Can connect to iPhone hotspot only if option “Optimize compatibility” is enabled

[giuliobenetti]

Checklist

• Checked the issue tracker for similar issues to ensure this is not a duplicate
• Read the documentation to confirm the issue is not addressed there and your configuration is set correctly
• Tested with the latest version to ensure the issue hasn't been fixed

How often does this bug occurs?

always

Expected behavior

It should connect without enabling that option possibly.

Actual behavior (suspected bug)

Need to enable Optimize option

Error logs or terminal output

Steps to reproduce the behavior

Setup an iPhone as hotspot and don’t enable Optimize compatibility option

Project release version

Latest

System architecture

ARM 64-bit (Apple M1/M2, Raspberry Pi 4/5)

Operating system

Linux

Operating system version

Buildroot BSP

Shell

sh

Additional context

No response

[mantriyogesh]

Yes, if you're using an iPhone hotspot, please ensure "Maximize Compatibility" is enabled in your iPhone’s hotspot settings. This setting forces the hotspot to:

• Use 2.4 GHz only (ESP32 is typically limited to 2.4 GHz)
• Disable Wi-Fi 6 (802.11ax)
• Use 802.11 b/g/n modes that are compatible with embedded Wi-Fi devices

Without this option, the iPhone may switch to 5 GHz or Wi-Fi 6, which some ESP slave cannot connect to, if limited by their Wi-Fi capabilities.

May we know the slave ESP chip used?

[giuliobenetti]

@mantriyogesh I'm using esp32c6, it should handle WiFi 6 in some way even if in 2,4Ghz and AFAIK iPhone without option enabled should use some fallback mechanism to get to connection. It's known that many IOT devices don't work with it, but it would be great if esp-hosted could handshake with iPhone until it works, if possible of course.

By the way, I don't have esp32c5 here, but could it work with iPhone instead?

[mantriyogesh]

@kapilkedawat Can we please try to test C5 with iphone?

[kapilkedawat]

@Shreyas0-7 please help..

[giuliobenetti]

Yes, if you're using an iPhone hotspot, please ensure "Maximize Compatibility" is enabled in your iPhone’s hotspot settings. This setting forces the hotspot to:

• Use 2.4 GHz only (ESP32 is typically limited to 2.4 GHz)
• Disable Wi-Fi 6 (802.11ax)
• Use 802.11 b/g/n modes that are compatible with embedded Wi-Fi devices

Without this option, the iPhone may switch to 5 GHz or Wi-Fi 6, which some ESP slave cannot connect to, if limited by their Wi-Fi capabilities.

I've scanned with nmcli and yes with "Maximize Compatibility" option enabled:

• Use 2.4 GHz only (ESP32 is typically limited to 2.4 GHz)
• Disable Wi-Fi 6 (802.11ax)
• Use 802.11 b/g/n modes that are compatible with embedded Wi-Fi devices
• enable WPA2 only, more likely supported by IOT devices

while with option disabled:

• Use 5Ghz only(that's crazy to me because AFAIK WiFi 6 should have both 2.4Ghz and 5Ghz enabled) and so AP can't be scanned from a 2.4Ghz Station like ESP32*(Probably except ESP32C5, waiting for @Shreyas0-7 results)
• Enable Wi-Fi 6 (802.11ax) only with 5Ghz. It should expose both 5Ghz AND 2.4Ghz as best practice since it's not seem to be mandatory to have both for WiFi 6
• Use 802.11ax mode that should be compatible with embedded Wi-Fi devices if the devices theirself support it, note ESP32C5 where it's stated "ESP32-C5 features a dual-band Wi-Fi 6 (802.11ax) radio with 802.11a/b/g/n/ac backward compatibility. "
• enable WPA2 AND WPA3 that should be sufficient for IOT devices

So I expect ESP32C5 to work.

[Shreyas0-7]

Hey @giuliobenetti, thanks for your observations,
I have tested esp32c5 ng with iPhone 14 Pro Max, iOS version 18.4. ESP32c5 can connect with the iPhone hotspot with "Maximize Compatibility" option disabled.

