Hi, im new to ARM mbed. Im using LPC 1768. Anyone know is it suitable apply in network-on-chip project? if no which platform is more suitable. What type of network-on-project i can do. Any suggestion.
If you have the mbed board, then the only additional component you need is a RJ45 connector, which is compatible.
Unfortunately, I do not know which ethernet PHY is on the board, so I don't know which RJ45 connector type would work.
If you're inter-connecting two mbed boards without connecting them to a LAN or WAN (that is: connecting the two boards to eachother), then you can skip the RJ45 connectors.
You can use LPC1768 for 10/100 netowrk if you wish.
-But you'll need an ethernet PHY.
For instance, Micrel Inc.makes some great PHYs.
You can choose between two interfaces:
1: MII (abbreviation for "Media Independent Interface")
2: RMII (abbreviation for "Reduced Media Independent Interface")
The difference is that RMII will use less pins, but RMII will give you 100 Mbits/sec, where MII will use more pins and give you 1.25 Mbits/sec.
Here's a couple of examples:
The KSZ8051 can be used for MII.
The KSZ041 can be used for both RMII and MII.
The above two PHYs are compatible with HanRun HR911105A.
-You can find other PHYs from this or other vendors, which will do just as well.
If speed is not an issue, then you can connect virtually any microcontroller to ethernet by using ENC28J60. Ready-made modules exist with these fitted.
If you need WiFi instead, you can use ESP8266 with your microcontroller. Ready-made modules cost around $4.
Thank you for your helpful answer. May I know which version and platform (board) suitable for me to do a project on network-on-chip.
Another board which is very easy to get started for networking projects, is the LPC4088 Quick Start board by Embedded Artists.
https://developer.mbed.org/platforms/EA-LPC4088/
LPC4088 and LPC1768 have good compatibility, as well.
I think that the mbed LPC1768 board is probably the most obvious one to use.
-It's almost "ready-to-go" as it is; you only need to connect the RJ45 socket.
Make sure you read David Smart's answer on this question.
He links to this site and speaks about a base board with RJ45 connector, for instance one of these:
-There might be more, but those were the ones that quickly caught my eye.
David Smart also designed his own board; it seems he offers the blank PCBs as well (in case you want to fit only the necessary components):
SmartBoard Baseboard
Comparison between cortex-M and cortex-A which one is more suitable to do a project? and apply in which board?
This is my view on the question...
If you want to make your own designs, then Cortex-M is the easiest to go for.
A Cortex-A board typically requires you to have external SDRAM and NAND-flash on the same board.
I often see that when people make network applications for existing Cortex-A boards (such as CubieBoard, Raspberry Pi, Odroid, etc.), they tend to install Linux first - this is probably because it's easier to use BSD-sockets than writing your own firmware.
I would expect that the Cortex-M solution would most likely require a smaller PCB-space than the Cortex-A; however, it really depends on the choice of chips.
The Cortex-A processor is no doubt the fastest, thus if you have a lot of complicated jobs to do, which requires quick response, the Cortex-A might be a better choice.
If you want to be able to make your own PCB, I recommend using Cortex-M, because the Cortex-A needs much more precision in your design. This would also be the quickest route to a PCB you design yourself.
If you're just looking for ready-made boards, you will probably find that getting started with a Cortex-A board and Linux is the quickest route to a "Hello World!" string sent over Ethernet.
If you're looking for building a compile-farm or other type of computing-farm, there's really only one option, that's Cortex-A. It would be more expensive to build a clustered Cortex-M4 board which delivers the same performance as a Cortex-A7 running at 1.8 GHz.
If you can provide more details about what the job of the microcontroller needs to do, I think we can get you closer to a solution.