[Oberon] [EXT] Re: WiFi module survey

[Skulski, Wojciech]

Are these messages going to the list? I attached the PDF to my message, a 15 kB file. Are the attachments going through?
________________________________________
From: Oberon [oberon-bounces at lists.inf.ethz.ch] on behalf of Skulski, Wojciech [skulski at pas.rochester.edu]
Sent: Saturday, September 26, 2020 12:07 PM
To: Joerg
Cc: josef.templ at gmail.com; Paul Reed; PeterMatthias at web.de; waltergallegos at vera.com.uy; Oberon List; Andreas Pirklbauer
Subject: Re: [Oberon] [EXT]  Re: WiFi module survey

Joerg:

  thank you for the e-mail. Meanwhile I wrote an response to the previous one. I made an extensive commentary in the form of yet another schematic page to be released with the board package which is now ready. The page is attached in PDF. I apologize for not using the proper nomenclature with OSI levels.

Yesterday I added the WRL-13678 pin connector to the board. This module is probably the dumbest WiFi out there. The only reason I chose this one was that it uses the exact same connector as the Nordic radio, though pins are wired differently. It is also a similar size. It costs $7 at DigiKey. So it is really easy to plug it it in. Furthermore, I overlaid the connectors such that you have to choose which module you want to use. You cannot use both radios at the same time. The reason is shortage of FPGA pins. (I guess I should upgrade the FPGA, but I have no time for doing it now.)

The most important thought from my analysis is that the modules which offer the ASCII script interface are a one way street that looks inviting, but there is a steep price to be paid later. The ASCII commands are an invitation to a sloppy, quick path to initial success. If you follow the Instructables [ref. 7 in my writeup page on WRL-13678], then you can put your board on WiFi in half an hour or so. And then what? Such an approach is a quintessence of a "hobby quality". So it is not a solid path to go. On the other hand, there is a great appeal in a quick success, if one does not overvalue such fast achievements. It can be a good way to introduce wireless WiFi networking to students, which can be a next step after Nordic radio.

So with some hesitation I added the WRL-13678 socket to the board.

Concerning the access points, there is at least one module which can work both ways. It is Lantronix xPico XPCW100x, which can be an access point for four other WiFi devices. This would allow building a cloud of four WiFi boards served with a fifth server board, which is perhaps a worthy project for students.

Thank you,
Wojtek
________________________________________
From: Joerg [joerg.straube at iaeth.ch]
Sent: Saturday, September 26, 2020 4:10 AM
To: Skulski, Wojciech
Cc: Andreas Pirklbauer; Michael Schierl; Oberon List; chris at cfbsoftware.com; Hellwig Geisse; josef.templ at gmail.com; magnus at saanlima.com; Paul Reed; PeterMatthias at web.de; Tomas Kral; walter at waltergallegos.com; waltergallegos at vera.com.uy
Subject: [EXT]  Re: WiFi module survey

Wojtek

I read your summary now. Wired Ethernet and WiFi define OSI layer 1 (PHY) and 2 (MAC). those two layers differ indeed: PHY obviously and MAC as well (Ethernet= collision detection, WiFi= collision avoidance)

If I’m right, non of your mentioned chips exposes PHY and you have to implement MAC yourself. So, on OSI Layer 3 (IP) the difference is neglectable.
From „socket“ (OSI layer 3 and 4) point of view there is hardly any difference.(you have to set SSID and channel and off you go)

Ethernet is symmetrical (the same chip can be on both sides of the cable), but WiFi is assymmetrical (STA client and AP host). One choice you have to make: can the chip work in both modes? I mean, shall the Oberon system be a client-only or shall it be possible to be an access point as well.(assuming the SW for the AP side existed)

I did not check the data sheets on the WiFi modes the chips support.

BTW: The same choice has to be made for the USB chip; USB is asymmetrical as well. The USB on my board can be client-only. I can connect the board to my PC via USB but I will never be able to connect a USB mouse to my board. Even if the USB SW were ported to Oberon-07.

br
Jörg

> Am 25.09.2020 um 01:33 schrieb Skulski, Wojciech <skulski at pas.rochester.edu>:
>
> Dear All:
>
> I am sending this e-mail to a long list of names off-list because the list server is not happy with attachments. Please feel free to post it to the list if you can push the attachment through.
>
> I was not able to do much real work because it is now grant reporting season. But I did some research on WiFi. The results are presented in several attached pages, starting on page 4. The other pages present my thought and investigations on the Nordic radio module and Oberon software, and some general remarks. All these pages are part of the RiskZero Rev 1 schematic. Since I am in no hurry, I keep pursuing information and adding notes to the schematic package to be eventually released "really soon now".
>
> This investigation was triggered by Joerg remark that he could add WiFi to the System, using Wiznet WiFi module. So I started digging in. The findings were different from expectations. I found that the Wiznet WiFi modules are rather poorly documented. A better choice could be the Sparkfun module WRL-13678 with Espressif ESP8266 because there is more information and code examples, although of mixed quality. The Espressif tools seem good. Some of the Arduino community contributions seem sketchy. The largest reservation against WRL-13678 is low speed.
>
> The landscape of the WiFi modules is very mixed. There seems to be no common denominator. Low performance application can be done with AT commands. Both WRL-13678 and Wiznet modules support these commands. It is not clear whether these two use an identical set of AT commands, or two slightly different sets. Silicon Labs provides its own script BGScript. Lantronix provides yet another command interface named LANCIS. If one develops with one of those, one will be out of luck with the others.
>
> Perhaps the most reasonable approach would be using the ATWINC15x0 module from Microchip and define a common API with Wiznet W5500 because both chips use a similar architecture. But even this is not clear because WiFi is different from the wired Ethernet after all. It may not be warranted to use the same driver if the details do not closely match.
>
> All of these thought are expressed in the attached pages. I also collected all the data sheets and app notes which I am referring to. The zipped archive is 50 MB. I can send it to anyone who tells me how to share such a large file. Having all these documents handy can spare quite a bit of investigative work. If you want to receive these files, please send me a dropbox link or something of this sort.
>
> Thank you,
> Wojtek
>
>
> <WiFi_Modules_Overview-merged.pdf>