• Please review our updated Terms and Rules here

Next Lomas project - Thunder 186

RichCini

RichCini

Veteran Member
Joined
Aug 7, 2005
Messages
565
Location
Long Island, NY
Since I have the Color Magic reproduction working, I was going to tackle the Thunder 186. This is an 80186-based all-in-one card with a floppy controller and 256k of DRAM. There are only two PALs on it, and they seem to be simple address/status decoders, so I figure they will be easier to reverse.

I spent some time drawing the schematics (originals attached) and there’s something I need some help with. Like the ColorMagic, Lomas used stacking memory sockets, in this case allowing for 36 DRAM chips (4164-15L, total of 256k of parity RAM) in the space of 18. They are arranged in even/odd rows using one CAS for all and two RAS signals from the TMS4500/VL4500 DRAM refresh controller for the lower and upper chips in the socket. The TMS4500 tops-out at 64K devices, but the board has a DIP-48 footprint which allows the alternate controller, the VL4501, to be mounted. The VL4501 can handle 256K devices. The TMS4500 is bottom-justified in the socket so the signals line-up.

Given the board density and the lack of stacking sockets, I need to convert this to something that could use only one or two rows of unstacked chips.

The manual makes no mention of the alternative DRAM controller, with the only reference being in the schematics. I’m thinking that if I switch to the VL4501 and use only 18 41256-15L chips (256kx1) it might work. But, that’s what I wanted to ask the group about. The memory schematics are on page 3 of the attached. I guess I could make a mezzanine board with 36 chips, but I'd like to avoid that -- I'd need 32 pins (2x8 pin header times 2), but that might make routing the main board easier. I know several PC/ISA memory expansion boards did that, like the 4MB memory board in my Compaq DeskPro/386.

I understand that AST Research used the VL4501 in their PCjr sidecar memory, but I can’t find a manual or schematic of it.

If anyone has suggestions, I'm all ears. Thanks!

Rich
 

Attachments

  • Pages from LDP Thunder 186 manual.pdf
    912.9 KB · Views: 5
RichCini said:
Since I have the Color Magic reproduction working, I was going to tackle the Thunder 186. This is an 80186-based all-in-one card with a floppy controller and 256k of DRAM. There are only two PALs on it, and they seem to be simple address/status decoders, so I figure they will be easier to reverse.

I spent some time drawing the schematics (originals attached) and there’s something I need some help with. Like the ColorMagic, Lomas used stacking memory sockets, in this case allowing for 36 DRAM chips (4164-15L, total of 256k of parity RAM) in the space of 18. They are arranged in even/odd rows using one CAS for all and two RAS signals from the TMS4500/VL4500 DRAM refresh controller for the lower and upper chips in the socket. The TMS4500 tops-out at 64K devices, but the board has a DIP-48 footprint which allows the alternate controller, the VL4501, to be mounted. The VL4501 can handle 256K devices. The TMS4500 is bottom-justified in the socket so the signals line-up.

Given the board density and the lack of stacking sockets, I need to convert this to something that could use only one or two rows of unstacked chips.

The manual makes no mention of the alternative DRAM controller, with the only reference being in the schematics. I’m thinking that if I switch to the VL4501 and use only 18 41256-15L chips (256kx1) it might work. But, that’s what I wanted to ask the group about. The memory schematics are on page 3 of the attached. I guess I could make a mezzanine board with 36 chips, but I'd like to avoid that -- I'd need 32 pins (2x8 pin header times 2), but that might make routing the main board easier. I know several PC/ISA memory expansion boards did that, like the 4MB memory board in my Compaq DeskPro/386.

I understand that AST Research used the VL4501 in their PCjr sidecar memory, but I can’t find a manual or schematic of it.

If anyone has suggestions, I'm all ears. Thanks!

Rich
Click to expand...
If you would be satisfied with a functional reproduction of the Thunder186 (I certainly would) would you consider simplifying the design by replacing all the DRAM chips and associated refresh circuitry with a single 512Kx8 SRAM chip like the AS6C4008?
 
That’s a great idea. I would need 2 to get the 16-bit data width and it would save a ton of chips. Let me look into that more. 512k would definitely be better than 256k. It wouldn't have the parity function but does that really matter? Likely not.

I did a similar, but easier, substitution on the ColorMagic, using two higher density RAM chips to replace 4.

More to come. Thanks!
 
RichCini said:
That’s a great idea. I would need 2 to get the 16-bit data width and it would save a ton of chips. Let me look into that more. 512k would definitely be better than 256k. It wouldn't have the parity function but does that really matter? Likely not.

I did a similar, but easier, substitution on the ColorMagic, using two higher density RAM chips to replace 4.

