Drives for the IDE Plus 2

[flashjazzcat]

It's interesting -- with MyDos, the setting for Write with Verify setting has no effect, suggesting it always verifies. The R/W times are identical with/without Verify.

Does this not suggest that the IDE BIOS never verifies (meaning that SIO commands 'W' and 'P' have the exact same effect as far as the IDE driver is concerned)?

Edited July 7, 2011 by flashjazzcat

[+Larry]

Hi Jon-

 

Yes, you are exactly right! I incorrectly "remembered" that there was about 7 KB/S difference between verify and no-verify that I had seen previously. I just checked with the Sandisk card and it also shows no difference between verify and not (just faster, period).

 

What I actually had confused was the difference between verify and no-verify with the MyIDE 4.6 (which I just checked) at about 5.5 KB/S difference. Too much hardware!

 

-Larry

 

It's interesting -- with MyDos, the setting for Write with Verify setting has no effect, suggesting it always verifies. The R/W times are identical with/without Verify.

Does this not suggest that the IDE BIOS never verifies (meaning that SIO commands 'W' and 'P' have the exact same effect as far as the IDE driver is concerned)?

[+Larry]

Followup on Toshiba 1.8" MK2006GAL HDD that only showed 125 MB instead of its 20 GB actual capacity.

 

Kurtm was kind enough to link to me a couple of tools used to restore capacity "locked out" by the mfgr/supplier. The capacity restore tool did see the full 20 GB, and only 125 MB had been allocated. Windows partitioning software could not see the true capacity, it said "Bad Drive." But after using the tool to restore the capacity, then the partitioning software saw the full 20 GB also, and was able to allocate the rest of the space. The drive continues to work very well with the IDE+2, and the "restore" operation did not even overwrite the KMK partitioning info.

 

The drive is so quiet and vibration free, it is impossible to hear anything, and one can just barely feel the drive spinning. Interestingly, I purchased another identical drive from the same supplier, and it showed 20 GB right out of the box. (?)

 

-Larry

[Rybags]

Write + Verify is just an Atari disk drive command. As far as CIO / SIO / the OS is concerned, nothing extra is done.

 

AFAIK, there's not an ATA/IDE equivalent - so the write+verify if used is just executed as a normal write operation.

[kurtm]

Kurtm was kind enough to link to me a couple of tools used to restore capacity "locked out" by the mfgr/supplier. The capacity restore tool did see the full 20 GB, and only 125 MB had been allocated. Windows partitioning software could not see the true capacity, it said "Bad Drive." But after using the tool to restore the capacity, then the partitioning software saw the full 20 GB also, and was able to allocate the rest of the space. The drive continues to work very well with the IDE+2, and the "restore" operation did not even overwrite the KMK partitioning info.

 

That is excellent news!

 

--Kurt

[+bf2k+]

I ran into a CF card that doesn't work with IDE+2.0.

 

SANDISK 256mb

 

Identifies itself as 'Hyperstone ATA'

 

This is an older card that I bought to use with MyIDE years ago.

KMKDIAG fails on the MBR test.

 

Also says LBA not supported.

[+remowilliams]

The Kingston 4G 'flower card' I was initially using failed writes intermittently. Trying a 8MB Sandisk and a 4GB Transcend afterwards has been flawless so far.

[+Larry]

More on backing up fixed disks for the +2.

 

Although it is quite simple to "cable up" a 3.5" master-slave arrangement; not so easy with the 2.5" laptop connector. I did find pieces that should work. But in the meantime, I remembered another solution. Not quite so elegant as sector copying from a master to a slave, but still very doable and likely at less cost.

 

I attached my IDE+2 1.8" Toshiba system drive to a non-powered IDE to USB drive adapter and also plugged in a 2 GB compact flash card in a USB reader in a second port.

 

Then the magic -- I ran the Miray Software HDClone backup/cloning program on my PC. I'm still using version 3.2 that I purchased several years ago, but their new 4.x FREE version looks like it would work as well. Total backup and verification time for 50,000 sectors is under 1 minute. After the copy, I booted my Atari off the CF card and ran (MyDos) VTOC FIX to verify the copy. Perfect Copy (as expected).

 

Although the free version is limited in its features, it look to be more than adequate for our use with the A8.

For anyone interested, here's a link to Miray's HDClone:

 

http://www.miray.de/products/sat.hdclone.html

 

-Larry

 

P.S. does anyone understand the significance of their logo? Looks like a sheep on the HD platter, so thought it might have something to do with the German word for "sheep" or "wool?"

[Rybags]

Wasn't a sheep the first animal to be cloned ?

 

I've not played with backup or import solutions yet... my plan is to create a setup using a HDD image in Altirra then put onto flashable media and see if it goes OK on the real thing.

