hiroshima wrote:

I also managed to find a BIOS/FW upgrade package v1.17 for the controller but I don't have an EPROM programmer to try it.

Hi, if you have a socket 7 or slot 1 (or even socket 370 or socket A) motherboard, couldn't the hotswap trick work in order to program a new eprom ? All you need is a motherboard that uses the same kind of rom chip (even if it has more pins than your controller card) and normally you should be able to flash it using a software called Uniflash.
I fixed my bricked my pentium 3 using a slot 1 motherboard and I also flashed a boot rom for my 3com network card and it works flawlessly !

Deksor wrote:

hiroshima wrote:

I also managed to find a BIOS/FW upgrade package v1.17 for the controller but I don't have an EPROM programmer to try it.

Hi, if you have a socket 7 or slot 1 (or even socket 370 or socket A) motherboard, couldn't the hotswap trick work in order to program a new eprom ? All you need is a motherboard that uses the same kind of rom chip (even if it has more pins than your controller card) and normally you should be able to flash it using a software called Uniflash.
I fixed my bricked my pentium 3 using a slot 1 motherboard and I also flashed a boot rom for my 3com network card and it works flawlessly !

Hi,

The hotswap method is too risky than hardware eeprom programmer like a Genius G540.

Moreover, I am not absolutely certain that by updating the BIOS / FW I'll be able to solve the problem because I'm not 100% sure the problem regarding the size partition resides in current BIOS / FW limits.

Tekram user manual isn't too clear to me and online documentations are sometimes cloudy.

This is why I'm asking some advice 🤣 😁

M.

Deksor wrote:

hiroshima wrote:

I also managed to find a BIOS/FW upgrade package v1.17 for the controller but I don't have an EPROM programmer to try it.

Hi, if you have a socket 7 or slot 1 (or even socket 370 or socket A) motherboard, couldn't the hotswap trick work in order to program a new eprom ? All you need is a motherboard that uses the same kind of rom chip (even if it has more pins than your controller card) and normally you should be able to flash it using a software called Uniflash.
I fixed my bricked my pentium 3 using a slot 1 motherboard and I also flashed a boot rom for my 3com network card and it works flawlessly !

EPROM isn't the same thing as flash EEPROM. The old stuff was only erasable by UV light and was programmed differently. The one time I had to have one of these chips burned, I got the help of a nice fellow Vogon here.

I really don't know about this SCSI stuff. It's meant to have fewer compatibility issues with big drives than IDE, so there's probably something going wrong.

Edit: Found something in the manual. DOS 5 Support Enable/Disable: "It is required to enable this option in order to work with drive capacity
over 1GB under DOS environment."

firage wrote:

EPROM isn't the same thing as flash EEPROM. The old stuff was only erasable by UV light and was programmed differently. The one […]
Show full quote

Deksor wrote:

hiroshima wrote:

I also managed to find a BIOS/FW upgrade package v1.17 for the controller but I don't have an EPROM programmer to try it.

Hi, if you have a socket 7 or slot 1 (or even socket 370 or socket A) motherboard, couldn't the hotswap trick work in order to program a new eprom ? All you need is a motherboard that uses the same kind of rom chip (even if it has more pins than your controller card) and normally you should be able to flash it using a software called Uniflash.
I fixed my bricked my pentium 3 using a slot 1 motherboard and I also flashed a boot rom for my 3com network card and it works flawlessly !

EPROM isn't the same thing as flash EEPROM. The old stuff was only erasable by UV light and was programmed differently. The one time I had to have one of these chips burned, I got the help of a nice fellow Vogon here.

I really don't know about this SCSI stuff. It's meant to have fewer compatibility issues with big drives than IDE, so there's probably something going wrong.

Edit: Found something in the manual. DOS 5 Support Enable/Disable: "It is required to enable this option in order to work with drive capacity
over 1GB under DOS environment."

Hi,

This option is already enabled as I had to change controller BIOS address.

M.