More to come. Thanks!
Click to expand...
For this application, RAM parity checking can be removed. The 74LS280 parity checker (U51) can then be eliminated, and the "parity error" signal it generates can be forced to an "always OK" state.
 
Last edited:
Ok, here's what I have so far. This is the rough re-draw that I did, but it's not fully verified. The alternative SRAM setup is on page 7, but I think this might make it easier to compare than the original scans. I'm hoping to get better scans soon. I also plan on hand-verifying each schematic net.

Thanks!
 

Attachments

  • schematics.pdf
    1.1 MB · Views: 4
If you're willing to do surface mount you can get 16 bit wide chips (with high/low byte selects) that would give you all your RAM in one chip. The AS6C8016 is an example of a 1MB (512Kx16) chip, if you really only need 512K there's the CY71041G. (Both of these are 5v.)
 
I guess I'm not opposed to that for a single chip for me, but I'm thinking of ease for various builders. It seems that anything from Alliance that's 16-bit wide is the TSOP package.
 
Yeah. I bought some of the CY chips (in the 64Kx16) size for a project that I keep not finding time for, but I'm not looking forward to soldering them.
 
Al — the manual for the T186 references a driver disk of some sort but I don’t have it. It also doesn’t have a complete BIOS listing so I can’t tell how “compatible” it is. But, it will run DOS 3 and I’ve run a graphics-heavy game or two. I have to try some other games like from Sierra, and maybe a CP437 program like Norton commander.
 
Al Kossow said:
what software did Lomas release for the 186 card?
Click to expand...
Lomas offered MS DOS 2.1 and Concurrent DOS. When paired with the Color Magic video board, you could run IBM PC-DOS or compatible. I was able to run modern FreeDOS on the Lomas 286 and ColorMagic combo and even Linux (ELKS), I don't know if the 186 would have any problems or not with the same.
 
The images of DOS 2.1 I have don’t seem to boot with my Gotek’s for some reason. I have a version of DOS 3 which boots perfectly and is pretty snappy.

{Ed: this came from @new_castle_j, who also supplied the original boards used for testing and comparison}
 
RichCini said:
The images of DOS 2.1 I have don’t seem to boot with my Gotek’s for some reason. I have a version of DOS 3 which boots perfectly and is pretty snappy.
Click to expand...
I think Lomas did not include the 55 AA boot signature in their boot sector on the MS DOS 2.1 distro for the T186. I remember having to add that in manually in order to read the disk in an IBM PC, perhaps the Gotek needs that 55 AA as well?
 
new_castle_j said:
I think Lomas did not include the 55 AA boot signature in their boot sector on the MS DOS 2.1 distro for the T186. I remember having to add that in manually in order to read the disk in an IBM PC, perhaps the Gotek needs that 55 AA as well?
Click to expand...
Hmmm. Interesting. I’ll look at that. There might be a way to have it ignore the signature.
 
Ok, some interesting info. The first DOS 2.0 image is clearly LDP custom -- I don't have any programs that can read and interpret the boot sector. I added the AA55h signature but that didn't matter for the GoTek/FlashFloppy.

The second image has a an OEM system name of "MSWIN4.1" which tells me it was formatted 360K on something Win9x. This barfs with bad opcode errors. So, likely 386 instructions in it.

The third image, DOS 3, looks like a standard PC-DOS boot sector and that boots just fine.

Tomorrow I'm going to look through a sector dump of the LPD disk and see if I can parse through it. I dumped the sector and ran it through Sourcer. On quick look, it seems to make calls into ROM @FE00:0004 to boot the disk. I'd bet those routines don't exit in the ColorMagic ROM, but I'll look into that this weekend. Maybe they exist in the standard monitor ROM.
 
Also…the LDP DOS 2.0 disk is 320k, not 360k, based on the parameter block in the image. Based on this I’m going to try some of my disk tools again with custom parameters.
 
Are you using "raw" images? I wonder if you might have more luck using one of the more explicit image formats, like HFE? I remember having some issues with 8-sector images when playing with some DOS 1.x stuff on a Gotek, but that was many moons ago.
 
They are raw images converted to HFE. However, now that I actually parsed the DPB, I need to remake the HFE with the right parameters. I assumed it was a 360k image.
 
I also just noticed that three fields of the DPB - # of FAT sectors, # of heads, and # of hidden sectors — seem to be byte swapped (MSB, LSB) as compared to a normal DPB. So it would reports SPT as 0800 rather than 0008.

I’m going to look for a generic DOS 2.11 image and look at it. Not sure any existed in 320k format though.
 
You must log in or register to reply here.
Back
Top