Funny CF reader behaviour: MCDISK-G-SUB with 3000

I finally spotted an external SCSI PCMCIA reader.

Box only has a "CSM" sticker on it. I opened the case and found out the reader inside is a single slot MCDISK-G-SUB manufactured by MPL (www.mpl.ch). Label also reports MED-10097-001 Rev. D.

I connected the drive to my 3000, set the correct SCSI ID and termination. Popped up a 1GB CF card with a PCMCIA adapter and fired it up. The MPC seems to likes it, DISK-4 shows an active drive. I went on with formatting the card, took a fairly long time, over 20 minutes. 26 partitions of 23345K each are created. I can select them in DISK-0.

So far, so good. Now here's the nasty bit.

I cannot save samples to disk. When I try to save a sample I get this message: There is no disk in the SCSI hard disk drive. Please insert a disk. [FF80]. Red light on the card reader blinks, confirming it's a SCSI error we're talking about here.

While the message shows saving was unsuccessful, when I go to DISK-7 (Load Menu) the file I tried to save earlier is shown in the very same partition I previously selected, dimension shows 0K (sometimes 1K, don't know the logic behind that).

If I try to load the following message reads: This file cannot be loaded since it contains invalid data. Try loading a different file. [F500:306/28/00]. Sometime I'll get a different error: Sorry, this file (or possibly the entire disk) contains damaged data. For safety, transfer any other good files on this disk to another disk and stop using this disk.

Now, any file "saved" this way can be renamed and erased fine.

The most confusing aspect is I can successfully save, load, rename and erase sequences. This really puzzles me. Perhaps it's got to do with the dimensions of the file being saved? I'm clueless.

In a desperate tentative to get it to work I messed with the jumpers and tried the following configuration:

- Parity ON
- Paritiy OFF
- Full CIS mode
- Suppress CIS Level 1 writing
- No CIS formatting by default
- No CIS recognition

None worked, I either got the same symptoms described above or the drive isn't recognized at all.

At this point I don't know if it's worth to keep on fighting with this drive or I should just put it in the bin.
I thought I should post this story, hopefully some of the SCSI experts on this board can help me figure out what's the problem with this unit. If all else fails this will add up to the MPC/SCSI Knowledge Base in the "not working with 3000" chapter.

Update:

Tried to format the CF card with MPC Editor under Windows XP and transfered some .snd files to it. These samples will load correctly in the MPC.

I guess this promotes the MCDISK from the "Not Working" to the "Read Only" league.

Mandino wrote:I finally spotted an external SCSI PCMCIA reader. .

Yaaaaayyyy !!!

Mandino wrote:Box only has a "CSM" sticker on it. I opened the case and found out the reader inside is a single slot MCDISK-G-SUB manufactured by MPL (www.mpl.ch). Label also reports MED-10097-001 Rev. D..

These are a cracker drive from the industrial security world.

Built like tanks.

Nice ciruit boards, quality parts.

A reason they cost $$$$.00 ...

They were designed for harsh condition data logging and other smilar tasks.

Mandino wrote:I connected the drive to my 3000, set the correct SCSI ID and termination. Popped up a 1GB CF card with a PCMCIA adapter and fired it up. The MPC seems to likes it, DISK-4 shows an active drive. I went on with formatting the card, took a fairly long time, over 20 minutes. 26 partitions of 23345K each are created. I can select them in DISK-0...

The MPL drive accepts the SCSI format unit command and, wait for it ... unlike a Microtech PCD-25; actually performs a low level format and block check of a FLASH card.

Mandino wrote:So far, so good. Now here's the nasty bit.

I cannot save samples to disk. When I try to save a sample I get this message: There is no disk in the SCSI hard disk drive. Please insert a disk. [FF80]. Red light on the card reader blinks, confirming it's a SCSI error we're talking about here..

The MPL has crashed in response to being unable to process the large chunk of data being sent by the MPC.

MPL drives in my experience behave a bit odd when working with certain devices.

Here goes...

The MPC3000 uses SCSI-1 commands.

SCSI-1 read/write commands do not have the provision of allowing a "cached mode" of data transfer to be initiated. ( to tell the target drive that is should be caching data for the read/write process )

The MPL accepts SCSI-1 and/or SCSI-2 commands.

SCSI-2 read/write commands allow setting of a "cached mode" of data transfer in the event that a target device allows such. ( the MPL does in this regard )

By default the MPL "seems" to operate in an uncached data transfer mode when receiving/processing SCSI-1 read/write commands.

In this instance the CF media card is responsible for the caching and not the actual MPL.

CF media cards are not all made equal and most can not support large transfers in one go unless aided by a drive that is caching transfer in a productive way.

This thread for some more info:

http://www.mpc-forums.com/viewtopic.php?t=75557

Mandino wrote:While the message shows saving was unsuccessful, when I go to DISK-7 (Load Menu) the file I tried to save earlier is shown in the very same partition I previously selected, dimension shows 0K (sometimes 1K, don't know the logic behind that). .

Only part of the data got saved before the MPL spat the dummy in response to its limitation.

