Inverting the "mode" switches, the computer no longer boots. No post beeps or anything. Putting them back to normal brings it back to life.

Pulling the "0WS" jumper and moving J4 "pen" from 1&2 to 2&3 got it to show up in TesX.

(I moved to another monitor to use my main monitor for another project)

The 2.5MB window now shows up. Changing all of the switches in bank 1 to off makes no difference, and changing all of the switches in bank 1 back, then all of bank 5's switches to off gives this:

Found another one on ebay, can't quite tell what the switch positions are, look different though...
Trying to read..
B1 11001000
B2 11000100
B3 11000000
B4 10111100 not sure on 2nd pos
B5 too unclear
Mode 11011111
Could be reversed if the mark is a notch rather than a pip

Could you link me to this eBay listing? If I can get this one working, I'm probably going to try and pick up the other card as well.

Hehe, you are quite fast for testing 2^48 combinations :-p :-p :-p

Good work, you have it almost running! You now need to move your 2.5 MB window from starting adress 3.875M to 2.0M to fill that unused memory area.

So you already know (as far as I can see on your photo...)

B1 11x111xx 3.875M
B2 11x11xxx 4.375M
B3 11x1x1xx 4.875M
B4 11x1xxxx 5.375M
B5 11xx11xx 5.875M

Dont change all DIPs at once now, only 1 after 1 and note the results. With some mathematics you can already solve the puzzle. I would be lazy and get some more data first...

Look at the AT-Rambank manual (setting of SW2). It gives you an idea how it works. You need to find the corresponding DIPs on your card.
download/file.php?id=53151

I have no explanation for "changing all of the switches in bank 1 back, then all of bank 5's switches to off gives this". Think you selected a not-allowed DIP-combiantion which gives weird results...

It was a great pleasure for me 😉

Too bad these boxes don't come with a manual 😒

My proposal:

RAM Start BIN DIP 3-8
0,500 1000
0,625 1001
0,750 1010
0,875 1011
1,000 1100
1,125 1101
1,250 1110
1,375 1111
1,500 10000
1,625 10001
1,750 10010
1,875 10011
2,000 10100
2,125 10101
2,250 10110
2,375 10111
2,500 11000
2,625 11001
2,750 11010
2,875 11011
3,000 11100
3,125 11101
3,250 11110
3,375 11111
3,500 100000
3,625 100001
3,750 100010
3,875 100011 x111xx
4,000 100100
4,125 100101
4,250 100110
4,375 100111 x11xxx
4,500 101000
4,625 101001
4,750 101010
4,875 101011 x1x1xx
5,000 101100
5,125 101101
5,250 101110
5,375 101111 x1xxxx
5,500 110000
5,625 110001
5,750 110010
5,875 110011 xx11xx
6,000 110100
6,125 110101
6,250 110110
6,375 110111
6,500 111000
6,625 111001
6,750 111010
6,875 111011
7,000 111100
7,125 111101
7,250 111110
7,375 111111

hwh wrote:

Too bad these boxes don't come with a manual 😒

Indeed, it would make it a lot easier.

Predator99 wrote:

Hehe, you are quite fast for testing 2^48 combinations :-p :-p :-p

I changed two jumpers and it decided to show up. Stroke of luck.

Dont change all DIPs at once now, only 1 after 1 and note the results. With some mathematics you can already solve the puzzle. I would be lazy and get some more data first...

Look at the AT-Rambank manual (setting of SW2). It gives you an idea how it works. You need to find the corresponding DIPs on your card.
download/file.php?id=53151

I know how the address switches work, I've been messing with my AST Advantage 128, which is also having troubles. The motherboard has 2MB of RAM onboard. The first 640K are conventional, the next 384K are not showing up at all (even as UMBs... Weird), and the last 1024K are XMS. The AST Advantage 128 used to be bringing it from 512K to 640K and the next 1024K on it were for XMS.

So now, I have 1024K XMS from the motherboard and 1152K XMS from the AST.
The motherboard shows RAM to end at 1664K instead of 2048K like it physically does. Installing the AST at 1664K gives memory conflict. Same at 1920K. Setting it to start at 2048K... It doesn't appear at all. The TestX shows it starts at 2176K, instead of 2048K like it is set to. Is this a misprint in the AST manual?

I have no explanation for "changing all of the switches in bank 1 back, then all of bank 5's switches to off gives this". Think you selected a not-allowed DIP-combination which gives weird results...

I see.

It's probably not showing up as UMBs, because you need a software memory manager to access it. I think something like QRAM might do the trick, or you can run DR-DOS which has EMM286.sys. Alternatively, if you have a 386 or 486 upgrade module you can run EMM386.

FAMICOMASTER wrote:

Predator99 wrote:

Hehe, you are quite fast for testing 2^48 combinations :-p :-p :-p

I changed two jumpers and it decided to show up. Stroke of luck.

Nothing to do with "luck", there is a system behind it. You have difficulties to admit that you were not right with your estimation, correct? 😉 Doesnt matter, happy that the card works.

FAMICOMASTER wrote:

I know how the address switches work, I've been messing with my AST Advantage 128, which is also having troubles. The motherboar […]
Show full quote

Dont change all DIPs at once now, only 1 after 1 and note the results. With some mathematics you can already solve the puzzle. I would be lazy and get some more data first...

Look at the AT-Rambank manual (setting of SW2). It gives you an idea how it works. You need to find the corresponding DIPs on your card.
download/file.php?id=53151

I know how the address switches work, I've been messing with my AST Advantage 128, which is also having troubles. The motherboard has 2MB of RAM onboard. The first 640K are conventional, the next 384K are not showing up at all (even as UMBs... Weird), and the last 1024K are XMS. The AST Advantage 128 used to be bringing it from 512K to 640K and the next 1024K on it were for XMS.

