Mediator 1200Tx issues

[Vypr]

I recently obtained a new ElBox Mediator 1200Tx Black and am having problems getting it to work. In short, I think I have bought an expensive paperweight.

The board is set up on the following system:

Rev 1d4 board.
MTech 1230/28 accelerator with 8Mb RAM
Voodoo3 2000 PCI graphics card
Kickstart 3.1 ROMs
Amiga OS3.9 with both Boing Bags installed.

My issue is that the board is not being seen by the Amiga at all. It does not appear in early startup or sysinfo. The passthrough board appears to work as the MTech card is detected and the system boots to it OK and the power on from the ATX power button works fine.

I have tested the board on two other Amigas with the same problem. I have also tried all the jumper settings an even booting with the accelerator disconnected. No joy.

Can anyone think of something else I can try before admitting defeat and returning the card?

[Daedalus]

Really only the obvious things - make sure the Mediator is seated fully in the connector on the adaptor board, that all the MACH chips are properly seated, and that nothing is shorting out on the back of the Mediator, e.g. Scandoublers, clock port add-ons etc. It should show up in the early startup without and software support, so it's something pretty fundamental going on with it.

Have you tried it without any PCI cards in it?

[Vypr]

Originally Posted by Daedalus

Really only the obvious things - make sure the Mediator is seated fully in the connector on the adaptor board, that all the MACH chips are properly seated, and that nothing is shorting out on the back of the Mediator, e.g. Scandoublers, clock port add-ons etc. It should show up in the early startup without and software support, so it's something pretty fundamental going on with it.

Have you tried it without any PCI cards in it?

I teste dit on a bare board with no addons apart from the accelerator board.
Pushed down hard to ensure it was securely seated as well as I heard that this can be a problem but the fit si snug, there is less than a hairline gap in the connector on both sides.
MACH chips appear fully seated too.
I did try it unpopulated with PCI cards as well but no joy.

[Vypr]

Got a reply from Elbox regarding this issue-

Hello,
Thank you for the MedReport.txt report.
The problem is related to your turbo card:
RAM (Hardware Design Udo Neuroth): Prod=2192/32($890/$20) (@$200000 8meg Mem)
This card cannot work with any PCI or ZORRO cards, which need memory space accesses, as the memory of this turbo card is assigned as a memory expansion in the motherboard memory adress space: $20000 - $9FFFFF.
The only PCI card, which can work with this turbo is RTL8029 ethernet card as it use I/O accesses only.
In order to make use of all Mediator features you need to change the turbo card to the card, which allow to use more than 8 MB of the memory. cards. Such cards do not use mtherboard memory space for their memory.

However I am calling bullshit on that because I believe it should still see the expansion card in early startup. Also, their manual specifically mentions my accelerator card with no mention of problems with the memory expansion:

[quote]
MEMORY WINDOW
Set 0 MB If you use 8MB of RAM installed in the memory expansion or in the turbo card configuring this memory in the lower boundary (e.g. M-TEC 1230 turbo card).

[Kin Hell]

I understand it's not a PCMCIA friendly card. You'll need an accelerator than can address more than 8Mb.

[Daedalus]

Ah, I never thought of that, but the Mediator does sit in Z-II space along with the accelerator. It is strange that it doesn't show up with the window set to 0MB (which will cripple the Mediator, but it should still work for some functions).

It should show up in the early startup regardless, but I don't know if that accelerator card is playing nice with autoconfig. If not, there's a good chance it's blocking the Mediator altogether. If that was the case however, I would expect it to show up in the early startup menu when the accelerator is removed.

Hmmm, Have you asked them why they list the accelerator in their documentation if it isn't compatible?

[Vypr]

Well, it turns out that if I remove the accelerator card and open the config jumper, the mediator can be seen.
Have just decided to bite the bullet and splash out on an ACA 1233n card which should arrive tomorrow.

[Daedalus]

Ah, at least you've gotten to the bottom of it. You should get a decent price for the M-Tec card if you were to sell it - one went for £117 in February with 4MB, which would offset a decent chunk of the cost of the 1233n...

[Vypr]

OK, so ACA card arrived today. Fitted and now early startup can see three devices.
However the system locks up just after loading Workbench If I set the memory to anything other than 0Mb
medconfig can see the two PCI cards when it's on 0Mb so I have to be missing something!

[Daedalus]

Hmmm, there's probably a conflict in memory mappings so - the ACA cards have a special memory map designed to increase compatibility with old games, but some of this may be mapping to the same Zorro-II address space, causing a crash as soon as it gets used. I don't know the ACA cards well enough, but there are config tools that let you fine tune the card - perhaps there are options there for rearranging the memory map?

Looking at the ACA memory map, I can't see any overlapping space, but there are some options available with the downloadable tools, so maybe you might have some luck with that.

