Re: Manufacturing New 650x - Need help

jac_goudsmit · Post by **jac_goudsmit** » Fri Dec 16, 2016 11:56 pm

Giebels2609 wrote:

I know... Therefore I'm going to produce rebuilds of commodore computers and produce all chips for them...
Second I will produce ROM, RAM and things of that nature

Even though Commodore doesn't exist anymore, there are still people and/or companies that own the Commodore brand and the Commodore intellectual property. Not to mention the code inside the Commodore ROM chips is under copyright for the next 100 years or so (probably longer, because Disney will keep lobbying governments to extend copyright expiration)

It's probably very difficult to generate enough demand for Commodore replicas to make a profit; C64's are still cheap and ubiquitous on eBay and elsewhere so why would anyone want a replica? The more serious you are about making Commodore replicas into a business, the more likely you will show up on the radar of the current Commodore IP owners, and you will get sued for doing it.

I wish you good luck, but I think you're underestimating the amount of work and overestimating the chances of success.

===Jac

barrym95838 · Post by **barrym95838** » Sat Dec 17, 2016 2:07 am

@Giebels2609: It took me less than 30 seconds to discover that your first name is probably Kevin, and you probably live somewhere in Belgium, but I didn't pursue it further. I don't know the extent of your finances and abilities, but I wish you the best, in whatever you decide to do. And your English is good, or at least good enough. ttlworks is correct about the subtle nuances of English causing minor problems for non-native speakers ... English is a huge language, full of diverse influences and irregularities, perhaps more so than any other language on the planet (human or otherwise, with the possible exception of x86 assembly language).

Mike B.

Arlet · Post by **Arlet** » Sat Dec 17, 2016 5:19 am

barrym95838 wrote:

@Giebels2609: It took me less than 30 seconds to discover that your first name is probably Kevin

It must have been the "Hello, my name is Kevin" at the top, right ?

barrym95838 · Post by **barrym95838** » Sat Dec 17, 2016 5:52 am

Mike B.

KC9UDX · Post by **KC9UDX** » Sat Dec 17, 2016 8:50 am

Arlet wrote:

barrym95838 wrote:

@Giebels2609: It took me less than 30 seconds to discover that your first name is probably Kevin

It must have been the "Hello, my name is Kevin"

https://youtu.be/WuUnS-h2PoU

I found him on the internet, too.

Oneironaut · Post by **Oneironaut** » Sun Dec 18, 2016 7:36 pm

6502 already has a replacement, but the SID does not.
A good chip to start with in order to build up some funds.

I would order 1000 units, if they were less than $14.95.
For real.

Brad

Giebels2609 wrote:

Yes, I know... My fab will be smaller than MOS Norristown... Just a small one

KC9UDX · Post by **KC9UDX** » Mon Dec 19, 2016 3:09 am

A wise thing to do if they genuinely mimic SID behaviour, rather than just digitally emulate it.

But of course, which SID would you want?

ttlworks · Post by **ttlworks** » Mon Dec 19, 2016 8:08 am

Some info about the innards of the SID can be found here:
viewtopic.php?f=8&t=4150&sid=8f45721ef2 ... a91e859642

Be warned that the SID is a little bit tricky... but maybe less tricky than a 6502,
since the chip layout for the three oscillator sections seems to be almost the same.

Good luck with trying to manufacture chips, Kevin.
It's not as easy as it looks, not to mention that the chip has to be put into a package later, see here:
viewtopic.php?f=4&t=4310&start=60&sid=8 ... 642#p48713

Maybe it would be a good idea to aim at a 74HC00 first for testing if your manufacturing process
leads to the intended results.

Oneironaut · Post by **Oneironaut** » Mon Dec 19, 2016 1:42 pm

With the kind of MASSIVE budget that would be required to produce an IC, it should be possible to lure the original designer of the SID back into the fold to help! When you are willing to throw millions of dollars into something like this, it will be important to have key people available. After reading the story of Commodore in the book "Spectacular Rise and Fall", perhaps you could even get Chuck Peddle (The Father of the 6502 and Home Computer) back in the fold! After all, Jack Tramiel robbed him of the credit and financial rewards he was so deserving of. Seriously... you will be dealing with millions of dollars if you intend to cut an IC, so why not surround yourself with people that can make it happen?

