PET 4016 Upgrade options?

[Hutch]

Something is escaping me here. I'm looking over this mod, but I cant find anything about the need to increase the video RAM on a 40 col. machine. I suppose that it's not a requirement if you're modifying an 80 col. machine - but in the case of a 40 col. it's a different story.

I'd initially assumed that the additional RAM would be on the daughterboard, but that only holds logic chips. Could someone please supply additional details on this?

You'll need to populate the chips in this area. You'll effectively convert it to an 80 col machine with the ability to switch back.
IIRC it will require a custom EDIT ROM to initialize the CRTC for 80 columns, but will maintain the same keyboard mapping.
vRAM is 2114 static RAM and there's some additional logic chips.

 

[daver2]

The Universal PET would work in either 40 or 80 column mode. It was selectable via wire links.

I suspect Commodore produced this card so that they only held stock of one (1) PCB that could be configured for two (2) personalities. As a result, I would assume, that a Commodore Universal PET configured for 40 column mode may not have the additional RAM etc. - but this is only a guess on my part.

I think someone devised this little board to be fitted to the Universal PET (with the full complement of video RAM etc. fitted) to be able to switch from 40 column to 80 column "on the fly". The board itself only contains solid state switches (the '157 chips). All the 'smarts' is done on the Universal board.

Dave

 

[WimWalther]

The Universal PET would work in either 40 or 80 column mode. It was selectable via wire links.

So this 40/80 Switcher board takes the place of those wire links, replacing them with logic chips so that the change can be made by flipping a single switch which controls the logic?

I suspect Commodore produced this card so

Sorry, which card is this? The 40/80 Switcher board is a 3rd party item, no?

 

[WimWalther]

IIRC it will require a custom EDIT ROM to initialize the CRTC for 80 columns, but will maintain the same keyboard mapping.

So the 'EDIT ROM' is the built-in screen editor thats required to operate BASIC, the ML monitor, etc?

vRAM is 2114 static RAM and there's some additional logic chips.

2114 SRAM.. just like the C64 and probably other CBM machines. I can only assume that CBM / MOS produced them in-house.

ETA: But 2114 in the C64 was used for color RAM, not screen RAM. And while MOS may have produced 2114 at some point, the parts in my PET are 3rd party

Though now I'm questioning the whole upgrade plan. The main reason for bumping the RAM up to 32K was to play PETSCII Robots natively on a PET. Now I come to find thst a disk system (or what subs for one) is an absolute requirement for the game.. and I'm just not sold on the idea of buying one for a single use.

 

[daver2]

>>> Sorry, which card is this? The 40/80 Switcher board is a 3rd party item, no?

Sorry, I meant the Commodore Universal PET main logic board. Not the 3rd party switcher card. I assume Commodore had no interest in making their product 'universal'. They would want to sell two (2) PETs!

Commodore then just installed the appropriate devices, set the links and installed the correct EDIT ROM and they have either a 40 column or 80 column PET.

Dave

 

[Hutch]

So the 'EDIT ROM' is the built-in screen editor thats required to operate BASIC, the ML monitor, etc?

The screen editor
The EDIT ROM contains all the code that is unique to the hardware of a specific PET.
It initializes the CRTC chip to 40 or 80 columns, plays the startup chirp if the PET has a speaker, clears the screen
and it contains the keyboard matrix translation for the N or B type keyboard or foreign language keyboard types.

It's usually the only ROM chip that is socketed so it can be changed for other countries or 40xx/80xx operation.
Commodore never produced 2114 chips in house. MOS produced 6550 static RAM chips used in the original chicklet PETs, but they had yield and reliability problems so they switched to 2114 around mid 1978 and never used 6550s again.

 

[Hutch]

Though now I'm questioning the whole upgrade plan. The main reason for bumping the RAM up to 32K was to play PETSCII Robots natively on a PET. Now I come to find thst a disk system (or what subs for one) is an absolute requirement for the game.. and I'm just not sold on the idea of buying one for a single use.

