Ethernet Kit Includes Free Stack

Silicon Labs has assembled a kit that lets engineers and product designers explore a way to add Ethernet-communication capabilities to a circuit or product. The hardware comprises two boards, one with an 8051-type microcontroller and a second with a SiLabs CP2200 single-chip Ethernet controller. The kit comes with several demonstrations and code examples that use the royalty-free MicroNet TCP/IP stack from CMX Systems. By now, SiLabs will have revised the kit User's Guide to correct minor errors and add omitted information.

I liked this kit, in spite of a few minor hurdles. The ready-to-go CMX stack and the SiLabs wizard front end make it easy for engineers to jump into an Ethernet project. SiLabs provides the source code for each project - but not the stack - so you can see how they work. Software flow charts would help, though. If you want to go beyond the kit's basic exercises, review application note AN237, "TCP/IP Library Programmer's Guide," which explains the TCP/IP API and how to use it.

The 8051 MCU (C8051F120) oversees the CP2200 Ethernet controller and uses its external-memory bus for read and write operations that involve the CP2200's internal transmit and receive buffers, and internal control registers. Programmers will likely use the SiLabs API, but you can perform operations on the registers and buffers if you wish.

It took little time to set up the hardware and print the Ethernet-DK User's Guide from the accompanying CD-ROM. The User's Guide (UG) instructed me to install the SiLabs integrated-development environment (IDE), examples, and a TCP/IP Wizard. I also elected to install documents and application notes from the CD. You get a choice to install additional tools from Keil or Raisonance. But the UG information doesn't mention installing these tools, so I didn't.

The tutorial information starts with basic Web-server code and the use of the SiLabs TCP/IP Wizard, which I liked very much. The wizard gives you an easy-to-use front end for the CMX TCP/IP stack - you don't get bogged down in Ethernet bits, bytes and packets. Check boxes in the wizard let me choose physical layer, Internet layer, transport layer, application layer and system settings. I also set IP addresses for the development kit, my gateway and subnet mask. Upon completion, I saved the "profile." The wizard then created a complete project that supplied the skeleton of a working program I could compile and run. (To find the IP address on a Windows PC, go to Start-->Run... and type in cmd to open a command-prompt window. At the prompt, type /ipconfig /all and you'll see network-adapter and other network information.)

When I tried to run the sample program I ran into a two problems. The first time I ran the IDE, it asked me if I wanted to "connect" some windows. I clicked "OK" but didn't understand what the message meant and still don't. This step didn't cause a problem, but an explanation might have overcome the anxiety of seeing an unexpected and cryptic message.

Second, when I opened the first project and tried to build it, SiLabs IDE couldn't locate the assembler, compiler or linker, so it displayed an error message. The software people at SiLabs told me I had to install the Keil tools. But the UG did not explain the need to load these tools nor did it explain how to connect the IDE with the Keil tools. I tracked down an application note (AN104) that explains the process and some settings you must check.

Unfortunately, the tool-location settings apply only to the current project unless you set up and name a Tool Definition Preset. So, after you follow the instructions in AN104, and before you click "OK" in the Tool Chain Integration window, click on "Save As" and give the Keil settings your own name. Then when you start a new project, go to Project-->Tool Chain Integration ... and locate your named settings in the "Preset Name:" window to set up a new project with the Keil tools. Keep in mind the distributed Keil tools have a 4-kbyte code limit, so you can't create large projects unless you buy the full - and expensive - version of Keil's tools for 8051 MCUs. Of course you can use other assemblers, compilers and linkers, but it could take a lot of work to integrate them with the SiLabs IDE. A second CD-ROM in the kit supplies other software tools from IAR and Tasking. I did not try these tools.

The first project set up a one-page Web server on the 8051 and let me "ping" the server to determine I had a good connection. Next the project code displayed a simple Web page on my lab PC. This demonstration worked well.

The second project exceeded the code limit of the Keil compiler by 34 bytes, so the build ended with an error message. Perhaps I made an error, but the SiLabs-provided code for this project also exceeded the 4-kbyte limit. I went on to the third exercise in which I compiled code that let me set the IP address for the board set via a serial port on the 8051 board. This exercise worked well, although the UG authors refer to the Web-server-2 project when they mean the Web-server-3 project. Also, none of the exercises include a step to stop the 8051 processor before going on to the next exercise. Be sure to click on the red stop sign or go to Debug-->Stop to stop the 8051. Other minor documentation glitches can cause confusion. In one section, for example, the UG explains the need to replace a 2-ohm SMT resistor with a 0-ohm resistor, but it fails to mention which resistor needs replacing. I simply bridged a piece of wire across resistor R1, based on schematic-diagram information.

Also, the User's Guide notes:
"To force an IP address update hold the SW2 (P3.7) switch on the C8051F120 Target Board down and reset the MCU."
Do I reset the MCU through the debug interface or somehow in the hardware?

The UG should state something like:
"To force an IP address update, press down and hold the SW2 (P3.7) switch on the C8051F120 Target Board. Then press and release the Reset button and release the SW2 button."

The final exercise let me modify the simple Web-page contents in HTML and have the server display it on my PC. The murky instructions didn't offer much help. I suggest kit users who complete the third exercise use it at the basis for their Web-page changes. Otherwise you could wonder what project to start with. This final exercise notes use of an "HTML2C" utility program to convert HTLM Web "pages" into an array of characters and produce two files that the server will use. Unfortunately, the UG doesn't explain where to find the HTML2C program. Find it in the VFILE_DIR folder for each of the projects. CMX provides the HTML2C utility for its TCP/IP stack. Just run the HTML2C.exe program and you can figure out how to use it.

The final exercise includes a demonstration of active Java Web pages and it worked well. Just remember that if you use the Web-server-3 project as your starting point, you must go through the steps to establish an IP address for the board set. A flashing green LED on the 8051 board serves as a reminder. When you want to access the dynamic Web pages, I suggest you open a new browser and retype the IP address for the board set. If your browser doesn't clean out its cache, you'll see the previous Web page and wonder why the exercise doesn't work. The browser "sees" the old IP address and simply pulls the old page out of its cache. Caching depends on your browser settings.

By arrangements between SiLabs and CMX, engineers can use the CMX TCP/IP stack binary code for no fee and no royalty in their products. The stack supports FTP, HTTP, SMTP, DHCP and other application-layer protocols, as well as and transport- and Internet-layer protocols. If you need support, technical assistance, source code or the latest version of the MicroNet TCP/IP stack, contact CMX Systems directly for purchase and licensing information.