Single-Chip NIC Solutions?

scdeimos asks: "I started out working life as an electrical engineer, but subsequently moved into software development due to the lack of 'interesting' design jobs in the EE industry...one manufactory/power plant control system is much like another. Nowadays I find myself heading back to electronics as the field becomes more and more interesting with PIC?s and STAMP?s that are more powerful than many desktop systems were just a few years ago. Companies like Future Technology Devices International make useful single-chip USB solutions that allow your hack to appear as a serial port (FT232BM) or a parallel port (FT245BM) connected to your favourite CPU for device intelligence. This lets you build useful test equipment like computer-controlled voltmeters, logic analyzers and CRO?s for not much outlay.Which brings me to my question, which centers around NIC solutions. What are people using out there today for providing single-chip NIC connectivity? What benefits do you feel your chip preference has over the competition? Do any have a sockets (TCP/UDP) implementation built-in, or do you still have to write your own protocol libraries in the support CPU?"

Comments

Single chip is harder...

[morcheeba]

I haven't seen any single-chip solutions yet... that's probably because the magnetics already add some parts, and probably also because the high-speed digital processor and the relatively noisy ethernet connection process technologies don't coexist.

You may want to check out Rabbit Semiconductor's core modules. There's also the xport, that, while small, has got limited I/O.

XPort

[GoRK]

I have been very eager to try some projects with the XPort from Lantronix. It's technically not single chip but really it's about as close as you can get for what it does. It fits in an ethernet hood, so it's small enough to put into most anything that needs ethernet. Very cool little gadget...

~GoRK

Re:XPort

[BiggerIsBetter]

I've got one. It's neat alright, if you want to network something that's already got serial comms in it.

Basically it provides a baby webserver, and lets you pass serial commands over the IP link. It's got an 8051 onboard, but the idea is to store a Java Applet onboard and present that via the webserver. The user can then use the Applet to talk to the device. The XPort then shuttles data to and from the serial over Ethernet

I guess you can do trickier stuff if you want to hack the firmware, but for me, it's just fine serving up a web interface to my previously serial interfaced EFI computer.

Re:XPort

[BiggerIsBetter]

I didn't get their Dev Kit, because the Sample kit was enough for what I needed, however the docs do cover this. The Java approach seems to be the standard easy way of using the device, but the Dev Kit does support custom applications and firmware. The Datasheet shows that "OEM" code on a layer above the device server, and the FAQ says it is written for a Borland Compiler. Looks like writing a truely embedded app running on the XPort should work just fine.

Re:TINI from Dallas Semi

[ummcdou4]

Yeah the TINI is really neat. My Engineering Thesis (all Java and VHDL code included) used it interfaced (via parallel) with an EPLD (20k gates plus memory cells) from Altera.

I used it to simulate and control a home automation system over the net. It is a really interesting piece of software but as a sibling piece has mentioned it has pretty severe loading issues. In one case, making a ring with your thumb and middle finger around a wire tied to one of the address pins and moving your hand was enough to cause the TINI to reboot. (replicatable).

As well, they are exremely sesitive to static electricity. Our Prof fried 2 that he was using to do a lab with and I fried one during R&D for my thesis. On the plus side, Dallas Semiconductor is really helpful and actually shipped me a replacement before I sent them my fried one (you don't see that too often)

At the end of the day I just find it really cool that you can run a web server on a SIMM form factor.

G

Vintage = more power to ya

[poptones]

Single chip computers are fun, but if you're not working on something that HAS to be single chip I don't understand why you'd go there. An old P200 machine can be had pretty much for free, and the ISA bus is incredibly easy to use (not to mention those "legacy devices" like serial ports and parallel ports). Boot it up in freedos with a network stack and you got a basic "programmable controller" that'll do just about anything you want. And even at max warp a PIC ain't gonna outrun a P200.

Example: I wrote an interface in turbo pascal that'll use the parallel port as an 8 channel DA system with no extra parts except resistors and caps and a pair of quad op-amps (which you need anyway unless you like burning up parallel ports) and it'll sample 8 simultaneous channels every 150uS with 8+ bits accuracy and absolute phase integrity. Not ground shaking performance, but considering the platform and the software is free and very easy to work on you just can't beat something like it. And if you need better A/D you can always stick in a ten dollar sound card - or build an even faster one; the EISA bus is a piece-a-cake.

BTW I have a stack of NICs in the other room that are, essentially, one chip. There's another "chip" that's the transformer, but the NIC itself is just an RTL8139. There's also an 8 pin DIP on there but I'm pretty sure that's just an eeprom which could likely be replaced by any controller you might use. I've seen at least one project that used these cards with a microcontroller (the cards are cheaper than you can buy the chip unless you're planning to build a few million devices - that's why it's so easy to have so much fun with old peecees).

Re:I've had great results with RealTek

[stefanjo]

Quoting from the FreeBSD source for the RTL 8139:

* The RealTek 8139 PCI NIC redefines the meaning of 'low end.' This is
* probably the worst PCI ethernet controller ever made, with the
possible
* exception of the FEAST chip made by SMC. The 8139 supports bus-master
* DMA, but it has a terrible interface that nullifies any performance
* gains that bus-master DMA usually offers.
*
* For transmission, the chip offers a series of four TX descriptor
* registers. Each transmit frame must be in a contiguous buffer,
aligned
* on a longword (32-bit) boundary. This means we almost always have to
* do mbuf copies in order to transmit a frame, except in the unlikely
* case where a) the packet fits into a single mbuf, and b) the packet
* is 32-bit aligned within the mbuf's data area. The presence of only
* four descriptor registers means that we can never have more than four
* packets queued for transmission at any one time.
*
* Reception is not much better. The driver has to allocate a single
large
* buffer area (up to 64K in size) into which the chip will DMA received
* frames. Because we don't know where within this region received
packets
* will begin or end, we have no choice but to copy data from the buffer
* area into mbufs in order to pass the packets up to the higher
protocol
* levels.
*
* It's impossible given this rotten design to really achieve decent
* performance at 100Mbps, unless you happen to have a 400Mhz PII or
* some equally overmuscled CPU to drive it.
*
* On the bright side, the 8139 does have a built-in PHY, although
* rather than using an MDIO serial interface like most other NICs, the
* PHY registers are directly accessible through the 8139's register
* space. The 8139 supports autonegotiation, as well as a 64-bit
multicast
* filter.
*

Doesn't sound that good.

Here's One Use

[Ed Almos]

These guys are obviously experts at cramming NIC interfaces into small spaces. How about a web server built into an RJ45 socket ?

http://www.commanderx.com/productinfo.html

Ed Almos