wifi.radio.tx_power does not support float

[askpatrickw]

The Wemos docs say to use 8.5db to work around the WIFI connect issue.

Workaround

If you use 8 or 9 you can still connect.

Code/REPL

>>> import wifi
>>> wifi.radio.tx_power = 8.5
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
TypeError: can't convert float to int
>>> wifi.radio.tx_power = 8
>>> wifi.radio.connect(SSID, PASSWD)
>>> wifi.radio.ipv4_address
192.168.1.53

Behavior

TypeError: can't convert float to int

Description

No response

Additional information

No response

[askpatrickw]

Related to PR #6542

[askpatrickw]

@todbot any chance you want to look at this one?

[todbot]

This was by design. The underlying API is int8_t so I figured this attribute should be that way too. Is that a bad assumption?

[todbot]

Like, what do they mean exactly "set to 8.5dBm" when this is the API : https://docs.espressif.com/projects/esp-idf/en/latest/esp32c3/api-reference/network/esp_wifi.html#_CPPv425esp_wifi_set_max_tx_power6int8_t

Oh but wait there's that div by 4 in there now.

Okay I'll see what I can do.

[dhalbert]

The units are 0.25dBm, but the actual power setting is coarser than that, not even in 1dBm units:
https://docs.espressif.com/projects/esp-idf/en/v4.4.1/esp32s2/api-reference/network/esp_wifi.html?highlight=esp_wifi_set_max_tx_power#_CPPv425esp_wifi_set_max_tx_power6int8_t

esp_err_tesp_wifi_set_max_tx_power(int8_t power)
Set maximum transmitting power after WiFi start.

Attention

1. Maximum power before wifi startup is limited by PHY init data bin.

Attention
2. The value set by this API will be mapped to the max_tx_power of the structure wifi_country_t variable.

Attention
3. Mapping Table {Power, max_tx_power} = {{8, 2}, {20, 5}, {28, 7}, {34, 8}, {44, 11}, {52, 13}, {56, 14}, {60, 15}, {66, 16}, {72, 18}, {80, 20}}.

Attention
4. Param power unit is 0.25dBm, range is [8, 84] corresponding to 2dBm - 20dBm.

Attention
5. Relationship between set value and actual value. As follows: {set value range, actual value} = {{[8, 19],8}, {[20, 27],20}, {[28, 33],28}, {[34, 43],34}, {[44, 51],44}, {[52, 55],52}, {[56, 59],56}, {[60, 65],60}, {[66, 71],66}, {[72, 79],72}, {[80, 84],80}}.

Return
ESP_OK: succeed

ESP_ERR_WIFI_NOT_INIT: WiFi is not initialized by esp_wifi_init

ESP_ERR_WIFI_NOT_START: WiFi is not started by esp_wifi_start

ESP_ERR_WIFI_ARG: invalid argument, e.g. parameter is out of range

Parameters
power: Maximum WiFi transmitting power.

[todbot]

So what I copied from MicroPython is incorrect and that tx_power argument is a nearest-neighbor table lookup?

[dhalbert]

I don't know what's going on internally - that code is in a binary blob. It could be something about how the radio hardware works. But in any case there's significant quantization. I too am puzzled about the Wemos instructions.

Have either of you tried wifi at various power settings to see if some don't work at all?

[todbot]

They worked at all the 6 or so settings I tested, showing corresponding changes in TX power using a WiFi sniffer.
This is not a big change, it's just frustrating because Patrick is right this should be a float. :)

[dhalbert]

I agree, a float would be good. No issue changing it.

[todbot]

okay, PR incoming shortly