[giuliobenetti]

Hey @giuliobenetti, thanks for your observations, I have tested esp32c5 ng with iPhone 14 Pro Max, iOS version 18.4. ESP32c5 can connect with the iPhone hotspot with "Maximize Compatibility" option disabled.

Thanks a lot for the test, this sounds very good!

[giuliobenetti]

@Shreyas0-7 can you please test the throughput at 5Ghz and 2.4Ghz? Thanks in advance

[Shreyas0-7]

Hey @giuliobenetti we are doing some internal testing right now, will update throughput soon

[giuliobenetti]

Hey @giuliobenetti we are doing some internal testing right now, will update throughput soon

Thanks a lot. I’m going to setup an embedded Linux board(RK3562) with have with C6 using C5 with very short lines to keep a decent Signal Integrity and test as well.
Then let you know the results too.

[giuliobenetti]

I finally found time to put my hands back on this. Unfortunately when I try to flash with this command:
esptool.py --chip esp32c5 --force --port /dev/ttyS2 --baud 460800 write_flash \ 0x0 /lib/firmware/bootloader.bin \ 0x8000 /lib/firmware/partition-table.bin \ 0x10000 /lib/firmware/network_adapter.bin

I get this result:

esptool.py v4.10.0
Serial port /dev/ttyS2
Connecting...
Failed to get PID of a device on /dev/ttyS2, using standard reset sequence.

Chip is ESP32-C5 (revision v1.0)
Features: WiFi 6, BT 5, IEEE802.15.4
Crystal is 48MHz
MAC: 30:ed:a0:ff:fe:e4:4f:c4
BASE MAC: 30:ed:a0:e4:4f:c4
MAC_EXT: ff:fe
Uploading stub...
Running stub...
Stub running...
Changing baud rate to 460800
Changed.
Configuring flash size...
Unexpected chip id in image. Expected 23 but value was 0. Is this image for a different chip model?

A fatal error occurred: /lib/firmware/bootloader.bin is not an ESP32-C5 image. Use --force to flash anyway.

Is there a fix or a workaround? I'm using esptool 4.10 and --force is not an option even if pointed in log

[mantriyogesh]

The command generated after idf.py build for me, for esp32c5 after following

rm -rf managed_components/ dependencies.lock sdkconfig build/ sdkconfig.old; idf.py set-target esp32c5
idf.py build

is

python -m esptool --chip esp32c5 -b 2000000 \
  --before default_reset --after hard_reset \
  write_flash --flash_mode dio --flash_size 4MB --flash_freq 80m \
  0x2000  build/bootloader/bootloader.bin \
  0x8000  build/partition_table/partition-table.bin \
  0xd000  build/ota_data_initial.bin \
  0x10000 build/esp_hosted_mcu_wifi.bin

Never use force.

Some options in this command even if not mentioned, would be auto populated by idf.py.

Your effective command after changes could be:

esptool.py --chip esp32c5 --port /dev/ttyS2 -b 460800 \
  write_flash 0x2000 /lib/firmware/bootloader.bin \
               0x8000 /lib/firmware/partition-table.bin \
               0x10000 /lib/firmware/network_adapter.bin

There is major diff that you are pushing the bootloader on 0x0.
Please ignore partitions that you do not have and change the partition binary names as per your application name.

[giuliobenetti]

@mantriyogesh thank you. I've successfully programmed it after applying this patch pending:
#669

Unfortunately I have SI issues because of wires too long and crossed. So I'm going to setup an A20 board with 2.54 pins and use Jumper Wires. Then I come back to you.

[giuliobenetti]

Nothing to do neither with wires less than 5cm and far from each other. All 47K pull-ups are there, 22R series for CLK.
It begins to work only at 5Mhz with 1-wire but when it tries with wpa_supplicant to connect to either 2.4 or 5Ghz network it fails with error 2.

I proceed remastering to be sure to have correct SIignal Integrity.
If you @mantriyogesh have any other ideas to give a fast try please let me know.
To do this try I had to setup an Olinuxino A20 micro using SDC1 port. Unfortunately all pins are grouped together and no GND pin is close. The best would have been to have GND-CLK-GND-D0-GND-D1-... and so on, but nothing to do.

At this point I trust your ESP32C5-MINI-1 in place of C6-MINI-1 module and remaster carefully.