in there can affect RX performance) and antenna placement. For the latter, have a look at the hardware design guidelines, it has a section on PCB design and module placement. If you're willing to share it (privately) with us, we also offer free design review services.

[eriksl]

TS isn't responding right now, myself I am using a Lolin Wemos S3-mini board, fed by USB-C, so that's not the first place for me to look. Howeve,r who knows and maybe something isn't quite right in the design there. The interesting thing is that the board associates at the full 802.11N speed of 65 Mbps. I am also considering I am doing something wrong in software, something that just unexpectedly takes a lot of time. I can't avoid copying every frames of ~4k two times within PSRAM, is that a very slow process? The 8266 can do it quite quickly in my experience, so I wouldn't think bad of it. I guess I'll need to do some timing there. Otherwise the CPU clock (80-160-240 Mhz) hardly makes any difference, so I wasn't thinking of a CPU-bound process.

Sorry for hijacking this topic!

[chandramauli]

Hello ESP_Sprite

Thanks for your feedback on this. I just wanted to summarize that the issue of WiFi packet drop occurs only when noise is added on specific frequency. In normal scenario packet are received as expected. Can you please help to answer this query. Is it possible for you to add any RF expert into this loop for further debug?


Regards,
Chandramauli