XEL-CF Compact Flash Adapter for 1088XEL (formerly XEL-I3)

[+Spaced Cowboy]

Quicker test, in fact. If everything in the "Logical drives" menu is dimmed, the PBI BIOS is disabled or somehow unresponsive.

 

 

So on the "troublesome" card, I have logical drives D1..D9 and DJ..DO as 'SIO' drives that appear to be selectable and not dimmed out.

 

Initially I had problems getting my Mac to read the card formatted by SDX, but I got that working using:

simon% diskutil list
/dev/disk0 (internal, physical):
   #:                       TYPE NAME                    SIZE       IDENTIFIER
   0:      GUID_partition_scheme                        *251.0 GB   disk0
   1:                        EFI EFI                     209.7 MB   disk0s1
   2:                  Apple_HFS mini                    250.1 GB   disk0s2
   3:                 Apple_Boot Recovery HD             650.0 MB   disk0s3

/dev/disk1 (external, physical):
   #:                       TYPE NAME                    SIZE       IDENTIFIER
   0:     FDisk_partition_scheme                        *8.0 GB     disk1
   1:             Windows_FAT_32                         5.2 GB     disk1s1
   2:                       0x7F                         2.8 GB     disk1s2

[note: the partition in question is disk1, partition 1 for me. If you're doing this yourself, THIS MAY BE DIFFERENT!]

simon% diskutil eraseVolume FAT32 XFER /dev/disk1s1

... at which point I had a useable volume called 'XFER'

 

On a brighter front, using a 30 MB/sec 8GB card from Sandisk works just fine (although I haven't worked out how to preserve the D1:, ... Dn: logical drives through a power-cycle yet. At the moment, if I do:

 

• set D1: as the first 8000 sectors in FDISK
• power off then on again
• Type D1: then DIR

 

... I'll get another "138 Device does not respond". If I then type FORMAT and write a new directory with 'B', I can list my directories again. I can even get a directory listing after removing and re-inserting the card, then doing a shift-reset to reload the card config

 

Still, this is progress

[flashjazzcat]

Are you saving the partition table in FDISK? If you format a volume in SDX (or build directory), the entire folder tree is nuked. Yet you say it allows you to list your folders again.

[+Spaced Cowboy]

Are you saving the partition table in FDISK? If you format a volume in SDX (or build directory), the entire folder tree is nuked. Yet you say it allows you to list your folders again.

 

 

Yep, at least I think so. It's been a while since I used spartados

 

Basically I create n partitions, n typically being 4 or so, and within FDISK I then assign them each a drive. Then I go to the menu, select 'write', and then finally select 'exit'

 

This dumps me back to dos, at which point I can type FORMAT, select unit 1 (which comes up as 'A'), and then type B to build the directory, go through the 'are you sure' prompts and then hit Esc to exit.

 

At this point I can DIR D1:

 

Now if I power off, then power on again, at the D5: prompt, if I type DIR D1: ... I get the 138 message.

 

I'm just going to go and try this again to make sure the above is accurate

 

 

[Edit] Gaaah. I'm a moron. I must have screwed up something along the way because the above just worked for me without problems. Sorry for the noise...

[flashjazzcat]

Right: you've got a problem if the partition table isn't loaded on a cold boot, for sure.

[+MacRorie]

I just discovered this SanDisk Store run by Digital River that has some very attractive pricing on the 30 MB/s Ultra CF cards.

 

Buy_SanDisk.png

 

So I placed an order for all 3 sizes, and will be running tests to see if they are compatible.

 

Assuming that I actually do receive them (fingers crossed), and discover that they do work, I'll be sure to post an update. BTW, if someone else attempts to purchase some of these, don't click on the PayPal button, but instead use the Check Out button. From there you will be able to enter your details and select PayPal as your purchase preference or a credit card.

 

Note: The landing page is not secured, but the cart is.

Just for completeness sake, I have attached a spreadsheet of the results I have had from a variety of CF cards.

