• Please review our updated Terms and Rules here

Forthcoming XT-IDE Board - Cast Your Vote

Forthcoming XT-IDE Board - Cast Your Vote

  • As original XT-IDE, with a 40-pin header only

    Votes: 4 10.5%
  • With a 44-pin header and board space to mount a 2.5" IDE HDD (i.e. a hard-card)

    Votes: 7 18.4%
  • With an optional Compact Flash socket (as master or slave) and a 40-pin header

    Votes: 26 68.4%
  • With a Compact Flash socket only

    Votes: 1 2.6%

  • Total voters
    38
  • Poll closed .
At this point, I would have to agree that having a CF card alone at least for me, would be perfect. I am actually thinking of installing an XTIDE into my system, but I have some questions about implimentation.

Initially, I'm going to spend the time and money on buying a Western Digital MFM controller, and another full height drive. I'm doing that for later purposes Then I'll want to swap out the controller and hard disk drives for an XTIDE. BUT here's something I'm not sure has been addressed. I saw two adapters that interest me highly. One is a rear expansion slot mount for a CF card, and the other is a dual CF card adapter. With tqo CF card extenders I could actually install 2 of the expansion slot cover brackets, so that there would be two CF cards sticking out of the rear of the expansion unit. (Does this make sense?)

Will the XTIDE support two CF cards in this manner? or will the dual CF adapter make it appear that there is only one CF card?

My intentions are to use one as a boot device loaded with DOS, any and all configuration files and drivers. The other would be partitioned into as required by the OS into several partitioned and formatted so that I have one partition for DOS utility and shell programs, another partition for all of my business applications, and finally one or more for my games collection.

Would the XTIDE allow a dual CF card setup like what I described?


It's awkward to have 40-pin on it too, and personally I'm not convinced there is any point since CF cards are so cheap and have more capacity that could ever be used in an XT.
 
Will the XTIDE support two CF cards in this manner?
Any IDE controller, XTIDEs included, all support 2 devices, as long as they are properly configured as master+slave. A CF adapter merely changes the interface from CF to IDE-the controller doesn't know the difference.
I'm uncertain of how a dual CF adapter works, but I'm going to guess that they are just putting both master+slave on the same device, so that dual CF adapter will consume the 2 available devices that the controller can handle.
 
Last edited:
If you're going to put an XTIDE into it why fool around? Just put it in with an IDE drive and be done with it! Nobody will ever see the IDE drive anyway. :) Only you will know what's actually 'under the hood'. You'll never have any storage problems as an IDE drive's capacity would far exceed anything you could ever require with a PC. And, if you get lucky, you can put the IDEXT next to the MFM/RLL controller and run *both* drives simultaneously. With this setup you'll have a choice of which drive you choose to boot from, as well.
 
Well, my new board only has a CF socket:

xf-cf.jpg

As said CF cards in true-IDE mode can be master or slave depending on their CSEL pin at reset, although it is widely noted that many cards do not work properly in slave mode. It's not something I've ever tried or experienced though.

For me I can't see much point in using anything other than CF since they're so cheap and so fast and have more capacity thatn you'd ever need for any 8-bit system. Having two is interesting, but why not just have one? It's not like they're hot pluggable anyway so just carry the (tiny!) OS around with each card. Against just my opinion :)
 
I'm using CF cards (in adapters) for a lot of 16/32 bit systems as well. And, as you said, for vintage OS use, a 4 GB CF is huge. I've got one now in a system with Win98SE, Warp 3 and Warp 4 and still have lots of room left over.
 
Question: Can the CF holder be moved further in or on top? then w/ a CF card in place this guy can just sit behind a slot cover. Not superbly stable, but then I doubt to many people are lugging their PC/XTs/Clones around on a daily basis.
 
The idea with it is that the CF card is accessible through the slot opening, so the 'disk' can be changed easily (with the system powered down). I've had some custom ISA brackets made up for it with a slot for the card.

But the Eagle sources are all available, anyone can do with it whatever they like :)
 
Last time we went through the bracket thing there wasn't one available that worked. Did you have them made in quantity or just for your own needs? If its the latter I am guessing most people will leave theirs open to air. And honestly how often do you need to "change" the disk? Yes, some people will but most people will use it as a HDD w/ one OS and a set of programs.
 
I discovered that with suitable AutoCAD file, it can be made in Qty 50 via laser cutting or CNC punch-press quite cost effectively; I had a quote that was about £3 each (similar price as the Keystone 9202). I've had a sample made, but not received it as yet, per the sketch on my wiki.
 
I discovered that with suitable AutoCAD file, it can be made in Qty 50 via laser cutting or CNC punch-press quite cost effectively; I had a quote that was about £3 each (similar price as the Keystone 9202). I've had a sample made, but not received it as yet, per the sketch on my wiki.