I never had any issue with the hotswap technique. (But I understand that you don't want to take the risk). I also know that you can't flash UV-roms with a motherboard. However doesn't the UV rom have the same pinout as flash eeprom ? I won't take the risk myself to erase the original rom if something goes wrong with that, however I would try to flash another rom chip (harvested from a dead mobo or bought on ebay) and put it in the motherboard

They're not even the same form factor, as the flash ROM's in motherboards are 32-pin and those UV EPROM's are 28-pin.

Well the hard drive geometry should change when you select the max DOS drive size.

I think based on the geometry the whole disk is accessible when you select 2GB mode, because it should have 1003 cyls, 128 heads and 32 sectors. Not so when you use the default, 2006*64*32, because through BIOS you can only use 1024*64*32 which is exactly 1GB.

If you use the 3GB or 4.5GB options, you'll start to lose addressing of the last sectors and there is few megabytes less that can be accessed.

Are you sure you have:
-first deleted all partitions so the drive is empty (except there is the MBR, but it should be empty)
-reboot and change the MAX DOS setting to something
-use "FDISK /MBR" just to be sure you write a fresh MBR
-finally run FDISK normally to see if you can now create with new parameters

Jepael wrote:

Well the hard drive geometry should change when you select the max DOS drive size. […]
Show full quote

Well the hard drive geometry should change when you select the max DOS drive size.

I think based on the geometry the whole disk is accessible when you select 2GB mode, because it should have 1003 cyls, 128 heads and 32 sectors. Not so when you use the default, 2006*64*32, because through BIOS you can only use 1024*64*32 which is exactly 1GB.

If you use the 3GB or 4.5GB options, you'll start to lose addressing of the last sectors and there is few megabytes less that can be accessed.

Are you sure you have:
-first deleted all partitions so the drive is empty (except there is the MBR, but it should be empty)
-reboot and change the MAX DOS setting to something
-use "FDISK /MBR" just to be sure you write a fresh MBR
-finally run FDISK normally to see if you can now create with new parameters

Hi,

Already tried it with no luck except I have used LOW LEVEL FORMAT utility inside controller firmware.

TBH I already tried every single possibility... I really don't know what I can do.

M.

Just for the record, I even tried to boot up the machine with a custom boot floppy containing tekram ASPI drivers (aspibmv.sys and diskdrv.sys) and, using TFDISK (FDISK utility from Tekram) I can actually make 1 primary partition of 2Gb in size. This with the HDD controlled by BIOS Int 13h and NOT device drivers.
BUT, after formatting the partition, the MS-DOS 6.22 install process through MS own setup, after a while stops giving WRITING ERRORS.
I've tried the same procedure but this time with the HDD controlled by the device drivers* (aspibmv.sys and diskdrv.sys); I've successfully partitioned the HDD with TFDISK creating a 2GB primary partition and then formatted it. I created a custom floppy containing install files from 1st MS-DOS 6.22 floppy and tekram ASPI drivers but when I launch the install with the modded 1st floppy, setup process will stop reclaiming special drivers for the HDD (or something like this....)

* From the user manual:

BIOS INT 13H device: HD/HDRM/No. HD (default) option is for fixed disk drives controlled by BIOS INT 13H and will be ignored if […]
Show full quote

BIOS INT 13H device: HD/HDRM/No. HD (default) option is for fixed
disk drives controlled by BIOS INT 13H and will be ignored if no device is
installed. HDRM option is for fixed disk and removable disk drives controlled
by BIOS INT 13H. The No option is for drives controlled by the DC-
880B’s device driver instead of INT 13H. Note that removable disk drives
with the HDRM option selected can only function as fixed disks and are not
allowed to change media. Also, if you set the option as HD for a removable
media device, it can only be recognized by adding a device driver instead of
being controlled through BIOS INT 13H.

Again, sorry for my bad english, I hope you can understand what I'm trying to explain 😁

M.

I would not usually do the low level format, unless the sector size needs to be changed (it's 512 byte sectors, right? Are there any settings for the low level format?).