Mandino wrote:If I try to load the following message reads: This file cannot be loaded since it contains invalid data. Try loading a different file. [F500:306/28/00]. .

Correct ... as not all of the file was saved then certain structures on disk will not be present to allow reloading of said data.

Mandino wrote:Sometime I'll get a different error: Sorry, this file (or possibly the entire disk) contains damaged data. For safety, transfer any other good files on this disk to another disk and stop using this disk..

Not surprising. If the MPL bombed when writing part of the file allocation table this would be an additional side effect.

Mandino wrote:Now, any file "saved" this way can be renamed and erased fine. .

The chunks of data in this regard to be transferred during reading/writing processes associated with re-naming a file are small enough that this MPL drive will behave.

Mandino wrote:The most confusing aspect is I can successfully save, load, rename and erase sequences. This really puzzles me. Perhaps it's got to do with the dimensions of the file being saved? I'm clueless..

You're on the right track here.

In this instance also ...... The chunks of data to be transferred during reading/writing processes associated with loading/saving these smaller files fit within the MPL drives transfer limitations.

Mandino wrote:In a desperate tentative to get it to work I messed with the jumpers and tried the following configuration:.

- Parity ON ...... MPC3000 should have this set on where possible.

- Full CIS mode ...... This would need to be on so that the MPL would correctly read the CF card "Card Information Structure Blocks" and ID cards in a correct manner.

Mandino wrote:None worked, I either got the same symptoms described above or the drive isn't recognized at all..

Take care with jumper settings on an MPL. In certain situations with particular jumper configurations, CF cards can be destroyed.

Similarly, such configurations allow you to re-build faulty CF cards. ( where allowable )

Mandino wrote:At this point I don't know if it's worth to keep on fighting with this drive or I should just put it in the bin..

Keep it... Perhaps.

Mandino wrote:
I thought I should post this story, hopefully some of the SCSI experts on this board can help me figure out what's the problem with this unit. If all else fails this will add up to the MPC/SCSI Knowledge Base in the "not working with 3000" chapter.

I have talked about this before with certain drives. ( a mention of MPL models included )

http://www.mpc-forums.com/viewtopic.php?t=70030

http://www.mpc-forums.com/viewtopic.php?t=72102

http://www.mpc-forums.com/viewtopic.php?t=68134

Have a look at this image:



Within the MPC3000 Vailixi 3.50 OS.

Setting the "SCSI Multi Sector Transfer Mode" parameter to OFF allows the MPL drive and a few others to work. ( slowly but reliably )

Hope this helps.

Rohan.

Thanks for a brilliant reply Rohan. Some extremely interesting points there.

There are some aspects I found in the technical documentation I'd like to discuss with you (direct vs. transparent disk access for instance), I'll post them as soon as I'm back home.

I have another unit coming soon (an internal Altec iDrive v1.1 rev1.2), I might post a quick report of how it copes with the 3000 too, so we can keep the MPC card reader KB up to date.

Mandino wrote:Thanks for a brilliant reply Rohan. Some extremely interesting points there.

No problem at all.

Mandino wrote:There are some aspects I found in the technical documentation I'd like to discuss with you (direct vs. transparent disk access for instance), I'll post them as soon as I'm back home.

Sure ... go for it.

Mandino wrote:I have another unit coming soon (an internal Altec iDrive v1.1 rev1.2), I might post a quick report of how it copes with the 3000 too, so we can keep the MPC card reader KB up to date.

I am sure we would all like to hear about the results with this unit. So maybe in public on the forum first. ( just a thought )

Rohan.

Hello Rohan,

As I mentioned in my earlier post this unit can access data in "transparent" or "direct" mode. At first I suspected this could fix the 1K write limit condition; manual reports:

"In transparent mode, the MCDISK makes a lot of extra accesses
to the card to analyse the CIS, find out the card size etc. These
accesses might interfere with the user's intention. Therefore
the MCDISK provides a direct mode of operation. In direct
mode, only those accesses specified by the user through
SCSI READ and WRITE commands are performed. Every
byte of the entire PCMCIA 64M common memory and 32M
attribute memory address ranges can be accessed without
restriction in memory mode (OE and WE signals) or in I/O
mode (IORD and IOWR signals)."

Now, I'm not a tech guy and I could (and would) completely miss the point of this paragraph. Furthermore, my understanding is transparent/direct mode can only be set through a driver which has to to be specifically written for the intended use - I suspect this cannot be done with the MPC3000. I'll just forget about it for now.

I was also intrigued by your comment on CF card not being all equal. Does that mean I could spot a CF card that will read/write fine with this drive and the MPC? Are there any leads to follow in choosing one? What about PC ATA drives? I spotted some cheap Sandisk 1.2GB PC Card (part no. SDP3B-1280-101-50B-176185I) I could get hold of for testing, would that be worth the time/money?

Hope to read your comments soon.

Altec i-Drive arrived today. Have to say this has to be the sturdiest piece of hardware I even came across. Full metal case, I can literally walk over it. Unfortunately the same behaviour showed again: read fine, can't write. Shame.