VIC-20 16K Prototype

[CommodoreZ]

Howdy,

A few weeks ago at VCF East, I spoke with Robert "Bob" Russell, one of the VIC Commandos. I don't know what convinced him to do so, but he gave me a pair of later VIC-20 prototype boards.

First, the 16K VIC-20 Prototype. This machine is an internal prototype from Commodore of a VIC-20 variant that never made it to production. Some folks call it a VIC-16, but we have no official designation for it.

Back to the prototype, this seems to be the only survivor of the project. It was supposed to be sent out for FCC certification RF testing, but whether or not it made it that far is unknown. The machine does not include the top half of the case or keyboard, much less a serial number. It's based on the common 324003 production motherboard, with a few ICs rearranged to make room for 8x 6116 SRAMs in place of 8 of the existing 2114 SRAMs. If you pay close attention, you'll note that it has 17K of RAM, not 16K, plus the color RAM. When you start the machine, it reports 15871 bytes free. The power, 6522, CPU, regulator section, and VIC chip/RF cage otherwise look like a 324003. There are three stickers with notes on the system, and here's my best interpretation of what they say:

OK SMW 1-2-82
LATEST TEST
155 RESET

BAD 1/5/82 RAS

155
LARGE FUSE
BI BAD 6560

I'm not sure what the significance of 155 RESET is, but I have noticed this machine takes longer to display a READY prompt than most other VICs. Considering it does in fact run, whatever was wrong with the 6560 may have warranted a replacement. The ROMs on board were dumped and were confirmed to match production VIC-20 ROMs bit for bit. It has an MOS 6560E VIC chip, which is odd because the E suffix only seems to be present on PAL VICs. If you've seen an MOS6560E in something else, VIC or otherwise, I'd love to hear about it. The board is marked with a unique assembly number of 3016006.




The brown one is 324003 that has no solder mask or silkscreen, and was intended as an internal RAM test board at Commodore. Bob said that it was Jack Tramiel's idea to test the viability of Russian or Eastern Bloc 2114 SRAM, however this appears to have 10 Toshiba 2114s, and one unsocketed 2114 of indistinct marking.

I've taken to calling it the "brown board" because it has no solder mask or silk screen, and is just tinned copper on top of vaguely brown fiberglass PCB. Apart from the PCB, this board sports an early production heatsink for the 5V regulator, typically only seen on early Japanese VIC-1001s before they realized it burned users who reached in the cartridge port.

Bob says that Jack Tramiel wanted to test the viability of cheap Russian or Eastern Bloc 2114 SRAM. This machine was an internal engineering test bench to that end. However, the SRAM on board does not corroborate this claim. The 10 socketed 2114s appear to be made by Toshiba, and the remaining one is both soldered in place and has indistinct markings. It has no top half of the case or keyboard, but it does have a serial number of V1002001. It has a beautiful gold and purple ceramic MOS 6560 VIC chip, and some minor mods to the analog circuitry. The ROMs on board were dumped and matched to standard production VIC-20 ROMs. I've tried dialing in the color phase, but it's finicky -- otherwise the brown board works fine.



I've put all of my photographs and details I've discovered here:
https://www.commodorez.com/vic.html


The 16K machine was previously documented on Secret Weapons of Commodore - Super VIC.

[orion70]

CommodoreZ, you really have one of the nicest VIC-20 collections I have ever seen
You were really lucky to meet Bob Russell, and have these valuable gifts from him.

The VIC-16 is and interesting beast (you are the owner of one of Commodore's secret weapons!!!), but probably this project with internally expanded RAM was doomed by the contemporaneous development of the C64. The "brown board" is very valuable as an historical find too, as it marks Tramiel's try at Soviet products - no need to be picky when it comes to save some bucks

Thanks for having shared this precious collection with us!

[Mike]

If you pay close attention, you'll note that it has 17K of RAM, not 16K, plus the color RAM. When you start the machine, it reports 15871 bytes free.

Oh! I always had thought that particular machine would simply put RAM in $0000..$3FFF which would especially make the extra RAM at $0400..$0FFF available for access by VIC, much akin to my VFLI mod (except the expanded colour RAM).

However, as you described it, that prototype leaves $0400..$0FFF open. The two 2114s then populate 1 KB at $0000..$03FF and the eight 6116s then populate $1000..$4FFF, which gives a rather unusual RAM size halfway between a +8K RAM expansion (11775 BYTES FREE) and a +16K RAM expansion (19967 BYTES FREE). This poses the interesting question how that VIC-20 is going to handle external RAM in the half filled BLK2.

Normally, the 3 KB in $0400..$0FFF are locked from BASIC use as soon as there is RAM in BLK1, so my first assumed option would have reduced the available BASIC RAM to the same value as for a +8K RAM expander. But then, with $0400..$0FFF accessible by VIC, they could have quite as well put the screen at $0400, allowing BASIC to use $0600..$3FFF, which would then have given 14847 BYTES FREE, and up to 31231 BYTES FREE if RAM was added to BLK2 and BLK3. (On the cartridge port, /RAM1..3 and /BLK1 would then simply have to appear deactivated all the time).

Code: Select all

OK SMW 1-2-82
LATEST TEST
155 RESET

If you read that as "15s RESET", it makes more sense and is in congruence with the long start-up time.

[chysn]

If you read that as "15s RESET", it makes more sense and is in congruence with the long start-up time.

Wouldn't the long start-up time be explained by simply having more RAM to test? Every time I turn on a VIC with a 24K expander, I experience several seconds of dread!

[CommodoreZ]

You know, y'all might have something there. I measured the startup time, and it's closer to 19 seconds from when I throw the switch. 15S instead of 155 makes alot more sense now.

I'd like to be able to post a definitive memory map of this unique VIC, and I can think of no better place to ask than here! I've included what I've used so far to check, along with what it returns. Any suggestions for how best to approach this?

My intent here is preservation, so if you have questions about these machines, or things you want me to check, let me know.

[Mike]

Any suggestions for how best to approach this?

36869 has a different mapping from how you interpret it (it's not the high byte of any address, but rather encodes the screen and character base address in its higher and lower nibble, respectively) and of course FRE(0) 'misses' the space already occupied by your test program.

Other than that, the output is pretty much in concordance with what I wrote: the screen is at $1000 and BASIC RAM spans $1201..$4FFF.

[CommodoreZ]

36869 has a different mapping from how you interpret it (it's not the high byte of any address, but rather encodes the screen and character base address in its higher and lower nibble, respectively)

Then I will remove the 256 multiplier.

FRE(0) 'misses' the space already occupied by your test program.

A friend recommended that I toss it in, so I obliged.

[CommodoreZ]

Alright, here's the updated version that I plan to put on the webpage about these boards.

I've also taken some better lit photos of the mystery SRAM on the brown board:

It sure looks like an MOS part to me. MPS 2114, 30L 2280
Does that seem like a reasonable interpretation to y'all?