[+Larry]

Yes! First mammal? You win a -- oh wait, that was another thread!

 

Thanks, Gary. I never would have thought of that angle.

 

-Larry

 

 

Wasn't a sheep the first animal to be cloned ?

 

I've not played with backup or import solutions yet... my plan is to create a setup using a HDD image in Altirra then put onto flashable media and see if it goes OK on the real thing.

[drac030]

I came across another Kingston "the flower" which does not work correctly, i.e. passes all tests, but in DOS operations there are occasional errors 144 occurring during writes. Since the card was not mine, I didn't investigate the direct cause. A spinning platter disk (IBM Travelstar, IIRC) worked fine on the same setup.

 

Also, I came across a setup which had lots of files saved on the CF card (which proves it is working for the owner), but when the interface with the cards were plugged into my 130XE, it started to behave like the above Kingston. Symptoms: erros 144 and occasionally 138 during write. I don't remember the card type unfortunately. But it looks like a hardware problem again.

Edited August 4, 2011 by drac030

[+bf2k+]

I came across another Kingston "the flower" which does not work correctly, i.e. passes all tests, but in DOS operations there are occasional errors 144 occurring during writes. Since the card was not mine, I didn't investigate the direct cause. A spinning platter disk (IBM Travelstar, IIRC) worked fine on the same setup.

 

Also, I came across a setup which had lots of files saved on the CF card (which proves it is working for the owner), but when the interface with the cards were plugged into my 130XE, it started to behave like the above Kingston. Symptoms: erros 144 and occasionally 138 during write. I don't remember the card type unfortunately. But it looks like a hardware problem again.

 

That is exactly what happened to me using a brand new Sandisk Ultra 2gb CF.

 

My HDDs all work fine.

[+remowilliams]

I came across another Kingston "the flower" which does not work correctly, i.e. passes all tests, but in DOS operations there are occasional errors 144 occurring during writes.

 

I can confirm that. The 4GB Kingston 'flower' I originally had in seemed to work fine at first, but I had to switch it out due to errors.

 

 

[+bob1200xl]

Since the CF isn't reading anything during writes, and you have no kind of CRC or such, how can the hardware post a 144? The CF controller doesn't care what data you send. Flash memory has to go 'do things' sometimes. The software needs to allow for that and not get mad... Can anyone read out the regs when the error occurs?

 

Bob

 

 

 

I came across another Kingston "the flower" which does not work correctly, i.e. passes all tests, but in DOS operations there are occasional errors 144 occurring during writes. Since the card was not mine, I didn't investigate the direct cause. A spinning platter disk (IBM Travelstar, IIRC) worked fine on the same setup.

 

Also, I came across a setup which had lots of files saved on the CF card (which proves it is working for the owner), but when the interface with the cards were plugged into my 130XE, it started to behave like the above Kingston. Symptoms: erros 144 and occasionally 138 during write. I don't remember the card type unfortunately. But it looks like a hardware problem again.

[flashjazzcat]

Since the CF isn't reading anything during writes, and you have no kind of CRC or such, how can the hardware post a 144?

It's the device firmware / BIOS / DOS which will report an error 144. This can be caused by data transmission errors, resulting in a scrambled partition, which in turn results in DOS thinking a file is locked (or other unspecified behaviour resulting from scrambled directory entries).

 

The driver might reasonably return error 138 if DRQ didn't go high when expected, which can happen as a result of transmission errors resulting in dropped bytes.

 

EDIT: It's mainly DOS which will return error 144, although if the implementation supports partition locking and the partition table gets corrupted by a bad sector write, the driver itself might then return error 144 if a write was attempted on a partition whose write lock bit had been accidentally set.

Edited August 7, 2011 by flashjazzcat

[+bob1200xl]

Yes - but I think you are still saying that the CF card cannot report an error.

 

Is there any indication that there was a data error? (data was stored incorrectly?) If the sector is write-protected, that is outside the CF card hardware.

 

Is the error always on a write?

 

This is reasonably important since you may see errors on the Kingston card that you do not see on others, even though the problem happens on all. Even on a read, the only way to know if the data is correct is to do a compare.

 

In the bad, old days, tape drives read back the data immediately after writing it. (two sets of read coils in the head - only worked in one direction, of course - that's why you can't write backward but you can read backward) The new-fangled disk drive could not read-after-write, which bothered the heck out of large financial institutions. About the second time they wrote data and couldn't read it back correctly, CRC was added to the controllers. This, at least, gave them the ability to detect bad things and correct some minor errors. Eventually, they added mirroring and it's been downhill ever since.

 

Anyway, we can do a mirror if people are really worried about data integrity. It is slower, but in real-world applications it shouldn't matter too much. Would not slow down really intensive apps, like backup from CF-to-CF or HDD-to-CF. If there are no rotating devices, you don't have to sync up much, just write/read.