[Vypr]

I did a bit more testing, I installed OS3.1 on a spare SD card and tested that, and that seems to work fine with the accelerator. System booted to Workbench, drivers installed OK and pciinfo detected all the cards. It's only OS3.9 that freezes up after loading Workbench.
I gave up around 1.30am this morning so tonight I'll try it with a clean 3.9 install with no Boing Bags

[Vypr]

OK, so it's definitely an OS3.9 issue. System is fine with 3.1 but a completely clean 3.9 install freezes as soon as Workbench starts to load. I get a grey screen with no icons and the mouse cursor won't move.

[Kin Hell]

Originally Posted by Vypr

OK, so it's definitely an OS3.9 issue. System is fine with 3.1 but a completely clean 3.9 install freezes as soon as Workbench starts to load. I get a grey screen with no icons and the mouse cursor won't move.

Are you installing BB1 & BB2 on the OS3.9 Install?

Technically, you should be which then leads to another important question. Do you run from the "3.1 ROM on the Mobo" or the "Updates installed ROM" that gets kicked into RAM every time your Miggy Cold Boots?

You are asked which to choose during the update.

Another check should be the 3.1 ROM's on your Mobo are genuine C= ones & not Colanto's 3.X effort.

If it's all still questionable in functionality for eliminating these ideas, it might be worthwhile sending another Mediator.txt Report to Elbox now you have your ACA card in there. - If you bought the Tx new from Elbox, it shouldn't need any of it's MACH chips changing.

[Vypr]

Have tried with a fresh OS3.9 with no BBs installed and with both BBs. Same issue seen.
Also tried with and without the ROM update from BB2
The ROMs are from Amigakit (These ones http://amigakit.leamancomputing.com/...products_id=62)

I've already sent the medreport.txt with the ACA card to Elbox with a full description of what is happening and what has been tried. This is the reply back I got:

"Interesting"

I'll note that the mediator works perfectly in OS3.1, got the Voodoo card working and outputting to the monitor with no problems. In OS3.9 if I comment out everything between and including Set SCSIUpdate 1 - Set SCSIUpdate 0 in startup-sequence then the system doesn't crash, but then I can't use a lot of applications,
I tried updating scsi.device to 43.45 but again, no difference, the system still freezes when loading Workbench

[Kin Hell]

....

interesting
ˈɪnt(ə)rɪstɪŋ/Submit
adjective
arousing curiosity or interest; holding or catching the attention.
"an interesting debate"
synonyms:	absorbing, engrossing, fascinating, riveting, gripping, compelling, compulsive, spellbinding, captivating, engaging, enthralling, entrancing, beguiling; More

A lot of folk can't get in touch with them & when we hear someone successfully contacting them, their retort is merely to say, "Interesting"??
Naughty naughty Elbox

On the hardware side:

Has your 1D4 Mobo had timing fixes done?

http://www.ianstedman.co.uk/Amiga/am..._mobo_fix.html

There are some suggestions among the 1D4 mods for Apollo cards that removal of certain resistors may help stability issues with ACA cards too.
Now that you have an ACA card, is the 1D4 board 100% stable?
Maybe try that for a while & go back to the Tx setup after removing the Resistors. Something to do with Clock Cycle Phase Alignment....

On your Startup Sequence Rem event, I'm afraid I am not worthy, sorry fella.
Not taking the pi$$ either,
....but it is "Interesting" that Remming the SetSCSIUpdate stuff out lets it boot to Workbench....

[Daedalus]

Hmmm, is it just the scsi.device stuff that's disabled? If so, perhaps an alternative drive or patch might work, like IDEFix or something along those lines. You can also disable the updated scsi.device by setting the NOSCSIUPDATE environmental variable to anything, rather than disabling that whole chunk of the startup-sequence, but if it works (with IDEFix or other alternative) then stick with it.

[Kin Hell]

@ Daedalus

Doesn't 3.9 do some patch or other to the IDE/SCSI? or does it do it's own IDE99 patch?

I really can't remember.

[Daedalus]

Yes, and it also includes an extra bit of scripting so that the patching can be enabled or disabled using an Env variable. If the scripting is left as standard, it automatically takes IDEFix and some other patches into account instead of its own if it finds them installed.

[Vypr]

Well it appears the scsi section was a red herring. It did still hang from time to itme with that commented out.
I commented out the loadwb command and .... no freeze. so something in the Workbench GUI is causing problems.

[Daedalus]

Loading Workbench also loads the WBStartup files, which could be pushing memory usage into some problem area. There's a shortcut for disabling the WBStartup directory when the machine starts up, but I can't remember it and I'm not in front of my Amiga now. But temporarily renaming WBStartup will disable it too for troubleshooting.

Under 3.1, was it fully stable even if you opened lots of applications, i.e., pushed up the RAM usage?