Forthcoming XT-IDE Board - Cast Your Vote

[JT64]

Eye-fi support in bios

Eye-fi support in bios

Since a IDE-CF adapter do not need a driver, and i guess a SD-CF adapter do not either, you could make support for EYE-FI wireless SD card in ROM?

You would still need a tcp stack loaded in memory though, but maybe i am dreaming now?

 

[pearce_jj]

This is a GREAT concept - an 8-bit card that provides access to a drive image via WiFi all completely seemlessly and without any code needing to run through the aging PC's CPU. It would make file transfers as easy as they could possibly be - just drop into the image on the hosting PC.

I don't think the Eye-Fi card will do this for us though (I could be wrong but it looks to me more of a flash-drive-with-replication-elsewhere thing).

Food for thought certainly.

 

[JT64]

Well i actually think it support some tcp-ip layer making a share, because it says available on your homenetwork?
I think it is a FTP share or similar, right now i am on linux machine so i can not try it out.

But the manager is free to download can someone check if it is FTP, and if it is twoway connectivity?
http://support.eye.fi/support-resources/downloads/software/eye-fi-manager-desktop-software/

Found some more, there is a thirdparty server for more flexibility.
http://www.engadget.com/2009/03/17/sandalone-eye-fi-server-hack-one-ups-eye-fi-manager/

Eye-fi server python code
https://github.com/tachang/EyeFiServer


JT

 

[RWallmow]

Well i actually think it support some tcp-ip layer making a share, because it says available on your homenetwork?
I think it is a FTP share or similar, right now i am on linux machine so i can not try it out.

But the manager is free to download can someone check if it is FTP, and if it is twoway connectivity?
http://support.eye.fi/support-resources/downloads/software/eye-fi-manager-desktop-software/

Found some more, there is a thirdparty server for more flexibility.
http://www.engadget.com/2009/03/17/sandalone-eye-fi-server-hack-one-ups-eye-fi-manager/

Eye-fi server python code
https://github.com/tachang/EyeFiServer


JT

I am pretty sure the eye-fi's firmware would have to be rewritten completely, as it is it only sends JPG, RAW, or MPG files over wifi, their firmware is designed not to allow anything else to be sent (And you have to buy the higher end PRO or Video card to get RAW or MPEG capabilities, the base card ONLY sends JPG). Also its not designed to receive anything, but again if its firmware were rewritten I don't see why that wouldn't be possible either. Just saying out of the box it only pays attention to files written to it that are JPG, RAW, or MPEG formats, ignores everything else.

 

[JT64]

I read that there is CF network cards, so far i have only seen NE2000 compatible ones and some do not even need a driver?
Could the XTIDE project be made to use such a NIC via CF interface it would be nice. Because i am simply unable to get any NIC for my XT working.

Is there a reason why a CF NE2000 card would not work with a implemented CF reader in XTIDE? Pins, IO, IRQ, DMA?
I know a NE2000 do not work in ISA, but maybe it will work as CF or maybe they are NE1000 compatible?

And what about other types of CF cards, like audio?

http://www.broadcastwarehouse.com/maycom/n>trans-professional-cf-audio-card/1055/product

http://compactflash.org/faqs/faq.ht
CF+ I/O cards include modems, Ethernet, 802.11b WiFi, serial, Bluetooth wireless, digital phone cards, USB, laser scanners, VGA, etc????

JT

 

[RWallmow]

...Is there a reason why a CF NE2000 card would not work with a implemented CF reader in XTIDE? Pins, IO, IRQ, DMA?
I know a NE2000 do not work in ISA, but maybe it will work as CF or maybe they are NE1000 compatible?

And what about other types of CF cards, like audio?...

Yes, there is a reason, the XTIDE ONLY implements IDE, not general CF I/O. CF (and PCMCIA for that matter, since CF is just a shrunk PCMCIA card) have two modes, IDE and I/O, the XTIDE is only an IDE card, no general I/O support, just like most digital cameras or USB Flash card readers also only support IDE mode. CF and PCMCIA I/O Modes are 16-bit, so I am not even sure if it would be possible to implement it on an 8-bit ISA card.

 

[bettablue]

Is there any kind of time frame for when the new revision will be made available?

 

[pearce_jj]

I'm awaiting delivery of some revised prototype boards at the moment. Alan I think is further along with his.

 

[deathshadow]

Silly question, might it not be time to consider making XT-SATA?

Though given how simple SD cards are, with simple devices like the Arduino, Raspberry Pi and even crappy TI Launchpad being able to interface to them -- wouldn't it possibly be simpler to just make a dedicated XT-SDHC board... Given the limited capabilities of those devices I'd think it would be SIMPLER to just forget IDE altogether at this point.

After all when you can get a 3.5" floppy emulator that takes SD cards for less than $30 bucks SHIPPED, how hard could it be?

... and even the slow SPI accessing mode shoves data around faster than the 4.77mhz 8 bit XT bus.

I'd love to see a Tandy plus card form factor SD card only version. Even the 2 gig and smaller standard SD would be overkill in most XT class systems anyways.

 

[RWallmow]

Silly question, might it not be time to consider making XT-SATA?

Though given how simple SD cards are, with simple devices like the Arduino, Raspberry Pi and even crappy TI Launchpad being able to interface to them -- wouldn't it possibly be simpler to just make a dedicated XT-SDHC board... Given the limited capabilities of those devices I'd think it would be SIMPLER to just forget IDE altogether at this point.

After all when you can get a 3.5" floppy emulator that takes SD cards for less than $30 bucks SHIPPED, how hard could it be?

