ROM: How did they do it?

[Miles J.]

There are some questions I have been thinking about for quite a while, especially after reading 'How to Build a Microcomputer' by Sam Creason.

Imagine yourself living in the year 1976. You have bought a KIM-1, and now you'd like to expand it with your own 8 kb ROM (at $e000..$ffff). How would you do it? And by this I'm not refering to address decoding or connecting the ROM to the data bus.
What I'd like to know is: How would the ROM itself be implemented? The size is way too big for a DIPROM as described in the book. For 8 kb you'd also need 16 4 Kbit (512 Bytes) 2704 EPROMs, which is quite a lot. Now MOS and others offer 2 kb ROM chips (like the 2316 as found in the Apple II), but could a hobbyist even buy one of these? After all, the ROM chip had to be custom made for you. Could you simply send a letter to MOS saying 'Dear Sir or Madam, I would like to order a 2 kb ROM from you. For this I have enclosed 10 sheets of paper on which the contents of the chip are described in hexadecimal notation. Thank you very much. Yours sincerly H. Obbyist"? And if this were possible: How much would it cost you? Would it be affordable for a hobbyist? How did Steve do it?
Unfortunately, I have found a lot of datasheets about those old chips but so far no price list that could give the actual price and how to order. Maybe someone here in the forum knows more about it?

[Dr Jefyll]

Quote:

MOS and others offer 2 kb ROM chips (like the 2316 as found in the Apple II), but could a hobbyist even buy one of these?

I *did* own a KIM-1 in the 1970's. And for hobbyists, the solution was to use an EPROM such as the 2716, which can be programmed and erased in one's own shop.

ROM's (as you seem to suggest) were simply out of reach except for commercial firms with deep pockets. Because of the hefty setup charge, it doesn't make economic sense to order a batch of ROM's unless a rather large quantity is involved.

Yes EPROM's were affordable. I don't recall the prices, but you could refer to the advertising in period issues of BYTE magazine and others.

-- Jeff

[barnacle]

Very occasionally PROMs were used in projects: I recall a memory board expansion for my Microtan 65 that required a small prom blowing as part of the address decoding; I had to have it programmed by a third party recommended by the magazine that published the article describing the circuit. In that case, I think the author had already provided them with the code, but yes, I did provide a hex listing or something similar.

But in general, 2716 and (wonders!) 2732 EPROMs did the job. And the company programmer and UV eraser.

At the time - late 70s - I was developing a part using an 8048 or 8049 microcontroller; they had a one-time-only programmable internal memory, so I was using a UV erasable part - 8748 I think - to get the code right before blowing the *much* cheaper one-time parts.

Neil

[Miles J.]

Thanks for your answers!

Dr Jefyll wrote:

ROM's (as you seem to suggest) were simply out of reach except for commercial firms with deep pockets. Because of the hefty setup charge, it doesn't make economic sense to order a batch of ROM's unless a rather large quantity is involved.

Unfortunately, just as expected. At least it explains why EPROMs could be installed in the Apple II as an alternative.

This basically means that you had to
1) wait for the year 1977 when the 2 kb 2716 EPROM was released (or 1979 for the 2732),
2) then buy four of them for 8 kb and have some company program them for you.

Dr Jefyll wrote:

EPROM's were affordable. I don't recall the prices, but you could refer to the advertising in period issues of BYTE magazine and others.

Interestingly, according to an advert I found in "Micro - The 6502 Journal" from 1977, the EPROM 2716 costs $38 and the 2 kb 4116 Ram $42, only slightly more. This means that the memory alone (16 kb Ram, 8 kb Rom) costs 8 * $42 + 4 * $38 = $488 for a home-made 6502 computer back then. Not exactly cheap, but that's how it was.
(Just in case you wonder why I asked this question: I am currently working on a 6502 computer (right now emulation only, later perhaps FPGA), and I am curious if this computer could actually have been built in the 70s.)

[GARTHWILSON]

Miles J. wrote:

2) then buy four of them for 8 kb and have some company program them for you.

My first EPROM programmer was home-made, totally manual, where you set the address with switches, set the data, and press a button to generate a 50ms programming pulse.  It had to be manual, since I didn't own a computer of any kind that could control a programmer yet.  As you can imagine, besides being slow, it was terribly vulnerable to human error, and without a UV eraser, I had to put the EPROM out in the sun for a week to erase it.  I bought two or four EPROMs so I could be working on one while another one was out in the sun erasing.  That worked, but was obviously not terribly useful, even for hobby.  Soon a friend helped me out with a programmer he had made which was controlled by his HP-71B handheld computer; then not long after that, I could use an EPROM programmer at work that the company bought for $500 (back when $500 was a lot more money than it is today).

[BillO]

GARTHWILSON wrote:

Miles J. wrote:

2) then buy four of them for 8 kb and have some company program them for you.

I had to put the EPROM out in the sun for a week to erase it.

One of the benefits of living in S. CA.

I had a "black light" bulb I was able to buy at a local department store. I'm sure there was precious little UVB or UVC being emitted by it, but it would accomplish the job in about the same time. Not that we don't get 7 days of sunlight in a row here in central Ontario, it's just that you can't count on it.

[Dr Jefyll]

barnacle wrote:

At the time - late 70s - I was developing a part using an 8048 or 8049 microcontroller; they had a one-time-only programmable internal memory, so I was using a UV erasable part - 8748

The 8748 -- yes!! Ah, memories! (My own memories, I mean -- not RAMs or ROMs! )

Hard to imagine a time when the now-ubiquitous microcontroller was not yet a thing! But when the Intel MCS-48 family was introduced, it was one of the first products of its kind (or indeed perhaps the very first).

Die-hard home-brew fanatic that I was, I was able to build my own 8748 blaster based on info published in the Intel manual, and I did a string of mostly one-off applications for a client in the printing industry.

This all began before I acquired my first 8088 box, so there were necessarily some other homebrew aspects as well. The host computer was my grotesquely modified KIM-1, augmented by a home-brew floppy-disk system. And my homebrew MCS48 assembler was written in 6502 FIG Forth.

Appropriate -- for me, at least -- to see the 8748 mentioned in a thread in the Nostalgia section!

-- Jeff