Re: TI-H: 4-Way Doodad
[Prev][Next][Index][Thread]
Re: TI-H: 4-Way Doodad
Don't decide that mine is not good...it isn't that expensive, at 5$ I
barely pay for shipping...
Btw, that 4 port hub is a pos... I'm almost done with a new version taht
incorporates some better IO parts to allow upwards of 128 nodes operating
with the same protocall. There is no source to port since the network
routines are built into the hub, and the communication routines are built
into the calc.
I've almost got drivers made for the C64, C128, VIC (maybe), Apple II
series, IBM (8086 and up), and Mac. This means that HPs, TIs, IBMs,
Apples, Oranges :) can all share modems, printers, and a centralized server
can be made anywhere.
I might need some help with the commodore, but I've got an Apple II
bootstraping from the network.
Its so far 64 nodes per chip, with a max of 4 chips together. Then you can
zone the segments together. The hub operates with TI protocall when
talking to calcs, and it uses its own protocall with computers. It
basically has its own handshaking form that allows anywhere from 1bps to
400,000bps based on the speed of the computer/device.
I personally have it connected to all of my computers which is connected to
a parallel printer.
So, I've ditched the $10 4 port hub and came out with the 64 port hub. I
don't know how much it would be since the electronics distributor doesn't
have any of the chips in stock, but the 64 port firmware should cost about
$5, with all the supporting parts another $5. You can have a
multi-computer network with 64 ports for only $10! IMO, better than the
calcnet thing.
The centralized server consists of a 68HC11 (upgradable firmware), a LCD,
and 1MB of flash, and 512k of static RAM. The Apple or Commodore sends a 1
byte message to the server, and the server dumps the startup instructions
to the computer, so the comouter doesn't need disks! After the computer
starts up, it can send more detailed instructions to the 68HC11 which is
connected to a HD.
Every peripheral's, except the hub its self, ROM can be downloaded off the
net so you can build it yourself.
If there is any interest in this network, I will continue it. I've already
made a chat program that makes an 83 talk to an 8086 (about the best
computer I have working right now). The calcs driver is arround 500 bytes,
and the PCs is about 1.5k. Both are very crude.
If I continue this, by june I can releasee the ROM freely for a
Ethernet<->(I havn't named it yet) converter. The ethernet routines will
convert the network stuff and all you have to do is start a telenet
connection with it, then use the hubs protocall. I think the 68HC11 has
enough rescources to handle 10 connections at a time.
Oh well, if someone likes this, I will finish it... ;)
PS, with a test 2 port version, I was able to get .8MB burst over it before
the paper clips fell out of the computer. That was on 2 PIIMMX we had at
school. Boy did the CL teacher freak when she saw 8 paper clips and a
breadboard hanging out! (It was at school)
I don't expect 800K per sec, but maybe 400k with 10 computers, and 200k
with a bunch of calcs and a few computers. Thats pretty good since
10Base-T gets ~230k per sec burst...
Grant
>> I am interested in building a piece of hardware that allows 4 calculators to
>> plug in and:
>> a. Chat
>> b. Play Specially Designed Games (Compatible with the 4-Way Linkport)
>> c. Host a "server" in which other calcs could download files.
>>
>> I think that the Gameboy 4-way connector would be a good reference/possible
>> conversion. I will need someone to write drivers, preferredly in ASM for
>> Usgard, and possibly someone who is a little more knowledgible about this
>> stuff than I am. I'll try to have a schematic of the gameboy 4-way
>>connector,
>> and a possible schematic of my 4-way connector soon.
>
>MBus already allows this, and MChat shows that it really works. MBus uses
>I2C and a higher-level protocol to interface between calcs, and works with
>straight wire and up to about a hundred calcs (I've never tried it out with
>that many, though...). I admit that CalcNet is slightly faster and, at present,
>more compatible with other calcs than the TI-85, but it is also a lot more
>expensive and at least a bit harder to make. I see no need for a third,
>competing design, and instead I'd like to see more development on the MBus
>and why not CalcNet too. Some people have shown interest in porting the
>MBus routines to other calcs than the 85, but I haven't heard of any
>complete ports yet. A MBus multiplayer game by me is on the way. The
>current MBus drivers were made for ZShell, but I'll probably make an Usgard
>version soon because the game should be easier to make with Usgard. The
>differences won't be big, though.
>
>Your idea seems to be to connect calcs together in the same way as Gameboys
>are connected together. Although I don't know how Gameboys link together, I
>can assure you that you have to either do some pretty advanced electronics
>if you want to use the TI internal protocol (use a microcontroller as CalcNet
>does) or use a custom protocol with simple hardware (use I2C as MBus does).
>Both approaches have their good and bad sides, and I think both CalcNet and
>MBus are about as good as they can be, so we don't need a third system unless
>it relies on some really new and great idea.
>
>This blurb probably seems confusing, check out my MBus page at
>http://www.clinet.fi/~ozone/mbus/
>and Grant's CalcNet page for more info.
>
>-Ozone
>
>--
>*** Osma Suominen *** ozone@clinet.fi *** http://www.clinet.fi/~ozone ***
Follow-Ups: