Early IVS Trumpcard 2000

[Amiga87]

I've got a IVS Trumpcard 2000 which I've had sitting around for a while and thought I might give it a go in my spare completely stock A2000, OCS chipset with KS2.1.

The Trumpcard is one of the very early versions with the DMA jumper and the INT2/INT6 select jumper. There is no 1.2/1.3 boot select jumper as found on some newer versions of the card.

So far I've managed to get the card working with a SCSI2SD using a floppy boot disk. For future reference if someone is reading this, disable the low level format in the Trumpcard utility which is on by default - that would hang every time.

The card I have came with no bootrom installed on the card. I would really like to get this to auto-boot but I've had no luck so far. The only ROM I have been able to find is version 4.9. This just causes the machine constantly reset without starting, can't even get to the KS2.1 boot options screen.

Have also tried a link on the DMA jumper - no effect. Selecting either INT2 or INT6 causes the machine to hang whether there is a boot rom installed or not. Manuals for the later version of cards don't mention either of those jumpers.

So I think I need an early ROM version for the Trumpcard which i haven't been able to find. Would greatly appreciate it if someone could help me out.

[Toni Wilen]

Did you try Trumpcard Pro v4.9 or Trumpcard v4.9 ROM? (both are included in my expansion rom package, should be in eab file server somewhere)

There is chance non-pro variant will work.

Unfortunately I haven't found any older ROM versions but some Trumpcard driver disks include *.driver files that can be easily converted to official ROM images.

[Amiga87]

Hi Tony,

Thanks for the response.

I've only tried the non-pro variant of the 4.9 ROM as it is a very early version of the card.

After you mentioned the driver disk I managed to find a boot floppy that came with the machine. No .driver files anywhere but in the devs directory there is a IVS_SCSI.device with the following text early in the file: "IVS_SCSI.device expansion.library.SCSI 1.3 (14 Aug 1988)"

Is the .device file of any use for conversion to ROM?

[Toni Wilen]

Normal device drivers (.device) can't be converted to autoboot ROM, they almost always don't include autoboot code.

Could you attach/upload the install disk, photo of the board (Must be high resolution enough to see all chip markings) and output of this program: http://www.winuae.net/files/b/expdump.zip ? (card installed, no HD needed. Run the program twice, boot jumper set and not set)

This is quite old card, worth more detailed examination

[mark_k]

Are you able to try the install disk from amiga.resource.cx? That has the V4.12 driver (including .driver file which you could make a boot ROM from).

Does your board look different to the pics on that page? Maybe the DMA jumper is related to this text in the 1988 advert?
"EXCLUSIVE IVS HYPERDISK PORT CONNECTS TO IVS WARPSPEED DMA CONTROLLER CARD TO SPEED DISK I/O UP TO 8X!"

No later ads mention that feature.

[Amiga87]

Toni,

Please find image of card here...

https://drive.google.com/file/d/1ymFZNrNnzrdRTyeQeD9imMvXfnB4uZy6/view

Results from expdump:

BoardAddr : 00e90000
BoardSize : 00010000
er_Type : d1
er_Product : 30
er_Flags : 00
er_Reserved03 : 00
er_Manufacturer : 0840
er_SerialNumber : 00000000
er_InitDiagVec : 8000
{0xd1,0x30,0x00,0x00,0x08,0x40,0x00,0x00,0x00,0x00,0x80,0x00}

As you can see from the image of the card there is no boot select jumper.

Mark, yes the driver from amiga.resource.cx has the same date as the rev 4.9 rom that is available.

From a quick look at the board (i will have a better look tonight), the ROM socket does not seem to have A13 on pin 26, this suggests that it is a 8kB device, where as the 4.9 rom is 16kB.

Edit:

Have had a better look. A13 was a NC on the ROM socket. Early address lines on the ROM went through a '245 3 state buffer. Later address lines were wired direct to the zorro bus, so connected A13 on ROM to zorro BA14 (for reference ROM A12 is BA13 on the card). Unfortunately still no joy. It's still going through a restart cycle and guruing when ROM rev 4.9 is installed.

[Toni Wilen]

Thanks. Quite different layout.. Autoconfig data is identical to later variants.

One test would be to burn 16k (or larger) EPROM that contains some simple pattern (or even a text file!). Then check with any monitor program where in address space the ROM gets mapped.

Last 32k of address space (e98000-e9ffff) should be filled with ROM data. ROM data in even bytes, odd bytes either contain random garbage or only zeros or ones.

Most likely one of the PAL chips are part of address decoding and only maps the ROM in address space when first 16k is accessed.

Perhaps "DMA" jumper hardware address space was reserved for last 16k?

In emulation I also get a guru if I only load 8k of ROM. But it does not prove much because almost any issue can cause similar crash.

EDIT: Make sure first few bytes of test ROM image contain FF to prevent AmigaOS detecting it as normal boot ROM.

[Amiga87]

Thank you Toni. This little card has me intrigued. I suspect the DMA jumper is doing as you speculate. I will burn a test ROM tomorrow and report back with results. May just be a case of some simple logic to address 16k of address space on the rom. We shall see...

[mark_k]

I think the "DMA" pins are likely to be for the drive activity LED; that's why they are labelled - and +. My guess is the + pin connects to the 53C80 DRQ pin.

This page on stone-oakvalley-studios.com has info about various IVS products, and a page with ADF software/driver downloads. Does the ivs_scsi_wb13.adf disk work with your card? I'll download that and have a look at the driver code.

Edit to add: In the devs directory of that disk is IVS_SCSI.driver. Text in that says SCSI 1.3 (14 Aug 1988). You can build a boot ROM image from that .driver file.