 

Just as a side note: the 16G ***50mb/s*** cards will function in the loader far more "reliably" than at the CL/DOS. Mainly because FDISK does not see them. FJC, I know you are saying that the loader is less forgiving, but it will usually "see" even problematic cards whereas FDISK will not.

 

The 16G 30MB/s listed above are definitely compatible as I have a number of them from Amazon (the test results are in the spreadsheet). Now, that presumes that the place listed is selling the genuine article. I ordered some of each kind as well.

1088XEL CF tests.pdf

[+MacRorie]

Yep, at least I think so. It's been a while since I used spartados

 

Basically I create n partitions, n typically being 4 or so, and within FDISK I then assign them each a drive. Then I go to the menu, select 'write', and then finally select 'exit'

 

This dumps me back to dos, at which point I can type FORMAT, select unit 1 (which comes up as 'A'), and then type B to build the directory, go through the 'are you sure' prompts and then hit Esc to exit.

 

At this point I can DIR D1:

 

Now if I power off, then power on again, at the D5: prompt, if I type DIR D1: ... I get the 138 message.

 

I'm just going to go and try this again to make sure the above is accurate

 

 

[Edit] Gaaah. I'm a moron. I must have screwed up something along the way because the above just worked for me without problems. Sorry for the noise...

Simon, just out of curiosity, did you have anything else in the drive chain? (SIO2SD? SIO2PC?) That might give you variable results when accessing D1: I usually set all HDD drives to D5 and above (E:-O:) in the partition table. This way the A;-D: will be floppies. I mean it is *possible* to have D5-D15 as floppies, but that is generally not the case physically (e.g., 810/1050/XF551 can only do 1-4). That is just my practice, however. YMMV.

[+Spaced Cowboy]

Simon, just out of curiosity, did you have anything else in the drive chain? (SIO2SD? SIO2PC?) That might give you variable results when accessing D1: I usually set all HDD drives to D5 and above (E:-O:) in the partition table. This way the A;-D: will be floppies. I mean it is *possible* to have D5-D15 as floppies, but that is generally not the case physically (e.g., 810/1050/XF551 can only do 1-4). That is just my practice, however. YMMV.

 

 

Nope. I have no excuses.

[flashjazzcat]

Simon, just out of curiosity, did you have anything else in the drive chain? (SIO2SD? SIO2PC?) That might give you variable results when accessing D1: I usually set all HDD drives to D5 and above (E:-O:) in the partition table. This way the A;-D: will be floppies. I mean it is *possible* to have D5-D15 as floppies, but that is generally not the case physically (e.g., 810/1050/XF551 can only do 1-4). That is just my practice, however. YMMV.

That's good practice (I start HDD partitions at C:), but a HDD partition always overrides any SIO drive on the same drive number, just as an ATR mounted on D1: via the loader will obscure a serial device on the same ID.

[+Spaced Cowboy]

Just out of curiosity, the machine boots to the D5: prompt. Is there supposed to be anything provided internally at D5: by default ? Or is the idea that you create a CF card with D5: as the disk, and then you'll be "in" your CF card by default ?

 

(Because I get my old friend, error 138

[+DrVenkman]

Didn’t you say earlier you had set your boot drive option to D5:? Personally I leave the boot drive options alone and just redirect my CONFIG.SYS to the first drive partition on my CF card.

[+Spaced Cowboy]

D5: was just the default as it came shipped. I was just wondering if this was supposed to be a "special" drive with BIOS-related stuff pre-installed from the U1MB or something

[flashjazzcat]

Probably the way MacRorie left it after testing, since his partitions apparently start at D5:. You can nuke customisations with Save and Exit->Restore Defaults (or by pressing "D").

 

The reason there are two different boot options ("Boot Drive" and "CONFIG.SYS") is to allow easy switching between SDX and some other, disk-booted DOS. The CONFIG.SYS setting is used when SDX is active and the Boot Drive option is used when SDX is turned off. The OS awkwardly insists on resetting the SIO unit number to 1, but since SDX doesn't actually perform a disk boot, simply setting DUNIT to the desired value works. The CONFIG.SYS setting is - conversely - automatically ignored by the OS when SDX is not active, and in this case the Boot Drive setting takes over, forcing DUNIT to the desired value when reading the first three sectors of the disk while the boot flag is set. Some DOS packages are as obstinate as the OS, however, in as much as they expect everything to reside on D1:, so it can be a bit of a faff to get them to boot from a higher drive ID.

