ESP8266 Developer Zone

Re: AT COMMANDS parser code sources

2016-10-02T20:29:42+08:00

Great, you are very fair ( and this is Sunday....)
I don't wanna load you too much, let's say it might be a good starting point for me to know the average latency introduced by the AT parser.
But if you deploy some tests i will appreciate it!
Let me know if you use the UART from a computer or from a dedicated microcontroller in a secondary board.

Statistics: Posted by Debbio — Sun Oct 02, 2016 8:29 pm

Re: AT COMMANDS parser code sources

2016-10-02T19:48:05+08:00

I will do some tests for you and get back to you with the maximum data rate I could achieve with the CIPSEND command.
You can achieve transmission of about 1.2Kbyte blocks with wait states of less than 5 ms with custom AT commands though. I can confirm that would be possible, just in case CIPSEND does not work fast enough.

I'll let you know after a few tests.

Statistics: Posted by Guest — Sun Oct 02, 2016 7:48 pm

Re: AT COMMANDS parser code sources

2016-10-02T18:18:52+08:00

Thank you Pratik for the explanation,
the reasons for not releasing the sources are pretty clear to me.

Probably i need to be more clear , we already have a board and we are replacing the wifi module.
The module connection is over UART and we prefere to keep the board as is.
I need to know how to minimize the time for a single CIPSEND with a 200 byte payload with AT and UART.
Or at least i need to know about the minimum time achievable when i do a single send in this manner.
We don't need a huge throughput but we are looking for a fast single transmission ( and reception ),
If we don't get our requirements we will start rethinking our board, for example with SPI or even programming the espressif module.

Thank you for your time,
Have a nice day.

Statistics: Posted by Debbio — Sun Oct 02, 2016 6:18 pm

Re: AT COMMANDS parser code sources

2016-10-02T12:14:00+08:00

Hello,

Firstly, the source code for libat is confidential and is maintained internally by the engineers at Espressif. The reason being that we need to directly interface to the internal RF hardware to make the AT library efficient. And if we released the code publicly, people would have access to info on how to control the Wi-Fi radio, etc. on the ESP8266. And that would not be nice, because RF gone crazy with a buggy program on a chip can interfere with all other equipment in the vicinity. Hope you understand!

And if you want to know how to maximize throughput, you need to know that the limiting factor is UART speed. Not the Wi-Fi speed. The ESP8266 can easily do over 5 mbps in very noisy environments as well. Compare that to the 0.1 mbps of 115200 baud UART!
We recommend using SPI or SDIO based AT interface for better performance.

Statistics: Posted by Guest — Sun Oct 02, 2016 12:14 pm

AT COMMANDS parser code sources

2016-10-01T18:19:07+08:00

Greetings to you all,
i am new to this platform and i am testing AT Command parser capabilities through UART.
I wonder where are located the code sources for the libat.a library included with NON OS SDK V2.0.0.
I found an old source code in a GitHub account related to Espressif but it si really not up to date to the parser present in the last sdk.
I need to go deeper into the code to understand where exactly the 20ms delay acts, and how many times occours during an AT+CIPSEND.
My aim is to find the best throughput achievable in sending data with sequential CIPSEND.
Other aspect is the presence of NAGLE algo on TCP SEND / ACK , what is the minimum payload that permits to get an ACK immediatly for a TCP SEND?
Thank you.

Statistics: Posted by Debbio — Sat Oct 01, 2016 6:19 pm