... and even the slow SPI accessing mode shoves data around faster than the 4.77mhz 8 bit XT bus.

I'd love to see a Tandy plus card form factor SD card only version. Even the 2 gig and smaller standard SD would be overkill in most XT class systems anyways.

Good idea in theory, but most those programmable micro-controllers are surface mount and or expensive, which is beyond "kit" assembly, it would have to be a fully finished card, very few members here have the ability to surface mount solder.

Now with that said, many people here would probably be OK with that if the price was right, if you could build and sell the cards for similar prices to the old XTIDE's most people would still bite.

 

[kb2syd]

After all when you can get a 3.5" floppy emulator that takes SD cards for less than $30 bucks SHIPPED, how hard could it be?

Which one?

I'd love to see a Tandy plus card form factor SD card only version. Even the 2 gig and smaller standard SD would be overkill in most XT class systems anyways.

Some of the flash cards won't boot with the tandy versions of DOS. I've had a hard time getting anything larger than 32meg to consistantly format and boot with Tandy's DOS 2.11. There are a few Tandy specific commands in there that are nice to have.

 

[pearce_jj]

The big advantage to the IDE part is that very little electronics is actually needed, in other words IDE and ISA are almost one-and-the-same before we start. Really the XT/IDE just provides some buffering to enable a 16-bit protocol to work on the 8-bit ISA implementation in the PC and XT and of course the BIOS.

That's not to detract from it in any way - just that it is a lot more complicated to make SD work.

 

[Chuck(G)]

The big advantage to the IDE part is that very little electronics is actually needed, in other words IDE and ISA are almost one-and-the-same before we start. Really the XT/IDE just provides some buffering to enable a 16-bit protocol to work on the 8-bit ISA implementation in the PC and XT and of course the BIOS.

That's not to detract from it in any way - just that it is a lot more complicated to make SD work.

At this point, the amount of silicon is pretty much reelevant. By the time you've done the BIOS ROM address and I/O port mapping, you've got a small pile of SSI functions. Adding CF capability would scarcely take more--as a matter of fact, there are HDL libraries to handle a CPLD SD-to-8-bit-parallel. Or you could use a small uC and even add a few peripherals, such a serial I/O A-to-D, etc. Either way, you'd be struggling to fill a half-sized ISA card.

 

[pearce_jj]

Hi Chuck, I'm very interested in this, thanks for posting. Compact Flash integration has proved more difficult than I'd thought because of trace routing and board real estate.

If you don't mind I'll raise this on the DP development thread. Any further input you might have would be very much appreciated indeed.

 

[eeguru]

BIOS is in the works. I updated the schematic/board to Rev.A. I'm anticipating ordering the production batch of PCBs when folks return from Thanksgiving. But it's likely it will be a 4-week turn unless someone highly recommends another fab that can do hard gold fingers for under $20 a board shipped (qty 25) w/ electrical testing (4.1" x 3.2" x 2 layers).

Going to stick with the Atmel part for this run as they are coming back in stock in a few places. It's looking like PCB+parts kit (w/ pre-programming) will be around $50 including shipping when finally ready.

 

[hargle]

FWIW, I have at least 25 people in my queue waiting for XTIDE cards, so if you can get cheaper boards with a qty of 50, please up your order. I'll sell 'em for you.

 

[eeguru]

They drop about -$6 per board going up to 50 quantity. I haven't worked out the parts kit discount, but I would expect that to drop some too.

 

[sergey]

They drop about -$6 per board going up to 50 quantity. I haven't worked out the parts kit discount, but I would expect that to drop some too.

Where can I find the schematics for your board? (There is a link in the thread, but it doesn't work - it asks for authentication)
I am wondering if you have one or two spare pins left on the CPLD? I'd like to implement similar board, adding Intel 82077AA FDC and a 16550 UART, so if there are some available pins on CPLD I can use them for chip selects.

Thanks,
Sergey

 

[Chuck(G)]

Sergey, forget the Intel 82077AA; it's got flaws. Use a National Semi PC8477 instead. Same pinout and it actually works.

 

[eeguru]

Where can I find the schematics for your board? (There is a link in the thread, but it doesn't work - it asks for authentication)
I am wondering if you have one or two spare pins left on the CPLD? I'd like to implement similar board, adding Intel 82077AA FDC and a 16550 UART, so if there are some available pins on CPLD I can use them for chip selects.

Thanks,
Sergey

Here is a temporary link until I can get some time to organize and launch my project site; PLD source included with lots of comments. I can also send you the reflash and IDE test utility if you're interested in taking a lot from a software point of view. There are no free pins atm, but I currently have a mix of dedicated and muxed pins used for panning a second ROM access window around. You could either eliminate those or mux more of them with IDE signals and free up some. Really the meaning of the 16 bit IDE data lines and 3 address lines are chip select specific. The zero wait state line was a mistake on my part. It was introduced on the AT on the lower IDE bus connector and I doubt any XT compatible chipset offers variable wait cycles. So you can re-purpose that pin. Also the 128 MC PLD has lots of extra capacity. I can almost fit the design in the 64 MC part.. almost.

And before you ask, it was just easier from a routing perspective to put the flash part on the buffered data bus. Though now I kinda regret it as it is the reason I can almost but not quite fit the logic in a smaller part.

The current design linked above is tested and functional. Just the BIOS code is pending. I get 330 KB/s on a PCjr and 800 KB/s on a 386/DX33 w/ 8.33 MHz bus. In any case, it should finally be fast enough for 8088 Corruption to run w/o having to buffer from disk!

There is also a preliminary BOM for pricing.

And anyone willing to take a look in review, it would be appreciated.