XT-IDE on PCjr

[jmetal88]

Ah, thanks for that info!

I just did a Google search and found an application note on operation of the chip in 8-bit mode. Interesting stuff.

[jmetal88]

Oh, and Mike, let me know which changes specifically you want made to the XT-IDE circuit, when you have time, and I'll add them in the KiCAD schematic.

[jmetal88]

Alright, I went ahead and added the CS8900A ethernet controller to the schematic, wired up exactly as it appears in the 8-bit mode application note (also looking at the tech ref for some of the pins that weren't mentioned in the application note). Ran it through the autorouter with 8 mil trace width and 8 mil clearance, and all the connections can be made, so it looks like we're good with the ethernet thing.

It looks like the way they have it wired up in 8-bit mode by default has it using addresses 0x300 to 0x30F. Mike, does that conflict with anything that you know of on the PCjr? If it does, we'll have to add some extra logic to change the base address.

[MattCarp]

...awesome. of course, awesome.

Can you post a quick pic?

Also, how were you able to do it so fast? I don't have experience with high end autorouters. Is freerouter that good?

On the port address usage, I'd like Mike to chime in, but after a quick check of my "Compute!'s Mapping the IBM PC and PCjr", the only thing that shows up in that address range is the IBM Prototype Card, which I'd guess was a very unusual ISA-bus experimentation board. I've never seen it, so I would consider that a non-factor. My book was published in 1985, so I would say that it is possible that sound cards, networking adapters, and perhaps non-standard disk adapters (SCSI?) may show up in that area, but also seem to recall that 0x300h was a pretty standard address for network adapters.

The book does say that I/O ports 200-3FFh on the PCjr are reserved for the I/O channel, so that's a good sign.

I guess the only other thing I could think of is the PCjr Speech Adapter. I don't know what addresses that sits at.

-Matt

[jmetal88]

...awesome. of course, awesome.

Can you post a quick pic?

Also, how were you able to do it so fast? I don't have experience with high end autorouters. Is freerouter that good?

On the port address usage, I'd like Mike to chime in, but after a quick check of my "Compute!'s Mapping the IBM PC and PCjr", the only thing that shows up in that address range is the IBM Prototype Card, which I'd guess was a very unusual ISA-bus experimentation board. I've never seen it, so I would consider that a non-factor. My book was published in 1985, so I would say that it is possible that sound cards, networking adapters, and perhaps non-standard disk adapters (SCSI?) may show up in that area, but also seem to recall that 0x300h was a pretty standard address for network adapters.

The book does say that I/O ports 200-3FFh on the PCjr are reserved for the I/O channel, so that's a good sign.

I guess the only other thing I could think of is the PCjr Speech Adapter. I don't know what addresses that sits at.

-Matt

As long as you have a decent component layout, and don't forbid vias, freerouter is pretty fast with the initial routing. Unfortunately, it takes several hours to optimize the board once it makes all the connections.

I'll interrupt the post-route optimization after I eat dinner so I can get a rendering of the board posted. Since I'm not up to a final design yet, the optimization isn't really necessary anyway.

[Brutman]

Argh .. you guys are too fast for me to keep up!

The CS8900 was one of the candidate chipsets to use. But we need some flexibility on the I/O ports that it uses.

I was hoping to get some time to workout the changes on the XT-IDE address decoding myself; I need to get started with the tools. The port changes I was thinking of are pretty easy to do; adding another 1KB of memory is a bit more involved.


Mike

[jmetal88]

Alright, here's a quick render of the board with the network chip installed. Sorry it's so small, I couldn't figure out how to get it more detailed without leaving it in a somewhat proprietary format.

[jmetal88]

Argh .. you guys are too fast for me to keep up!

The CS8900 was one of the candidate chipsets to use. But we need some flexibility on the I/O ports that it uses.

I was hoping to get some time to workout the changes on the XT-IDE address decoding myself; I need to get started with the tools. The port changes I was thinking of are pretty easy to do; adding another 1KB of memory is a bit more involved.


Mike

Well, I have a lot of free time on my hands, so... yeah.

I can zip up a copy of my KiCAD directory for you if you want to try working on the files I've added things to already.

[jmetal88]

Oh, and I'd like to suggest that when the time comes we get the prototype PCBs done by Advanced Circuits at 33each.com. As the name suggests, prototypes are $33 each (not including shipping), which is probably the cheapest we can get boards done at this size without ordering large volumes. They state a minimum order of four boards (although I'm personally exempt from that, being a college student).

[MattCarp]

Argh .. you guys are too fast for me to keep up!

Fast? Nah, I'm just engrossed.

The CS8900 was one of the candidate chipsets to use. But we need some flexibility on the I/O ports that it uses.

Flexibility? You got it. Check out page 75 of the datasheet: "On power up, the default value of the I/O base address is set at 300h. (Note that 300h is typically assigned to LAN peripherals). The I/O base address may be changed to any available XXX0h location, either by loading configuration data from the EEPROM, or dur- ing system setup."

Pretty slick, eh?

On the ethernet controller, I think now the only question is, are the drivers that are available usable, or enough of a start to finish the job? Or, if not, can we roll our own? I think the general answer is "yes", so I think we're making progress here.

Oh, one question, jmetal88, did you drop in the 93C46 configuration EEPROM? It's optional, but to save the current configuration you need it. I could find the CS8900 on the picture, but the EEPROM may be too small for me to discern.

I was hoping to get some time to workout the changes on the XT-IDE address decoding myself; I need to get started with the tools. The port changes I was thinking of are pretty easy to do; adding another 1KB of memory is a bit more involved.

Whatch'ya thinking? I haven't followed the XT-IDE project, but I'd hope it uses fairly free I/O ports and memory addresses. I'm not sure what you mean about the 1KB reference.