Edited March 6, 2018 by flashjazzcat

[+MacRorie]

D5: was just the default as it came shipped. I was just wondering if this was supposed to be a "special" drive with BIOS-related stuff pre-installed from the U1MB or something

 

 

That's the way I set them up prior to shipment. You can change it to anything you want. Be aware that the D5: drive is a set as "bootable" in FDISK/APT as well. This would allow you to install a disk-based DOS to boot from. However, you are free to junk it all and start over! :-)

[+MacRorie]

 

I'm totally willing to believe that there are differences between manufacturers, and it's entirely possible the XEL-CF-2 is somewhat pushing the envelope of what is acceptable to all cards and that only some card manufacturers are within that cross-section.

 

However, it seems odd to me that the same hardware, running different programs, can either see or not see the same card depending on which program is running. The XEL loader has no problems, ever, accessing the FAT32 partition and scanning through directory hierarchies. FDISK can't even find the device. I'm not sure how "different manufacturers" can explain that unless I'm missing something.

 

 

Simon,

 

I went back up the thread to make sure I knew what we were all talking about, so please excuse the resurrecting a past response. One of the things I discovered was that some cards would, indeed, seemingly operate correctly in the loader and then not in the CL. HOWEVER, the loader behavior was not consistent over time. Eventually it fails or crashes. We developed a multi-step procedure to try and verify cards. I think it works, but only time and enduser experience will tell! I hope the ones you ordered from Amazon give you no trouble.

[+Spaced Cowboy]

Thanks for the update MacRorie

 

The card from Amazon (the 8GB, 30MB/sec Sandisk one) seems to work just fine. I haven't tried the SD-card adapter yet, in fact, because the CF card isn't giving me any issues at all

 

Cheers

Simon

[flashjazzcat]

On the face of it, the loader assumes the hardware works and makes no attempt to compensate for transmission errors. The PBI driver does attempt to compensate for transmission errors, but also transfers data at a higher rate and only by complete 512 byte sectors at a time. When the hardware doesn't work, short, slower, fragmented transfers (loader) may appear to work to some extent, or the error correction in the PBI driver may work better. In either case, it's probably better to run some empirical tests than to try and intuit reliability from observed behaviour. RWCRC run on APT partitions and large executables launched from the loader, if successful, are good indicators of a stable setup.

[+mytek]

Update on New XEL-CF board design coming down the road...

 

Flashjazzcat and I have been working diligently on making sure that his new U1MB 1088XEL BIOS will be able to detect the difference in the old vs the new CF board, so that one size fits all. To this end I gave the new version a hardware signature so that it can be easily distinguished, and thus allows the BIOS to adjust itself to suit the board in use. It's really rather nice how it all turned out, being automatic in nature and not requiring separate builds between the two types of CF boards. This is especially important since there will be quite a few of the original boards in use, and the new one that's still in development might be awhile before it's ready for prime time.

 

What does work?

 

Edit: Everything that was noted to be good in MacRorie's post, and ...

 

CF card compatibility is looking better, with full support for the Sandisk Ultra Series (25, 30, and 50MB/s), and also Transcend 80x cards are now on board. I also have had success with an old Lexar 32MB and a Crucial 128MB cards as well, so that presents a few other options. Lexar appears to be a re-branded SanDisk.

 

 

What still doesn't work?

 

The Kingston flower series with the triangle symbol on the label.

 

Although the nearly identical looking one with the medallion does appear to work, apparently being of a newer release.

 

And this Simple Tech card didn't work for me.

And last but certainly not least, any of the fakes coming from eBay have pretty much zero chance of working.