The easiest thing for disk access is an SD card reader that emulates a Commodore disk drive.
There are only two options available for sale at this time,

The SD2PET from Tynemouth/The Future was 8 Bit, which costs around $150 US and ships from the UK.
This is a high quality device that supports most file types and disk image formats, but has been discontinued so current stock is the last of them.

The other option is the PetDisk Max from Bitfixer, which costs $40 US, ships from California and is only available in Kit form.
This supports PRG files and .d64 disk images only but is still under active development. You would be able to run Petscii Robots fine from this one.

Also note, if you want to use the SNES controller adapter that comes with the boxed version of Petscii Robots, you need to provide 5v for the adapter to power the game controller since the PET doesn't have 5V available on the user port.
Link: Using a SNES Controller for Petscii Robots on the PET

 

[WimWalther]

The screen editor
The EDIT ROM contains all the code that is unique to the hardware of a specific PET.
It initializes the CRTC chip to 40 or 80 columns, plays the startup chirp if the PET has a speaker, clears the screen
and it contains the keyboard matrix translation for the N or B type keyboard or foreign language keyboard types.

It's usually the only ROM chip that is socketed so it can be changed for other countries or 40xx/80xx operation.

Yes, mine is socketed. It's ROM code 901499-01, which according zimmers is "Screen editor for BASIC 4, normal keyboard, CRTC, 40 columns, 60 Hz"

So what's the deal with the two different keyboard encodings? Do the 'business' class machines have different keyboard markings?

Just trying to understand why CBM did some of the stuff that it did..

 

[Hutch]

So what's the deal with the two different keyboard encodings? Do the 'business' class machines have different keyboard markings?

Different markings and a completely different shape and layout. The maxtrix is totally different also, pressing A or B on one doesn't equate to A or B on the other.
Commodore referred to these as "Typewriter Style" and "Terminal Type" keyboards.
Teletype and video terminal keyboards of the day were very different from typewriter keyboards and I guess computer users expected terminal type keyboards.
Or terminal type keyboards were cheaper and easier to get off the shelf in the beginning.
Because of the different size and shape, you can't swap the keyboards into a case made for the other type.

Business -B (top) and Normal -N (bottom)

 

[WimWalther]

The easiest thing for disk access is an SD card reader that emulates a Commodore disk (...)

The other option is the PetDisk Max from Bitfixer, which costs $40 US, ships from California and is only available in Kit form.
This supports PRG files and .d64 disk images only but is still under active development. You would be able to run Petscii Robots fine from this one.

The PetDisk seems like a very reasonable alternative to the $150 SD2PET. If its presently supporting. PRG & .D64 thats all you really need. I'm fairly sure there are utils to convert .TAP to .PRG, until they offer direct support.

It looks like the PetDisk also covers supplying 5V on the user port, as it also requires this. Weird that its not a stock feature.. seems so obvious & useful. Though I'm not totally cool with their spring clip approach. I'd just as soon solder that end of the 5V line.

 

[ScottishColin]

The PetDisk seems like a very reasonable alternative to the $150 SD2PET. If its presently supporting. PRG & .D64 thats all you really need. I'm fairly sure there are utils to convert .TAP to .PRG, until they offer direct support.

It looks like the PetDisk also covers supplying 5V on the user port, as it also requires this. Weird that its not a stock feature.. seems so obvious & useful. Though I'm not totally cool with their spring clip approach. I'd just as soon solder that end of the 5V line.

I have the PetDisk and can recommend it. It has two features that are better than the PET2SD - it has WiFi and can connect to a server (local or internet) that has .PRG or .D64 stored on it - I built a Pi server to do this; it also has the IEEE port extender so I can plug in my 4040 disk drive at the same time.

It took about 15 minutes to solder up and worked first time here.

Colin.