But for using the TFDISK to partition the hard drive, did you specify "TEKRAM" translation method, because the default settings use wrong translation for empty drive and default also will use whatever was on MBR if there was partitions defined so it would still select wrong settings.

OK, so few more ideas:
-SCSI card port 330h (though this is not required when DOS 5 support is enabled, and your MIDI may be at 330h already...)
-SCSI drive ID is #0 (though this is not required when DOS 5 support is enabled)
-SCSI controller ID is #7 (should be)
-cacheable device off so that format is not converted to read verify
-device type is set to HD, not HDRM or anything else
-erase all partitions again with FDISK and reboot
-verify DOS 5 support is enabled
-verify MAX DOS setting is 2GB

Then boot with a floppy, without any SCSI/ASPI driver, and use DOS MSD command or some other tool to verify the BIOS parameters for the drive, so that the next time you start FDISK and DOS format you know if you are partitioning with correct translation.

Well, it could also be that something is buggy and leaves you with translations that give only 1GB drive. Should be plenty for a DX4/120.

Jepael wrote:

I would not usually do the low level format, unless the sector size needs to be changed (it's 512 byte sectors, right? Are there […]
Show full quote

I would not usually do the low level format, unless the sector size needs to be changed (it's 512 byte sectors, right? Are there any settings for the low level format?). I can confirm this, 512byte/sectors

But for using the TFDISK to partition the hard drive, did you specify "TEKRAM" translation method, because the default settings use wrong translation for empty drive and default also will use whatever was on MBR if there was partitions defined so it would still select wrong settings. Used Tekram method of course

OK, so few more ideas:
-SCSI card port 330h (though this is not required when DOS 5 support is enabled, and your MIDI may be at 330h already...) Controller has a I/O port addr @ 130h. 330h is already used by MPU-401AT
-SCSI drive ID is #0 (though this is not required when DOS 5 support is enabled) Check
-SCSI controller ID is #7 (should be) Check
-cacheable device off so that format is not converted to read verify Already tried, no go
-device type is set to HD, not HDRM or anything else Check
-erase all partitions again with FDISK and reboot Check
-verify DOS 5 support is enabled Check
-verify MAX DOS setting is 2GB Check

Then boot with a floppy, without any SCSI/ASPI driver, and use DOS MSD command or some other tool to verify the BIOS parameters for the drive, so that the next time you start FDISK and DOS format you know if you are partitioning with correct translation.
Everything I try, both FDISK and MSD report a geometry of 1023c 64h 32c * 512

Well, it could also be that something is buggy and leaves you with translations that give only 1GB drive. Should be plenty for a DX4/120.

My guess is that HDD geometry translation (MAX DOS disk capacity in controller BIOS settings...) is only valid when the HDD is controlled by device drivers (aspibmv.sys and diskdrv.sys, BIOS Int 13h setting to NO) and using TFDISK with Tekram method to create a 2gb primary partition. If this is true, HDD is unusable as a MS-DOS installation, neither with floppy disks, nor using xcopy command to restore my backup. In fact both installation method fails requesting special drivers or reporting WRITING ERRORS.

So what remains is:

1. Give up. 1Gb space is enough for a MS-DOS machine;
2. BIOS/Firmware upgrade, hoping that updated one supports large HDD/LBA;
3. Going full IDE with a cache controller which supports large HDD/LBA such as Tekram DC-680C/CD
4. Using a DDO which eats at least 10kb of conventional memory. With my pc config, I managed to obtain 623kb of conventional memory using EMM386 for EMS.

@Jepael

This is what MSD reports:

I can't understand what those HDD geometry parameters at line nr. 8 are referring to.

Well I thought the 2G modes etc are available directly from INT13H interface without any drivers.

As what the manual says, it seems you need to load the ASPI drivers before you can partition it with TFDISK while setting correct geometry translation, using "special" mode, not auto, not Mylex, not Adaptec. After partitioning with TFDISK, reboot is necessary before the new translation is automatically detected, so if you did not reboot then old parameres were used and that's why install failed. After reboot, you can verify with MSD if parameters are correct. If yes, then normal format using DOS format should be possible.