So now, I have 1024K XMS from the motherboard and 1152K XMS from the AST.
The motherboard shows RAM to end at 1664K instead of 2048K like it physically does. Installing the AST at 1664K gives memory conflict. Same at 1920K. Setting it to start at 2048K... It doesn't appear at all. The TestX shows it starts at 2176K, instead of 2048K like it is set to. Is this a misprint in the AST manual?

I have no explanation for "changing all of the switches in bank 1 back, then all of bank 5's switches to off gives this". Think you selected a not-allowed DIP-combination which gives weird results...

I see.

Do not understand yet...did you test the DIP settings above and do they work in the Cheetah?

Anonymous Coward wrote:

It's probably not showing up as UMBs, because you need a software memory manager to access it. I think something like QRAM might do the trick, or you can run DR-DOS which has EMM286.sys. Alternatively, if you have a 386 or 486 upgrade module you can run EMM386.

I don't have a 386 or 486 upgrade, but I have been looking for one if you know where I might get one.

I know I need a memory manager to *use* the UMBs, but they just don't show up in MS Diagnostics. In the areas between the video memory and ROMs and stuff, it's shown as "Open" instead of "RAM"

I don't need an EMM driver, I have an EMS board installed that works just fine. BocaRAM XT with 512K on it. I don't have anything that actually needs EMS, but it's nice to have, I suppose.

Predator99 wrote:

Nothing to do with "luck", there is a system behind it. You have difficulties to admit that you were not right with your estimation, correct?

Everything to do with luck, pal. Picking a switch at random and it happening to work is not common. The guy in the thread you linked also got lucky by just flipping every switch to off.

Do not understand yet...did you test the DIP settings above and do they work in the Cheetah?

No, my motherboard XMS decided to just... Work. I've been re-configuring the AST Advantage to be above the motherboard's 1MB of XMS for now, but it's being troublesome. The first bank is 128K, but it only seems to have starting addresses in blocks of 256K... So having a 128K bank means it's got empty space, unless that first bank is overlapping with the motherboard's XMS. It's kinda annoying, I'm going to have to either get more 41256 DRAMs or I'm going to have to figure something out.

The Cheetah card will come after that - I also have 511000 DRAMs coming in for the motherboard to bring it to 4MB onboard. That way I can have 640K + UMBs + 3MB XMS + the AST's 1MB (maybe 1.5MB later) + the Cheetah's 2.5MB + BocaRAM XT's 512K EMS.
That should be way more than enough RAM for any and all programs that will ever run on a 286. If it's not, I have a fully populated AST RAMpage PC (shown earlier in the thread) with 2MB of EMS to bring it up further.

The reason I don't want to mess with settings for now is because I might end up just filling the gaps between with more memory. 4.5MB of RAM before the Cheetah might be enough to either make it overlap or to make it contiguous.

FAMICOMASTER wrote:

Predator99 wrote:

Nothing to do with "luck", there is a system behind it. You have difficulties to admit that you were not right with your estimation, correct?

Everything to do with luck, pal. Picking a switch at random and it happening to work is not common. The guy in the thread you linked also got lucky by just flipping every switch to off.

Guess who is the author of the thread 😉 Yes, its me :-p I have ...didnt count them ... but maybe 4 of these cards running with that method. These are all cards I own. Not a single one does not work with that method. If its only luck then I should start playing lotto at once 😀

There was no logic in me deciding "Let me move two jumpers at random" and it deciding to show up.

Here, I tested the setting to start at "2000" in bank 1. The switches are all the same but bank 1 is now 10100.
My motherboard memory covers the first 2MB of RAM for now.

Here's the result:

That cheetah card brought back memories instantly for me!

We had this exact same card with 2MB in our 286/12Mhz computer back in the day and I remember it enabled something in Wing Commander that otherwise wasn't there, but I can't exactly remember what.

Still cool to see this card again.

Salient wrote:

That cheetah card brought back memories instantly for me!

We had this exact same card with 2MB in our 286/12Mhz computer back in the day and I remember it enabled something in Wing Commander that otherwise wasn't there, but I can't exactly remember what.

Still cool to see this card again.

It was EMS which enables the speech pack 😉

FAMICOMASTER , you need to test some more settings to get the code...at least bank 1 has moved.

I'm on combination 5. The last 4 have been identical, no change in the configuration shown:
10100
10011
10010
10001
10000

They are having no effect on the starting location of the memory.

How about to continue with 111111 and to move it towards the end? That reduces the risk that the bank you have moved interferes with already existing RAM...think this is the reason that there is no change.

If I switched it into a position where it would interfere with other memory, wouldn't it just not appear or cause memory errors?

Just to humor you, I tried it anyways. Bank 1 is all 1s:
Absolutely no difference.

Just for kicks, I put all of the switches for "Bank 1" off.

Result:

FAMICOMASTER wrote:

If I switched it into a position where it would interfere with other memory, wouldn't it just not appear or cause memory errors?

Good question...I would expect memory errors?

FAMICOMASTER wrote:

Just for kicks, I put all of the switches for "Bank 1" off.

Result:

Here you see my table is not that bad. You moved bank 1 from
3,875 100011 x111xx
to
7,375 111111 (but also switched another bank..strange?)
--> Just see, you switched to 000000 not 111111 therefore the table is wrong 🙁

Can you try to switch also all other banks to keep it in one piece, e.g.
bank 1 4,375 100111 x11xxx
bank 2 4,875 101011 x1x1xx
bank 3 5,375 101111 x1xxxx
bank 4 5,875 110011 xx11xx
bank 5 6,375 110111