mjmahon@aol.com (Michael J. Mahon) wrote in message news:<20040224135522.24987.00000302@mb-m26.aol.com>... > Ironically, we are now in this very situation, with a third-party > Ethernet card for the Apple II line, with limited IIgs software > support and no //e software support. This may be why Apple > decided to drop their card. I have to wonder...the C=64 community has not one, but two working Ethernet adapters (TFE and RR-Net), while Apple II has one that kinda sorta works (but only really on the IIgs, and only with TCP/IP), is a closed design and costs 10x as much as a PC 10/100 card (as well as one that got to the late design stage, but was never completed). Now that it's 2004 and you can get Ethernet parts for cheap, I'd like to try. I don't think I could get the cost per board down to $15, but considering how simple the design I'm pondering is, it could easily be cheaper than the LANceGS. I went over to TFE's website (http://wwww.dunkels.com/adam/tfe), and looked at the schematics. The overall design of the thing is even simpler than the CFFA card; it's just a (third-party) board with the Cirrus CS8900, 20 MHz crystal and Ethernet isolation transformer on it, plus an address decoder (all of 1 74LS139!) on the cartridge PCB. It'd be easy enough to port to the II; just change the wiring so that it decodes to slot DEVSEL space instead of C=64 cart space (very simple). The CS8900 register window is 16 bytes long, so if my understanding of II DEVSEL space is correct, it should fit perfectly. It'd also be easy enough to program (since Contiki comes with working driver code for it), and since the CS8900 has full documentation available from Cirrus, modifying or improving the drivers would not be a problem. The one thing that bugs me is the US$70/qty 1 cost of the required CS8900 module. I'm wondering if it'd be cheaper to move the contents of the module onto the board, then put a 8-pin header on the end for a dongle (similar to what LANceGS does). Maybe if enough people want it, I could also include the header for the Embedded Ethernet module. Also, I'm wondering if we would need a working interval timer to use this board...I'm doubting it, since even on the C=64 it's not interrupt-driven, but it's something to look out for. Going from this, I should have a first draft of the II port in a couple of days, once I get PCB123 figured out (it's a bit funky at the moment). I'll post here again with a location once I have something. Comments, questions, please address them to me here. -lee Rob Steinmetz wrote: > Lee Cremeans wrote: > >> Now that it's 2004 and you can get Ethernet parts for cheap, I'd like to >> try. I don't think I could get the cost per board down to $15, but >> considering how simple the design I'm pondering is, it could easily be >> cheaper than the LANceGS. > > > I believe you're right, it could be cheaper. I looked at doing it a > while ago and never followed through due to the lack of software. The CS8900 itself is ~$7/qty 1 from Arrow, so a run of 10 should be quite cheap once I do the math. The biggest expense will almost certainly be the PCBs; even with PCB123's pricing, they're still kinda steep. >> The one thing that bugs me is the US$70/qty 1 cost of the required >> CS8900 module. > > > I wonder how quickly it would get cheaper say for 10 quantity or 50 or > more? Alternatively there are a lot of other embedded ethernet devices > out there, some are cheaper than $70 ea. For this particular module, I'm not sure of the pricing structure, though they do say they offer volume discounts; I should write the supplier and see. They're in Norway, but the shipping costs are quite reasonable (410 for the equivalent of Priority Mail). Using the module would make the board even simpler (basically the same schematic as the original TFE, just with the wiring changed to support Apple II addressing). That said, I do also wonder what the cost differential in using the board vs. integrating the CS8900 directly onto my design would be. (I'm kinda new at this; I've read plenty of schematics and assembly diagrams for work, but I've not yet designed something all by myself.) They give a schematic on their site so you can assemble your own Embedded Ethernet board; I may use that as a starting point to integrating the 8900 directly onto my board. >> Going from this, I should have a first draft of the II port in a >> couple of days, once I get PCB123 figured out (it's a bit funky at the >> moment). I'll post here again with a location once I have something. > > > That sounds great. Someone will have to develop (or port) some software > to use it. That was always the problem. A board is no good without > software and the software is no good without the hardware. With a Contiki driver already written and working on the C=64 (I tried it here in VICE and it worked), and with source available for that, it wouldn't be hard to write a Marinetti driver for it. I wonder what tools you need to develop for Marinetti...something tells me cc65 won't cut it, but we'll see. -lee Rob Steinmetz wrote: > Marinetti is a GS only package, It would be way cooler to have a board > for the Apple ][. Yeah...I would defnitely want to support both 8 and 16-bit systems. The big sticking point with the 8-bit systems, though, is that there's no defined interface for using networking resources (that I know of). It all depends on what you want the card to do; as it stands, the design would have the same limitation as the LANceGS, namely that Ethernet support would have to be implemented on a program-by-program basis. I mainly see the NIC as being a way to access TCP/IP-based resources, such as Web servers, FTP and Telnet (all things Contiki can already do), though any suggestions on other capabilities it could provide (like, say, implementing EtherTalk for machines that can't do AToIP) are welcome. I'm also wondering how the old Apple II Workstation Card handled things...I seem to remember it having some intelligence of its own. -lee