13

According to https://en.wikipedia.org/wiki/Second_source

MOS Technology licensed Rockwell and Synertek to second-source the 6502 microprocessor and its support components.

This makes sense; the 6502 would get more design wins with second sources available.

When was the first of these license deals made, and what exactly were the terms? I have not been able to find any details on that.

One reason I'm interested is because I'm curious about how the industry perceived the 6502 in the light of the Motorola lawsuit and subsequent sale of the company; at what point was the CPU perceived as something that was going to be a winner, and worth taking up.

Another reason I'm interested is because of the extreme willingness of MOS Technology to sell Atari the 6507 (a version of the 6502 with fewer address pins) at a steep discount from the already ridiculously cheap (by 1975 standards) $25 launch price of the 6502. To my way of thinking, it doesn't make sense to give that steep a discount when selling a product the prospective customer has to have, when you are already selling for a fraction of the price of your competitors (like Intel and Motorola in this case). So why did they? Was it because Chuck Peddle was a technologist, and we technologists always underprice? Or was it because they had already signed a licensing agreement, and Atari was able to say 'if you won't give us that discount, maybe your second source will'?

7
  • 3
    The MSRP of the Atari 2600 was IIRC about $200 when it launched. If one factors in the layers of markup between a manufacturer and consumer, every extra dollar the Atari had to spend on its CPU would likely have added $2-$4 to the MSRP. The competition for the 6502/6507 in the 2600 wasn't just from other chip vendors, but from other things the intended consumers could spend money on instead of a video game system.
    – supercat
    Dec 7, 2021 at 19:18
  • 3
    @Raffzahn It's a straightforward question about historical fact. I added some discussion of the reasons why I'm interested in the question, in case other people are also interested in those aspects of the history of the industry; if you aren't, feel free to ignore those parts! But the question of cost in this context is actually an interestingly tricky one, and worth raising...
    – rwallace
    Dec 7, 2021 at 19:35
  • 1
    @rwallace Not really, as it's a rather simple one, did they make profit at the price they sold the 6507 to Atari? Answer yes. Everything else is opinion.
    – Raffzahn
    Dec 7, 2021 at 19:37
  • 1
    @Raffzahn Just what is the cost of making a chip? The obvious accounting is what it costs you to run the fab. But that's incomplete. Because in a few years, you will need a new fab, and the cost of those increases exponentially with process technology. The sale price of your chips needs to cover, not just the cost of running your current fab, but that of building the new one. And it's not so clear that MOS Technology was profitable by that metric when they were selling the 6502 for $25. Of course, they soon became entangled with Commodore...
    – rwallace
    Dec 7, 2021 at 19:38
  • 2
    @Raffzahn So it's hard to assess the profitability of the chip business in isolation. But it is worth noting that MOS Technology did in fact fall behind in process technology, and ended up dropping out of the business of selling leading-edge chips, settling for making chips for Commodore, using obsolete process nodes.
    – rwallace
    Dec 7, 2021 at 19:39

3 Answers 3

18

When was the first of these license deals made, and what exactly were the terms? I have not been able to find any details on that.

All the relevant parts happened in a very short time between September 1975 and August 1976 - but most of them right around August to September 1975. A rough sequence might look like this:

According to Chuck Peddle, Synertek was their first Second source (*1), and they were quite proud about that step, as it was considered essential to be a 'real' microprocessor company (IIRC it was part of an oral history taping by the CHM). According to Peddle it was kind of an old boys contact involving John Paivinen (*2), one of the original founders of MOS and the head of Synertek Robert Schreiner, who knew each other from their time at general Electric. Synertek was, at the time, a semiconductor company focused on custom specific circuits and Schreiner saw it as chance to get into a more steady business with high integrated commodity chips. Short after the 6500 they also second sourced the Valvo/Signetics's 2650 (*3) and Zilog chips.