I remember reading someplace on the Parallax forum that the Prop was a 10 million dollar undertaking, but can't find the link. Here is a small discussion though...

http://forums.parallax.com/discussion/1 ... make-an-ic

Brad

Arlet · Post by **Arlet** » Mon Dec 19, 2016 1:50 pm

I'm guessing that $10 million doesn't include the fab, but is just the cost of the design, mask set, and a production run from an existing fab.

To operate your own fab would probably start at $1 billion, assuming you get some cheap used equipment...

Arlet · Post by **Arlet** » Mon Dec 19, 2016 2:51 pm

The cheapest fab I could find was the SUNY Nanofab 200 for $16 million start up cost in 1997, so yes, the $1b was a bit exaggerated. But 1 million at today's prices seems unreasonably low. And to become an Intel competitor, you'll need a few billion more. One of the more recent Samsung fabs cost $10 billion, for example.

Tor · Post by **Tor** » Mon Dec 19, 2016 3:15 pm

Arlet wrote:

I'm guessing that $10 million doesn't include the fab, but is just the cost of the design, mask set, and a production run from an existing fab.

Quite so. Parallax is fab-less.

Oneironaut · Post by **Oneironaut** » Mon Dec 19, 2016 4:50 pm

Looks like you are a man with a plan!
You will only need two more items...

http://www.professionalplastics.com/RUBYLITHFILM

http://www.leatherman.ca/en_CA/raptor-831742.html

You now will be retracing the footsteps of Silicon Valley circa 1980!

Brad

Giebels2609 wrote:

Arlet wrote:

The cheapest fab I could find was the SUNY Nanofab 200 for $16 million start up cost in 1997, so yes, the $1b was a bit exaggerated. But 1 million at today's prices seems unreasonably low. And to become an Intel competitor, you'll need a few billion more. One of the more recent Samsung fabs cost $10 billion, for example.

I'm going to use old machines from the 80s... Can get them cheap from east europe... And that with the intel competitor was assuming I hat the billion he said I would need.

Tor · Post by **Tor** » Mon Dec 19, 2016 6:54 pm

You'll also need masks for the 6502 etc.. I don't know if they are even available - visual6502 could have used the masks instead of having to de-cap a 6502 if so, maybe. If they are available (Commodore and MOS aren't around anymore) they would have to be licensed anyway.

kakemoms · Post by **kakemoms** » Tue Dec 20, 2016 8:56 pm

Giebels2609 wrote:

Tor wrote:

You'll also need masks for the 6502 etc.. I don't know if they are even available - visual6502 could have used the masks instead of having to de-cap a 6502 if so, maybe. If they are available (Commodore and MOS aren't around anymore) they would have to be licensed anyway.

Nope, I'm currently making preparations in CAD software for maikng new masks.

Try CLEwin. Its cheap, can handle many layers and for a 3500 transistor design you can probably do it manually within a couple of weeks. The only problem with that is debugging which becomes a manual thing.

For a more professional approach I would do the design in VHDL or Verilog (both of which have numerous cores available). Ask one of the people that has already made a core to license it. You can then translate that into a transistor-level design without even seeing a mask.

A small fab isn't very costly to set up, but it can become quite labor-intensive without the correct investments. I have once put up an in-house cleanroom with both litography, metallization and characterization equipment (it was for other things, not transistors). Using old equipment makes sense in the short-term, but for long-term you also need to consider maintenance, repair, engineering hours and having enough spare-parts for the machines to function. With 10 or 20 machines running, its enought that one of them stops working and your fabulous fab is down for hours or days. Fab down=no income.

A simpler option is to be a fabless design house. It means that you submit your design to one of the service companies of TSMC or alike. They don't want to talk to you directly, so you have to go through a service/design company that is already a customer of one of the large manufacturers (as TSMC is). These offer different ways to make prototypes or production runs, the latter requiring your own mask set (which will cost >100K USD for 150mm wafers). But with production runs you can make 65C02's for much less than in a small fab. Even a few thousand devices will be cheaper. Heck, they may even help you to put up a NMOS process if you really want that.

Anyway, good luck!