Sooooo when will you have some plug and chug units up for order ;P? Honestly, I don't know if I would ever use it but it is a useful thing to have on the "bench" just in case so I'd definitely "queue up" for one :)
 
The next step I think is to get pricing for an assembled or part-assembled shot-run, since as you hint at there isn't much apetite to do the SMT soldering 'at home'. For those willing to tackle SMT, boards are available now.
 
Last edited:
The next step I think is to get pricing for an assembled or part-assembled shot-run, since as you hint at there isn't much apetite to do the SMT soldering 'at home'. For those willing to tackle SMT, boards are available now.

Assuming they are under $50 assembled, I'd probably buy at least 2 of 'em.
 
Thank you for posting that. I was considering listing one on eBay too, to gauge market value.
 
Assuming they are under $50 assembled, I'd probably buy at least 2 of 'em.

Hmmm... $50 is a bit much for me but then again I have no idea how much the components cost and/or how long it takes to put one together. I'd say $25-$35 for me but as I said earlier I don't want one for immediate use. If I needed one now things would be different.
 
I can arrange a group buy here in Brazil, maybe 20 cards...
Please tell me who's in charge of selling these nice cards ;).
 
I own an ADP-50L w/2.18T BIOS. I put it in my Tandy 1000SX and attached it to a CF card using a CF-IDE adapter. I decided to benchmark it using Michael Brutman's iotest. Using the slow speed of the system (8088 @ 4.77MHz), I got an average read of 293.24K with iotest read 1024 5 and an average write of 260.56 using identical write settings. How does this compare with current XT-IDE tests?
 
lucasdaytona - thanks for that. The card isn't quite ready yet; I want to test omiting a load of the caps and then get a quote for a production run of boards with the SMT stuff done.

Great Hierophant - thanks for posting this; AFAIK this is the first opportunity to directly test XT/IDE against a commercial 8-bit IDE adapter. Today, that looks faster but all tests are different; would you mind running the tests using my simple file system based tester? (Documented here). Set BUFFERS=99 in CONFIG.SYS (and reboot) before running. Also please check the results broadly tally to real time - it transfers 4MB, so if the card will write at 260KB/s that test should take about 15 seconds or so.

There is some scope to improve the read speed through some loop unrolling I think, and the write speed I'm working on seperately. I would note that the new xtide-universal-BIOS is a bit slower as it's considering lots of possibilities. Ultimately I plan to do a custom 'thin' BIOS for the card, directly derived from the xtide-universal BIOS.

Many thanks!
 
Last edited:
Okay, the results using disktest with no parameters, buffers=99 and MS-DOS 5.0 on the Tandy 1000SX with the 8088 @ 4.77MHz :

Write Speed : 274.16KB/s
Read Speed : 316.05KB/s
8K random, 70% read : 9.9 IOPS
Sector random read : 14.1 IOPS

Average access time is 71ms.

I used the slow speed for easy comparisons with other testers. I have a 286 accelerator and when in use the speeds show a marketed improvement, up to 100% in some cases. I see the most direct comparison is with "IBM 5155 (PC Portable) SD card via modified XT-IDE 250 173 11.3 16.8 60ms". I would think that a CF card, which supports IDE natively, would be faster than an SD card, which does not. The results are extremely promising. I could let my ADP-50L go for an XT-IDE if results similar to that could be obtained. Saving a bay or a slot in a Tandy 1000 is always a good thing.

The ADP-50L uses standard 74 series logic chips, an EPROM and a pair of "mystery chips" with a white label stuck on each. Turns out those chips are standard PAL16L8s. I find it interesting that for a device that relies on memory mapping, there is no actual RAM on the board. It uses the DACK0 signal found on the 8-bit ISA connector, presumably to fool the system into thinking there is expansion ram on board.
 
Last edited:
Thanks for posting this! So the memory-mapped IO explains the speed. Using that technique the card doesn't need any RAM itself; to transfer the sector the card logic arranges itself to expect a read or write to a consecutive area of RAM space and it transfers those individual reads or writes to the IDE interface, collecting the 8-bit pairs via the shift registers you've spotted. Whereas with IO port mapping, the data is transferred continually to the same IO port (which is still the same address lines, albeit only the first 10, but /IOR or /IOW are asserted instead of /MEMR or /MEMW). Memory mapped is faster as the instructions are cleaner (word moves) and take less time to execute.

My compact-flash board might support memory-mapped IO at some point, in as much as all the electrical connections required are there, but the CPLD probably hasn't got enough logic capacity and in any case I'm still struggling with port-based IO at the moment, making writes faster. Today it does about 210KB/s read and 110KB/s write with v2 BIOS on 8088/4.77MHz; hopefully this can be improved to about 230, maybe 240KB/s for reads and writes with a combination of new logic and slimmed down BIOS.

Alan - how do the ADP-50L numbers compare to your expectations for memory-mapped IO?
 
Back
Top