 

Bob

 

 

 

Since the CF isn't reading anything during writes, and you have no kind of CRC or such, how can the hardware post a 144?

It's the device firmware / BIOS / DOS which will report an error 144. This can be caused by data transmission errors, resulting in a scrambled partition, which in turn results in DOS thinking a file is locked (or other unspecified behaviour resulting from scrambled directory entries).

 

The driver might reasonably return error 138 if DRQ didn't go high when expected, which can happen as a result of transmission errors resulting in dropped bytes.

 

EDIT: It's mainly DOS which will return error 144, although if the implementation supports partition locking and the partition table gets corrupted by a bad sector write, the driver itself might then return error 144 if a write was attempted on a partition whose write lock bit had been accidentally set.

[flashjazzcat]

Well, I might be wrong (I don't even have a Kingston "flower" CF card, and I haven't spent much time testing my IDE Plus yet, since results on my 800XL haven't been too positive), but in my experience over the past year coding up IDE drivers on the Atari, the card doesn't need to be experiencing hardware errors of any kind and probably won't be. With the MyIDE and SIDE, 99.9% of the time, 138 and 144 errors were down to hardware which is operating perfectly but which is dropping bytes because of timing issues (0.01% of the time it was a dying CF card! ). It should be noted that SIDE is ten times (estimate) less picky than MyIDE about what combination of CF card and Atari machine it will work stably with, so such errors are rare.

 

Sure - the CF card can report an error via the IDE status and error registers and this can be interpreted and relayed by the interface firmware, but I think the only hardware error I've ever seen and acted upon in driver code is effectively "sector number out of range". The rest of the time, the problem has been down to the drive going into an error state because of incomplete or overrunning buffer reads/writes, resulting in indeterminate DRQ / RDY states.

 

Of course it could be something quite different here: I'm not sure how the IDE Plus hardware works.

Edited August 7, 2011 by flashjazzcat

[drac030]

Since the CF isn't reading anything during writes, and you have no kind of CRC or such, how can the hardware post a 144?

 

Who told you it was the _hardware_ generating the error 144?

[Rybags]

My dual CF adaptor arrived the other day and I tried a generic CF card.

 

Also got some intermittent errors. Took a couple of times to partition and format 2 logical drives on the thing.

 

Tried the RWTEST benchmark on the 256 byte partition, results weren't exactly sparkling either.

[HiassofT]

The 4GB Kingston 'flower' I originally had in seemed to work fine at first, but I had to switch it out due to errors.

I have a similar 4GB Kingston card here:

 

When I tested Jon's SDX MyIDE driver with this card connected to my homebrew MyIDE-like interface I was experiencing similar problems (occasional errors and data corruption).

 

It seems like this card is especially picky about ringing on the control lines (I verified with my scope that there actually was some ringing on DIOR, DIOW and CS0). After adding series termination resistors to these lines (see ATA8 spec, page 18) ringing was significantly reduced and the errors disappeared. (since I didn't have any 22 ohm resistors here I used 33 ohm also for DIOR and DIOW, but this worked, too)

 

I don't have an IDE Plus interface, so I don't know anything about the signal quality. But it should be very easy to check, just hook up a 50+MHz scope to the pins of the CF connector or the IDE connector of the CF adapter (with the CF card plugged in, of course).

 

so long,

 

Hias

Edited August 8, 2011 by HiassofT

[drac030]

I have asked simius about this, and he replied that the resistors are already there, so the reason must be different. What reason - it is to be found out yet.

[venom4728a]

has anyone tried one of these yet? has dual card capability and looks great...

 

eBay Auction -- Item Number: 140523993608

 

Thats the cf adapter I purchased for mine, I have not tried it yet, Has any one found a nice case to put these in yet.

 

Robert

[Rybags]

That looks like it's got decent build quality. The dual one I got is mostly bare, only has metal brackets at the side to replicate the screw-holes that a 2.5" HDD gives you.

[flashjazzcat]

Thats the cf adapter I purchased for mine, I have not tried it yet, Has any one found a nice case to put these in yet.

Shame the UK shipping on that is $35.

[BillC]

Thats the cf adapter I purchased for mine, I have not tried it yet, Has any one found a nice case to put these in yet.

Shame the UK shipping on that is $35.

Here's basically the same adapter for $11.99 BIN + only $2 Economy Int'l Shipping, 99.8% Positive feedback

eBay Auction -- Item Number: 220784885513

 

For those who want an inexpensive dual-CF adapter, here's one for $1.00 BIN + $2.58 Economy Int'l Shipping, 99% Positive feedback

eBay Auction -- Item Number: 370529628891