[Toni Wilen]

Nice find, 8k ROM. Perhaps it works (but possibly it has other unexpected bugs..) Extracted ROM image uploaded to the zone.

In emulation it executes SCSI TEST_UNIT_READY command before hanging but it probably has some hardware differences that are not emulated.

[Amiga87]

Thanks gentlemen. Good find Mark.

I will burn 2 roms today, 1 with test pattern and 2nd with Toni's extracted ROM image and will report back tonight.

While looking at how they were doing the address decoding last night, noticed that pin 14 on U503 (74LS245) which is one of the address lines, had considerable corrosion (don't know how i missed that before, you can even see it in the photo above), seemed to be working fine on a scope but will replace that IC as well.

[mark_k]

Looking at the v1.3 .driver code (btw that load file has a symbol hunk...), it doesn't seem to add a level 2 interrupt server, unlike the later driver. So it must e.g. poll an IRQ register bit instead of expecting interrupts to be caused. And may well hang if there are unexpected interrupts???

Amiga87, was there a jumper on the INT2 position when you got the card?

The v1.3 driver does (seem to) support loading a later driver version from the RDB area. Presumably the IVS install software is able to write that to the RDB.

There's a PDF user manual for several Trumpcard products on the web page I linked above. It doesn't seem to cover the early board version though.

[Amiga87]

No links were on the card when I got it, but neither was there a boot rom.

Yes, I suspect that i will need to rebuild the drive (another uSD card) with early version of the tools.

[Amiga87]

Update:

- Replace U503 to be sure that we're not fighting a problem there. Board tested ok booting from floppy.
- yes Mark, the DMA jumper is in fact a HDD indicator as you speculated (thought i tried that before and din't see anything, anyway works now)
- Tried with v4.9 rom, same as before
- Tried the extracted v1.3 rom, just went into reboot cycle without accessing the scsi bus.
- Put the test rom in, pattern used was 16 bit value from 0x0000 to 0x1fff to cover 16kB address space
- data appeared correctly on monitor from 0xe98000 to 0xe9ffff (in even bytes)

conclusions thus far, this early card doesn't like the late version roms even with the extra address line added to the rom and extracting the rom from the .driver file is not working either.

[Toni Wilen]

I tried 8k rom again in emulation and it does work after few small tweaks. It even supports custom filesystems which is quite nice for 1988 filesystem.

Main difference seems to be how 5380 chip DMA request is handled.

Later trumpcard and trumpcard pro: hardware handles 5380 DMA protocol automatically when fake/pseudo DMA port is read or written to. Single byte transfer requires single IO access. Every fast PIO SCSI controller works this way (some better ones use word or even long word wide port)

Old trumpcard: software handshake. Software polls 5380 status register, when data is available, it reads one byte. Repeat. Every byte transfer requires multiple IO accesses (=very slow). Most very old PIO SCSI controllers worked this way (for example pre-bytesync/wordsync Supradrives).

I can't get both modes work at the same time so something must be missing from emulation if v4.9 on disk driver works with old trumpcard.

Where does DMA jumper pins go? I assume other pin goes to 5380 pin (which one?) but where the other pin go? (gnd/5v?) I still can't believe it is simple activity led connector with really weird name.

[mark_k]

I think later drivers can use two modes, controlled by the setting of the "Handshake Reads and Writes" option (see PDF user manual page 36). The Trumpcard Professional does handshaking in hardware.

Amiga87, can you try with Kickstart 1.3? Reading old magazine articles on the historical archive page, see Amazing_Computing_Vol_07_11_1992_Nov_page92.jpg

That's a letter from IVS which suggests the old boot ROM could have some problem with Kickstart 2.x.

[Amiga87]

Toni, DMA pin (-) goes to Gnd, (+) goes to pin 10 of the controller which is DRQ (DMA request output) and a pull-up. So they are using the dma request output to drive the indicator. Putting a link across the jumper makes no sense as the datasheet indicates that it is an output pin.

Mark, another nice find. I am using a 2.04 and was going to try 1.3 but it was getting late last night. Will give it a try tonight.

Interesting that it says

The 4.x series ROM works in our complete line of peripherals for the amiga...

when in fact the card in question is missing one of the address lines to the rom to actually work with the larger 4.x roms. That was one of the things i confirmed with the test rom last night.

[Amiga87]

Tried:
- ks1.3 with he extracted trumpcard 1.3 ROM, reboot and guru.
- ks3.1 with both 1.3 and 4.9 ROMS. reboot + guru.

May just have to wait to see if someone else shows up here with one of these early cards with a boot rom.

[Toni Wilen]

Originally Posted by mark_k

I think later drivers can use two modes, controlled by the setting of the "Handshake Reads and Writes" option (see PDF user manual page 36). The Trumpcard Professional does handshaking in hardware.

But v4.9 ROM uses hardware handshaking when used with non-pro in emulation (and it hangs if DMA transfer is not acknowledged automatically). Something is missing..

"If your drive does require handshaking" also sounds suspicious. Drive won't see the difference between software or hardware handshaking (except more tight timing), handshake is always needed. 5380 does not support synchronous transfers.

[mark_k]

Just a note about driver versions. What I've been calling the v1.3 driver since there's a string "SCSI 1.3 (14 Aug 1988)" inside it, probably isn't actually version 1.3. That string exists in other driver versions, so presumably was left over from an earlier version.

The "real" version should be given after the "IVSv" characters near the beginning of the file; two longwords give major and minor version number. It's 2.2 in this case.


I found a newsgroup posting relating to Kickstart 2.04 use with old Trumpcard.