The Atari deal came right after the agreement with Synertek during WESVON 75 (*4) with the deal being quite helpful to make the sale. Just, the starting point here was again different from just selling a CPU. It was in fact Atari asking MOS - or more exact it was a matter of days between adds placed by MOS for their $25 chip, publications in trade papers in August and September 1975, the WESCON75 in September, a meeting at Atari with Jay Miner and the sale.

The story is in part described in Curt Vendel's incredible book Atari INC- Business is Fun (p.194ff) as well as in Peddle's and Mensch's oral history recordings. Two engineers at Atari - or more exact, Cyan, their 'engineering think-tank', Steve Mayer and Ron Miler, saw the articles and went to WESCON to meet Chuck Peddle. After some talks he agreed to come over the next day to Cyan,where he met Jay Miner.

It was the early days of the VCS development and at that time the Atari management intended have the whole system made of one (or more) custom chips. Something Miner wasn't really into. With the 6502 he saw a chance to reduce the design effort by buying the CPU/IO/RAM/ROM standard parts and only create the additional Video/Sound logic, the part that ultimately became the TIA.

Miner had a budget of 12 USD for what he wanted. Chuck Peddle introduced the idea of using the 6507 to save cost and make that price point.

[The whole story is a bit longer than that, so it might be worthwhile to read Curt's book]

The Motorola Lawsuit came after WESCON (and the Atari deal) it essentially killed MOS as a company of its own, as the settlement was tied to payments MOS could not bring up in time. The Atari deal might have played a huge role here with Motorola sales believing that they already had the deal as told in Brian Bagnall book Commodore: A company on the edge (*5)

Commodore bought MOS after the settlement for cheap, as they needed the money to pay off Motorola.

Commodore 'spun out' WDC to reduce cost and allow Mensch to continue development on his terms, but with a goal to create a new very flexible chip (*6) for future calculators which should enable quick implementation of new functions to reduce cost and time to market. Effort and development time ended up to long for Tramiel's taste, so he cancelled the contract, leaving Mensch and WDC going an independent route from there.

WDC managed to get Rockwell as second source, as they could target a more professional environment. In the end it helped WDC to keep afloat.

That's essentially the story of how MOS ended being a company on its own early on, already inbetween the first and second licencee (not counting commodore who ofc kept a licence to the chips with the companies assets, not names CSG.

Another reason I'm interested is because of the extreme willingness of MOS Technology to sell Atari the 6507 (a version of the 6502 with fewer address pins) at a steep discount from the already ridiculously cheap (by 1975 standards) $25 launch price of the 6502. To my way of thinking, it doesn't make sense to give that steep a discount when selling a product the prospective customer has to have, when you are already selling for a fraction of the price of your competitors (like Intel and Motorola in this case).

This seems like a very strange sense of business. Kind of a monopoly point of view, where it's about pressing the most money of the least sales due being the only source. Isn't it? In a competitive market - and microprocessors were one - it's about making a sale or someone else will do so. The only business questions to ask is

  • will that sale generate profit and
  • would it stop me from selling to others (at profit)

The answer is clearly

  • Yes, Atari kept them afloat with their huge numbers
  • No, all ​other customers were tiny

It's rather moot to speculate if this is a good or bad business practice, and what other school of sales and company management one should follow.

Beside all of that, it's a number game.

Keep in mind that the everywhere reported 25$ price was for single 6502 in a box including S&H (*7). Manuals were 5$ each. At least in October 1975 as a very nice example shows. That is single quantity end user price. Not selling a case of 100, 1000 or 10,000 at once.

Even today it's safe to assume that reasonable volume can half the price. Heck, let's look at a 6502 today. Mouser Europe asks 10.81 Euro for a single W65C02 while a 1000-pack goes for only 6,140 Euro. Or 6.14 Euro per chip, Thus single sales have a 76% markup from wholesale (*8). I guess it's safe to assume that today's chip manufacturing is rather sustainable.

And yes, MOS did consider volume prices like this excerpt from a ca. September 1975 brochure mentions for the 6501:

enter image description here

If a single 6502 was at 25 USD including books, packing and mailing, it's quite safe to assume that a profitable price of the bare chip was considerable less than 10 USD. Way lower. And the Atari deal was about: High volume of raw chips. No packaging, no boxes, no S&H no nothing. Just boxes and boxes of chips.