 

[Eudimorphodon]

Commodore referred to these as "Typewriter Style" and "Terminal Type" keyboards.
Teletype and video terminal keyboards of the day were very different from typewriter keyboards and I guess computer users expected terminal type keyboards.
Or terminal type keyboards were cheaper and easier to get off the shelf in the beginning.

Those are terms that Commodore made up to cover for just how weird their “Normal/Graphics” keyboard was. It wasn’t even remotely “standard” or off the shelf. (Commodore made all its keyboards in house anyway.) Its bizarre layout with non-shifted symbols over the letter keys and numbers only on the 10-key pad is the result of Commodore directly translating the layout of the original PET’s tiny calculator-style keypad into something it was at least possible to type on.

The “Business” keyboard (which was *mostly* on 80 column PETs, but they did make a few 9” 40 column PETs with it as well, just to keep everything as confusing as possible), as noted, has a completely different matrix because of its different physical layout. (Commodore mostly arranged their matrixes to optimize for efficient trace layout, they didn’t do tricks to put the letters in alphabetical order, etc, like you’ll see on a lot of contemporary keyboards to optimize ASCII encoding.) And on the software side the EDIT rom has a more sophisticated scanning routine to handle the shifted number/symbol row, etc.

That’s really the big gotcha with that 40/80 column switching board. You can totally hack together a custom edit ROM that’ll work fine under BASIC in 80 columns with the “Normal” keyboard, but any business software that uses its own keyboard routine instead of the Kernel one will have issues. And on the flip side, if you install that board in an already 80 column PET so you can run games more easily (there are a lot more games and “fun” software for 40 column) you’ll likewise end up with scrambled keyboard input in a fair amount of software. If you’re just a “casual” PET user and already have a 40 column machine I wouldn’t actually consider it a fantastic investment.

 

[Hutch]

(Commodore made all its keyboards in house anyway.)

The keyboards were actually all made by an OEM, even for their calculators before the PET. They made very little in house actually.
No doubt they were made to Commodore's specifications to be sure, but if you look closely at vintage Apple 1 keyboards, or ADM terminal keyboards you'll see they are very similar.

The same OEM that made calculator keyboards for them make the original chicklet PET keyboard. The small size was chosen so they could fit the tape drive in there also, because they wanted an all-in-one turnkey solution. (google "commodore 1540 calculator") The later -N keyboard layout was made to be compatible with that original chicklet keyboard.

The biggest problem with the layout was the design choice to fit all of the special graphics characters on the keyboard, so the number keys are only available on the number pad and not duplicated to the top row. That's the thing that trips me up every time I type on one.

 

[Eudimorphodon]

The keyboards were actually all made by an OEM, even for their calculators before the PET. They made very little in house actually.
No doubt they were made to Commodore's specifications to be sure

“Made to specifications” is what I meant by “in house”; they were complete custom jobs in terms of both layout and key cap labels. Unlike, say, Apple, which used an off the shelf Datanetics keyboard; until the Apple II hit the sort of production numbers that justified specific OEM modifications it was literally the same unit anyone could buy out of a catalog. (Its wide availability is why Apple recommended it for the Apple I.)

Anyway, it’s clear why calling the PET “N” keyboard a “terminal” keyboard is at best highly misleading, because that implies that is complied with some kind of “standard”; it most certainly did not. There were some later computers that shipped with highly nonstandard (usually chicklet) key layouts with graphics characters assigned to shift positions, etc, that are very much in the same spirit as the original calculator PET, but it was not at all a standard thing when The PET came out. I will grant the possibility that someone may have made some kind of application-specific terminal that did something like PETSCII graphics (and there were programming Languages like APL on the IBM 5100 that used custom symbols, that I suppose you could argue were similar) before the PET, but your “typical” computer terminal keyboard was nothing like the PET’s.

 

[Hutch]

calling the PET “N” keyboard a “terminal” keyboard is at best highly misleading, because that implies that is complied with some kind of “standard”;

No standard at all. I've never seen any two terminals from the 70's that had the same keyboard layout. DEC, Teletype, ADM, Televideo keyboards are all completely different.
I don't think the marketing material meant to imply there was a standard, just that the keyboard wasn't a typewriter standard.