[john_q_atari]

Is it the case that more cards will work now with the current xel-cf after a U1MB BIOS update? Or did you mean to say that more cards will work with the new CF board design in the future?

[+mytek]

Is it the case that more cards will work now with the current xel-cf after a U1MB BIOS update? Or did you mean to say that more cards will work with the new CF board design in the future?

 

That's why I prefaced the post with "Update on New XEL-CF board design coming down the road..." so everything that followed was actually referring to the new hardware design that's still in the works. It's just not practical to 'fix' the problem in the software or BIOS side of things, thus the reason for the new board in development. As for what cards work reliably with the 'original' board, MacRorie's list he recently posted would be a good reference.

 

While designing the new board for increased CF card compatibility, it was also decided that implementing a Swap Button circuit into it just made good sense, so this new feature is an added bonus.

 

Latest test of new Swap Button Circuit with newest U1MB 1088XEL BIOS in development.

 

https://www.youtube.com/watch?v=8viTFqLYBEA

[flashjazzcat]

It's just not practical to 'fix' the problem in the software or BIOS side of things...

 

I'd have said 'not possible'.

[_The Doctor__]

Michael is very proud of his work, I heard it in his voice. Nice job.

[+mytek]

Michael is very proud of his work, I heard it in his voice. Nice job.

More like just plain happy to see it work. You guys are just seeing the tip of the iceberg when it comes to the development cycle on this new CF board design. At some point in the near future I'll have to zip up the BETA Group PM that was started about a year ago and upload it in the public area, it chronicles the entire back room development of the 1088XEL and the CF boards, and might make for a fascinating read.

[john_q_atari]

I just discovered this SanDisk Store run by Digital River that has some very attractive pricing on the 30 MB/s Ultra CF cards.

 

....

 

Assuming that I actually do receive them (fingers crossed), and discover that they do work, I'll be sure to post an update. BTW, if someone else attempts to purchase some of these, don't click on the PayPal button, but instead use the Check Out button. From there you will be able to enter your details and select PayPal as your purchase preference or a credit card.

 

Note: The landing page is not secured, but the cart is.

 

 

Were you able to determine if these new cards you purchased worked reliably in the current XEL-CF?

[+mytek]

XEL-CF3 News (an updated version of the IDE/Compact Disk Interface Board)

 

The XEL-CF3 Design has been completed and a test run of boards has been ordered from ALLPCB.

 

Schematic: XEL-CF3_3-14-2018_schema.pdf

 

 

Now works reliably with all these cards (and probably a lot more as well)...

 

The last ones that I tested are the 3 SanDisk Ultra 30 MB/s on the right (4, 8, and 16GB), which I did today (see video below). And some really good news is that the 10 fake SanDisk 1GB cards I purchased off of eBay late last year also now work in this new design, which is certainly a good sign that compatibility has greatly improved.

 

These Fake SanDisk 1GB cards now work as well. They ID as SMART CF not SanDisk.

 

Got these SanDisk Ultra 30 MB/s CF cards in the mail and gave them a quick test drive (see video). You can purchase them from the SanDisk Store which I talked about in this post a while back. I had originally assumed they were NEW items, but noticed that the word REFURB is engraved on the front. Although I'm pretty sure this was done in a SanDisk authorized facility, unlike what happens when buying from China via eBay where they say they are NEW, but then turn out to be something entirely different than what you thought you were buying.

 

I haven't tested the 30 MB/s Ultra's on the original XEL-CF-][ board, but rumor has it that they may work for that as well.

 

https://www.youtube.com/watch?v=vh3QwCspyME

 

Flashjazzcat explained the discrepancy in the XEL-Loader's reported FAT size vs actual CF Card size in a YouTube comment...

FAT overheads are rather large, especially if the cluster size is small. The reported capacity describes the volume size after formatting.

[+mytek]

Were you able to determine if these new cards you purchased worked reliably in the current XEL-CF?

 

I haven't tested them with the original XEL-CF-][ board, because my new prototype is setup on my only running system and I'm afraid to remove it just yet. However MacRorie should be getting some of these soon, and will likely post some results.