And volume it was. Everything ales at the time was peanuts.

  • Commodore delivered 500 PET in 1977 (*9)
  • Apple sold 600 Apple II in 1977
  • Atari sold >350,000 Units in 1977

MOS couldn't deliver enough 6507 to satisfy Atari's need, so they were quite happy about Synertek stepping up. It would be many more years before any computer usage could beat that - not to mention that VCS sales climbed quite steep, selling some 10M units until 1982 - in the US alone - more than double than all PET and Apple II combined.

Bottom line: At that volume, 12 USD per 6507/6532 pair was a quite profitable deal for MOS (now CSG) by whatever measurement one would calculate.


A Note about the assumption of Peddle being an "technologist, and we technologists always underprice" - he was always more the salesman of the team. With great social skills. He was the one who forged the idea to move and create their own chip, having the contacts to John Paivinen. He was the one doing most early sales (including personally selling the chips with his wife at WESCON. The is the connection between many of the actors around the 6502 and it's early success by talking to people and connecting them up. So not your classic techno-nerd. Ofc, it helped that he was an engineer, so he could talk and understand with all parties from Joe Decuir to Steve Wozniak.


*1 - As well supported by a 1975 MOS brochure mentioning only Synertek as second source, but not Rockwell.

enter image description here

*2 - John Paivinen was involved in another famous project later on, as he, Chris Fisch and Chuck Peddle founded in 1980 Sirius Systems Technology to produce the quite remarkable Sirius 1. It showed what a mass market 8088 system could have been - in Europe it gave IBM a hard time to establish their inferior PC.

*3 - The 2650 is an interesting point here, as Atari used it already in some of their coin ops before the 6507 or any other 6500. It's a very nice chip with a quite sophisticated instruction set. Notably following (or rather preceding) the same route MOS was going with their high integration peripherals. There was for example the 2656, a chip much like a doubled 6530 with 2 KiB ROM, 128 Bytes RAM and an 8 bit port. No timers but instead a mask programmable PLA, to it could generate custom specific functions - like doing all address decoding in a small system. Very nifty for video games. A BASIC system comes with just 3 ICs, 2650 CPU 2656 RAM/ROM and 2637 Video&Sound and could already include a basic game (much like the Vectrex did), saving on cost for the console which usually as sold with at least one game or some 'OS'.

So a really interesting question might be if Jay Miner had thought along the 2650 path and if yes, why he decided against? The 2637 could have done everything the Atari was intended to do

*4 - At that point history is a bit unclear as Peddle remembers the Syertek deal coming first, while other sources say it happened on behalf of Atari's wish to have a second source. Wither way, all seem to have happened in just a few days or weeks in September 1975.

*5 - Read it. Not as nerdy as Curt's collection but full of insight

*6 - They came up with a rather clever design of what they called a 'macro' processor, kind of a combination of a two level micro code execution scheme with the ability to add hardware with little effort on the same die. So the lower execution level would be kind of macro blocks of basic (mathematical) functions to be combined by high level code into more sophisticated functions presented to the user.

*7 - On WESCON the sale included data sheets but no box.

*8 - Not to mention that it may be even higher at larger volumes.

*9 - They collected prepayments for more than 1000, but production was lagging 'a bit'