The weird parameters you are seeing are what is relevant for IDE drives in BIOS, and they are obviously bogus because type is 0 means not installed.

Jepael wrote:

Well I thought the 2G modes etc are available directly from INT13H interface without any drivers.

As what the manual says, it seems you need to load the ASPI drivers before you can partition it with TFDISK while setting correct geometry translation, using "special" mode, not auto, not Mylex, not Adaptec. After partitioning with TFDISK, reboot is necessary before the new translation is automatically detected, so if you did not reboot then old parameres were used and that's why install failed. After reboot, you can verify with MSD if parameters are correct. If yes, then normal format using DOS format should be possible.

The weird parameters you are seeing are what is relevant for IDE drives in BIOS, and they are obviously bogus because type is 0 means not installed.

Hi Jepael,

I refer to these weird parameters:

CMOS Fixed Disk Parameters
247883 Cylinders, 30 Heads, 60 Sectors/Tracks

The parameters reported by MSD that refer to the C: drive

Fixed Disk, CMOS Type 0
1023 Cylinders, 64 Heads, 512 Bytes/Sectors, 32 Sectors/Tracks

are actually in use for the partition created with FDISK.

I don't know if it is a viable option for you but, what about an Adaptec SCSI card?

The ones that I'm using in my 386 systems (AHA-1540CP and AHA-1520B) are not that fancy (no caching, no VL-bus, just plain ISA SCSI) but work like a charm. I especially suggest 1540 over 1520 since it recognizes everything without a driver (1520 requires ASPI drivers to be installed).

To say the truth, I have bad experiences with the both Tekram SCSI cards I ever used in the past (DC-390F if I remember correctly).

Don't have much to contribute to your problem, other than disk cache controllers can be a pain to work with! I've tested some of the IDE VLB controllers and they are just as fincky to get working correctly in a 486.

vetz wrote:

Don't have much to contribute to your problem, other than disk cache controllers can be a pain to work with! I've tested some of the IDE VLB controllers and they are just as fincky to get working correctly in a 486.

Hi,

I don't have much trouble with the controller.

The only one is that it just doesn't support partitions and/or HDDs greater then 1GB.

I also tried the DDO way but it recognizes the hard drive with a capacity of just 1GB, so the limit of 1023 cyl is still there.

I can create a 2GB partition only when the hard drive is handled via controller's driver; in this case I can't boot from it because the motherboard doesn't support SCSI boot in BIOS sequence.

Last chance is the BIOS/firmware upgrade.

If there is someone capable of flashing some EPROM, would be great.

Mauro

hiroshima wrote:

I don't have much trouble with the controller. […]
Show full quote

vetz wrote:

Don't have much to contribute to your problem, other than disk cache controllers can be a pain to work with! I've tested some of the IDE VLB controllers and they are just as fincky to get working correctly in a 486.

I don't have much trouble with the controller.

The only one is that it just doesn't support partitions and/or HDDs greater then 1GB.

I also tried the DDO way but it recognizes the hard drive with a capacity of just 1GB, so the limit of 1023 cyl is still there.

I can create a 2GB partition only when the hard drive is handled via controller's driver; in this case I can't boot from it because the motherboard doesn't support SCSI boot in BIOS sequence.

The motherboard doesn't need to support SCSI boot in its BIOS to be able to boot from a SCSI drive. However, it's possible you can only use SCSI drives and no IDE drives if you want to boot from SCSI, and your SCSI card needs to have its own BIOS chip enabled (if you have a card without BIOS you won't be able to boot from SCSI, ever). You'll possibly need to disable the IDE channels. Of course, the SCSI card's hard disk size limitations will also apply. If your card has its own BIOS, be sure to check whether there's a function to enable large disk support...

Motherboards that support SCSI boot usually have either on-board SCSI connectors, or need special SCSI controllers that only work together with specific motherboards that have a special on-board SCSI BIOS.

The aforementioned Adaptec SCSI controllers have an 8Gb limit, by the way.