Wi-Fi6 communication speed difference caused by i.MX8MQ specification difference

[hbjeong]

Hi NXP team,

I have a question about the communication test using JODY-W377 on i.MX8MQ and i.MX6Q.

The test conditions are

i.MX8MQ - AP mode
(Run iperf -s -u -i1)
i.MX6Q - Station mode
(Run iperf -c 192.168.1.20 -u -b500M -i1 -t 20)

Running the test as shown above will give the following results depending on the part specification of the i.MX8MQ.

Test 1
MIMX8MQ6DVAJZAA
Commercial 0~95 °C
Cortex-A53 CPU speed level : 1.5GHz
Wi-Fi6 Communication speed: Approx. 510Mbps

Test 2
MIMX8MQ6CVAHZAA
Industrial -40~105°C
Cortex-A53 CPU speed level : 1.3GHz
Wi-Fi6 Communication speed : Approx. 250Mbps

(Attach the log files for Test1 and Test2)

The i.MX6Q board is the same for both Test1 and Test2,
i.MX8MQ is the same board, only the AP (i.MX8MQ) is different.

Test1 : Commercial
Test2 : Industrial

If you remove each commercial and industrial AP (i.MX8MQ) and rework them on opposite boards, the symptoms are reversed.
The symptoms are reversed (the speed difference is reproduced depending on the AP).

(For your information, JODY-W377 and i.MX8MQ operate with PCIE interface).

Regardless of the specific temperature, if all conditions (including temperature) are the same, changing only the AP from MIMX8MQ6DVAJZAA to MIMX8MQ6CVAHZAA reduces the throughput from 500Mbps to 250Mbps.
If all conditions are the same and only the AP is changed, you will see the same behaviour as above.
This has been confirmed by reworking several boards and APs.
I recently received two MIMX8MQ6CVAHZAA from NXP's website and reworked them all, but the symptom is the same.

In conclusion, the commercial version of the AP is capable of 500 Mbps, while the
the industrial version of the AP is not capable of 500Mbps speeds.
Can you explain this?

I tried to test it at -30℃, 0℃, 25℃, 50℃ using a temperature chamber, but the result is the same as above.

The temperature specification of the commercial version of the AP is 0℃ to 95℃.
However, when tested in a temperature chamber, the commercial version of the AP boots and communicates normally at -30℃.
Could you please explain this?

Kind regards.

[Rita_Wang]

Are you using the NXP reference board? Which version BSP are you using?

[hbjeong]

No, the boards we tested were developed by us.
The BSP version is 5.15.32.

[Rita_Wang]

Which Wi-Fi6 are you using?

[hbjeong]

Please find below a revised summary of our inquiry:

[Inquiry 1]
We have observed a difference in Wi-Fi 6 communication speeds between the Commercial and Industrial versions of the i.MX8MQ AP chip.

The primary differences between the two versions are the temperature specifications and the Cortex-A53 CPU clock speed (Commercial: 1.5GHz, Industrial: 1.3GHz).

In our tests, the Commercial version achieved a bandwidth of 500Mbps, while the Industrial version only reached 250Mbps, which is half the speed. We are curious if a 0.2GHz clock speed difference can cause such a significant performance discrepancy.

For reference, Wi-Fi communication between two i.MX6Q (1GHz) boards also resulted in a bandwidth of 500Mbps, so the CPU clock speed alone does not seem to fully explain this phenomenon.

We suspect other factors, such as the Wi-Fi 6 support features of the i.MX8M series or characteristics of the PCIE interface, might be contributing to the difference.

Could you provide a technical explanation from NXP on the root cause of the communication speed difference? Additionally, what recommendations would you make to improve product reliability in this context?

As briefly mentioned above, the Wi-Fi 6 interface is PCIE. Does the PCIE speed vary between the Commercial and Industrial versions?

[Inquiry 2]
The official operating temperature range for the Commercial version of the AP is 0~95°C, but we have found that it operates normally and can communicate at -30°C.

We believe that NXP may have conservatively specified the temperature range to ensure product stability.

Given that the actual operating temperature range is wider, indicating high product reliability, could you provide additional explanation regarding this aspect from NXP's perspective?

Thank you for your attention and we look forward to your response.

Best regards,

[Rita_Wang]

Hi @hbjeong ,

I see that our wifi engineer shaun are following yiour case, he is our wifi engineer, you can continue contact him, I am i.MX engineer.

https://community.nxp.com/t5/i-MX-Processors/JODY-W377-Wi-Fi6-communication-speed-difference-caused-...

Hope this can do help for you

Wish you have a nice day

Best Regards

Rita

[hbjeong]

We are using JODY-W377.