BFG 9060 conflicts with Amiga 3000

[mlenardon]

Amiga 3000 board with RTC damage cannot boot the BFG without rebooting once to detect SCSI. Setting to delay with SCSI Prefs worked. Board has Amber and RTC damage so I would like to not use this board until it can be fully repaired.

Re-Amiga board with the same chipsets installed has more issues.
SCSI drive detection after first boot corrupts partitions after a restarting or machine locks up mouse or freezes when accessing the hard drive. I am using 3.2.1 roms and a later SCSI WD chip not the prototype.

Is there an easy fix for these accelerators in the works or available?

I am going to borrow a SCSI card to test with and try a Zulu SCSI also to see if it is better.

I have a RC50 Rev 6 CPU installed with a SCSI2SD (smaller version) with 4.82 firmware installed.

[pipper]

Are you running at 50Mhz? 100Mhz and the A3k internal SCSI don’t work together.

[A10001986]

A3000+BFG9060 is a matter of luck as regards SCSI. The SuperDMAC is Commodore's worst chip ever, it has production/timing variances of epic proportions.

My conclusion, as it stands today firmwarewise: The BFG is perfect for the A4000, but not "production ready" for the A3000+SCSI (let alone the AA3000+). At any speed. It may work, it may not work.

I predict that this thread will now be flooded by people telling their success stories

[DarrenHD]

Originally Posted by A10001986

A3000+BFG9060 is a matter of luck as regards SCSI. The SuperDMAC is Commodore's worst chip ever, it has production/timing variances of epic proportions.

My conclusion, as it stands today firmwarewise: The BFG is perfect for the A4000, but not "production ready" for the A3000+SCSI (let alone the AA3000+). At any speed. It may work, it may not work.

I predict that this thread will now be flooded by people telling their success stories

I agree. I got immediate SCSI lockup in my A3000 with the BFG. I wasn't running the very latest firmware (maybe 1 version old), but I didn't want to waste time on it ripping machines apart, risking damaging things, so I just put it in my A4000 and it works fine.

[trixster]

I can get one of my BFGs to work ok in my AA3000+ at 50mhz. I had to remove the onboard passive terminators, add an active terminator to the external port, use as short a scsi cable as possible, and enable active termination on the acard 7720uw I’m using.

I’ve tried all the available firmwares to try and get 100mhz working - it will work ok for a bit but eventually the scsi bus will lock up.

Oddly wrangler can get his running at 100 in his aa3000+ without issue.

I haven’t tried a bfg in my 3000D.

[StompinSteve]

DarrenHD wrote "I predict that this thread will now be flooded by people telling their success stories"

No success story here.
I have an A3000D rev 9.1 Motherboard and a Buster 11. The SCSI Chip is a "WD33C93A-PL-00-08" (the latest version).
The Motherboard and daughterboard have been recapped, the 4 PAL chips replaced with GAL's. the whole nine yards.

I have tried the two BFG9060's I have (I normally use one in my A4000T). I have an older BFG card with a Rev 5 full 060 CPU (from the 4000T) and a brand new one with a Rev 6 060. Both at 50Mhz. Both run the latest firmware.

The 4 HDD's are virtualised on a ZuluSCSI RP2040 with the current firmware 1.2.1

Both BFG cards run perfectly fine in my 3000D but the onboard SCSI does not work properly (if at all). It can:
- see no devices at all (this is what happens most of the time)
- see one (1) of the 4 virtual HDD's and which one is random. So it can see DH2, DH0, DH1, DH3, again, totally random which one.
- If it happens to see DH0 that time, it will try to boot, but will throw random read errors at me like there is no tomorrow. Completely and utterly unusable.

I always run with an active terminator (GGlabs) on the external port. Motherboard terminator-packs are not installed.
My D800 too was soldered on backwards at the factory but this was fixed and the D800 replaced with a more modern part. It works absolutely fine and fast and stable. External SCSI Zip drive also perfectly fine (without BFG).

I can not emphasise enough, that there is nothing wrong with the onboard SCSI controller and ZuluSCSI solution.

Disconnecting the Zulu, thus running "floppy only" and both BFG cards are perfectly happy. No problems at all. Both cards are known to be good, because they work perfectly in my A4000T with its onboard SCSI.

So I too am a victim of the BFG having massive issues with the 3000D's onboard SCSI. If I remove the BFG and run with the stock 25MHz 030, the onboard SCSI controller and the Zulu are happy campers. No problems at all.

We are not alone. Chris Edwards also mentions the "BFG + 3000D SCSI Problem" in the comments of one of his video too.

I hope that someday, the issue can be fixed with a BFG Firmware. Until then, no BFG in my A3000.

[Channel-Z]

You get the same SCSI issues if you run 3660 @66MHz on A3000D.

[thebajaguy]

BFG-9060/50MHz in 2 different A3000D's - one with 60ns ZIP SCRAM, one with the DRAM-replacement PCB (limited run a few years ago) and something like 70ns 4Mx4 DRAM on it w/a GAL to make it work (I also removed the ZIP sockets, soldered it in). Both Ramsey-04, SDMAC-02, Buster-11. OS 3.2 or 3.2.1 ROMs

The SCRAM system has an AmigaNet card in it, and works fine (Roadshow stack). SCSI2SD v5.1. I put the SCSI2SD in the DRAM system to test. All behaves. I have more testing to do, but I have some other priorities to get to at the moment.

My A4000D is having system issues, as the GVP A4040/40 T-REX I (was the most recent card in it), the A3640(060 adapter), and the BFG only intermittently boot from the IDE now, but the WB boot screen comes up just fine, so it's likely past battery or cap issues coming back to haunt.

My BFG card was built by Acill, and he tested it to 100MHz, but I have yet to bump mine (have a Rev 6) to 100MHz. I'm waiting for a cooler sink/fan module to come in before I try. I have yet to give it a test drive in the A4000T where the TekMagic 68060/60 (ultrasound) card is - that is my super stable system, and likely to stay that way long term since the TekMagic accelerator is unique for that computer.

[fgh]

This might not be related, but I just had a scsi related experience that didn’t seem to make sense.
I moved my CS MK3 060 to a 4000D, replacing a warp engine 040.
I had MMUlib installed, and was of the (mis-)understanding the existing installation would work also with 060 (it had mmulib’s 68060.library), and it seemed to work until I tried using the MK3’s SCSI.

Partitions on any MK3 connected device would show up in early boot menu first time but then immediately be uninitialized on first access (even when booting from a device on the IDE)

If I went to hdtoolbox lots of warnings would pop up (from mmulib I guess) about phase mismatch, wrong dma transfer length, etc etc.
I then noticed scout reported my 50mHz 060 as 60mhz, so I threw in a 35mHz crystal just to rule out if the scsi was somehow running too fast. And at 35mHz things started half working. I could see the two partitions again, but have it freeze upon access, etc.

Then I put back the phase5 060 libs and all was back to normal.

So - as a lot of people use mmulib now with 3.1.4-3.2.2, and the BFG’s are probably replacing lots of 030’s or 040’s, and this seemed to work until I started using the scsi, I suppose some people might be in the same situation also with other DMA SCSI’s like the 3000D.

I didn’t try yet but I assume it would work perfectly if I installed MMUlib again with the MK3 attached, and answering yes to ‘Do you have a phase 5 card?’ during install.