 

[Eudimorphodon]

No standard at all. I've never seen any two terminals from the 70's that had the same keyboard layout. DEC, Teletype, ADM, Televideo keyboards are all completely different.

Certainly, there was no "standard" layout in the way we think of, say, the PS/2 101/104 key layout as "standard", but they did in the vast majority of cases comply with general conventions like "holding down shift and a letter inputs the opposite case of that letter" and "unshifted top row is numbers, shifted top row is punctuation", etc, unlike the PET.*

(* The punctuation thing is actually kind of interesting, because you'll notice if you compare the position of punctuation symbols on most 70's keyboards with where they live on 80's and later computer and typewriter keyboards you'll notice a lot of differences, like where the double-quotes and exclamation points live. This circles back to that thing I mentioned about optimizing keyboard layouts for "ASCII encoding". If you look at an ASCII code table you'll see that with the arrangement you find on 70's computers the punctuation starting at decimal code 32, space excluded, is in the same order as the numbers counting from 48. This is not a coincidence; taking advantage of situations like this where you can take a bit position off the matrix and add it to two different constants for shifted/not shifted saved a lot of hardware if you had to do keyboard encoding with discrete parts, tiny PROMs, or really brain-dead software. Even computers that used dumb matrixes, like the TRS-80, instead of decoded keyboards sometimes kept the ASCII-optimized matrixes. The PET is actually sort of "innovative" for just using arbitrary translation, like later keyboards with their ANSI punctuation layouts do.)

I don't think the marketing material meant to imply there was a standard, just that the keyboard wasn't a typewriter standard.

Sure. I'd just say in the context of this ad it's specifically just meant to distinguish it from the calculator style keyboard from the original (which was still on sale in that one), not any more than that. In other words, a polite way to say "The layout is still completely on crack, but it's at least type-able like a real terminal.".

Also of interest in this ad the "CBM 2001 Business Computer"; that's the 40-column + business keyboard model I mentioned. I'm curious how common those actually were. I have several PETs that I got at a huge warehouse dump a decade ago (they were former school machines) and I don't *remember* seeing any 9" business keyboard units out of literally hundreds of machines, but I certainly could have missed one. (There were 80 column units, I grabbed one. They mostly had engravings indicating they belonged to the college.) I vaguely recall reading somewhere they may have been more popular in Europe?

 

[WimWalther]

One more pic of the machine's internals after an appointment with Dr. Dyson




Plan now is to u/g the RAM to 32K and leave it at that. I will still order a small batch of the 40/80 Switcher boards anyway, and offer them for sale here & elsewhere, keeping one for myself if I change my mind later

 

[WimWalther]

Well THAT didn't work..





I'm not aware if the 15359 Bytes figure is correct for a 16K machine, but it is about half of the 32768 bytes that I'd roughly expect from a 32K machine.

Earlier in this thread I was told that no jumper changes are necessary.. that the new RAM would be autodetected.. so I've done nothing except install the new sockets & chips as in the photo.

ETA: What should the byte count be, 31743?

Ideas?

 

[ScottishColin]

My 3016 is 15359 bytes free. If I up it to 32k (using a Tynemouth board) I get 31743 bytes free.

Colin.

 

[WimWalther]

@Colin

Thanks, that agrees with what I'm finding.

@all ,

So I wrote a little BASIC to poke & peek the new RAM range (16384-32767) with 0, 1 & 255, and the PEEKs return garbage. It's as if the new RAM isn't even present.

I suppose it's worth trying each bit individually, which will be next. The keyboard is in desperate need of cleaning, which makes typing no fun at all.. should probably take care of that as well. But the RAM thing is bothering me.

ETA: The individual bit test (poke x, (0,1,2,4,8,16,32,64,128)) gives the same results.. just garbage data. Now the test is only up to 19000 so I'll let it run, but I'm not expecting much.