16
  • 1
    @davidbak even today full-on CPUs are sold at retail (not wholesale) in boxes. And, yes: pagetable.com/?p=1295
    – hobbs
    Dec 8, 2021 at 6:49
  • 1
    Maybe "that step" rather than "hat step"? Dec 8, 2021 at 11:08
  • 1
    Who is "Joe Decatur"?
    – Brian H
    Dec 8, 2021 at 15:12
  • 3
    @davidbak Erm, not really. Back then trays were much more common for serious delivery than another package methods. That is when ordering several hundred or more. Trays made from plastic or wood (!), then packages in paper boxes (usually one per tray, but as well stacks of tray in a single box) is what packager/factorys delivered their products. Tubes/rails were only used for small quantities, usually by distributors - part of their job is taking chips from factory packaging and stacking quantities ordered into tubes. Today trays are still used for large devices, while smaller come in reels.
    – Raffzahn
    Dec 9, 2021 at 0:11
  • 1
    About the pricing, it's reasonable to estimate, if Mouser sells a part for $2 each in lots of 10,000, that if you can buy it direct from the manufacturer you'll only pay about $1. At least nowadays. Distributor mark-ups in 1975 were probably even higher due to higher costs (printing paper catalogs, more salespeople giving quotes over the phone, etc).
    – The Photon
    Dec 9, 2021 at 18:12
9

A good indication of the date and the terms can be found on the first page of Synertek's "SY6500/MSS6500 Microcomputer Hardware Manual".

The date of this publication is clearly indicated as August, 1976. Furthermore, it indicates it's a revised edition of the First Revision 1975 document from MOS. Since Synertek was already branding their own documentation of the 65XX family, it's safe to conclude the second-source deal was made prior - so, probably by Q2 1976.

The first page also gives details about this cross-licensing agreement between Synertek and MOS.

enter image description here

2
  • 1
    Perhaps even earlier than 1976Q2, since August is the end of Q3 (calendar year). Depends somewhat on just how long a successful end to the negotiations was anticipated...
    – Jon Custer
    Dec 7, 2021 at 20:12
  • I wonder how how much demand there would have been for a chip which was similar to the 74LS138, but with some tweaks to facilitate common address-decoding scenarios, such as having an input that would force all outputs high, but otherwise having unselected outputs float. This would make it possible to "wire-or" outputs together without external logic, minimizing the need for things like address decoding PLAs.
    – supercat
    Dec 8, 2021 at 17:35
4

As to the pricing bit of the question...

Generally, you can separate early MPUs into two groups, those that were intended to be CPUs and those that weren't.

The former invariably includes design notes from a minicomputer they selected as a pattern - the PDP-8 and Nova were the favorites here, less so the PDP-11 and even some IBM machines. The CPUs often offered multiple register sets and/or stacks, variations on memory-keyed indirect addressing, auto-increment/decrement loops, and all sorts of branching.

The latter tend to be much lighter. An example of an early design aimed at the controller space is the F8. Although it was not cheap, due largely to being an early 1970s PMOS part not due to any design feature. A complete F8 could be implemented with only two chips, the F8 (3850) and a 3851 RAM/ROM. This sort of lightweight implementation is what we would today think of as a microcontroller. The 3851 was later merged on-die, making it really easy to use.

So then there is the mystery of the 6800. Motorola's traditional customers were car companies and industrial users. Yet the 6800 is very much like a CPU design as opposed to a controller. It was Peddle that was sent out to sell it, and he received considerable pushback from long-term Motorola customers about the price. They invariably stated that the price of the 6800, about $100 in quantity, was far too much to be used in, say, a fuel injection system. They all stated that the part would have to be under $10 before they could consider using it.

So when he returned to Texas, Peddle began thinking about what would have to happen in order to hit that price point. He outlined a design, and when Motorola proved uninterested in the idea, he left. The explicit goal was a chip that could be sold in quantity at $5. This was not because "was a technologist, and we technologists always underprice", it was because his customers had set the price and he knew anything over that simply would not sell.

And so, out went the second accumulator, out went a bunch of branches, out went indirect addressing, out went 16-bit indexes. In came Micraline, in came NMOS, in came a 165 mil chip. And in came the world's first $5 CPU.

I'm sure he was as surprised as anyone when it turned into the most popular processor of its era.

1
  • 2
    The 6800 is a rather schoolbook like CPU - exactly what one would expect from a big company when creating a new product. Making it middle of the road is one of two strategies such companies follow - the other is continue what we have, which Intels direction (noteworthy that the 4040 is a way more interesting concept than the 8080).
    – Raffzahn
    Dec 9, 2021 at 18:30

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.