Does anyone know where to get a set of PROMS? I'm putting together a board and I'm not sure where to get a set.
Anonymous
User login
Please support the defense of Ukraine.
Direct or via Unclutter App
Active forum topics
Recent content
Navigation
No Ads.
No Trackers.
No Social Media.
All Content Locally Hosted.
Built on Free Software.
We have complied with zero government requests for information.
What PROMs are you looking for?
If you're building an Apple II motherboard and need the System ROMs, it depends on which motherboard you have and what monitor/BASIC you want to have resident.
Many different options out there, but for the most flexibility and probably lowest cost, take a look at ROMX.
I think he is looking for an Apple-1. That's completely different than an Apple II. I don't thinik ROMX will work for him although as I mention below, there are similar options. He's looking for Bipolar PROMs. Those are much more difficult and expensive to find and also difficult to find someone who can program them. Since there are no more Uncle Bernie chip kits, I don't know of any good sources of the chips pre-programmed. Places like Unicorn and some of the places that cater to arcade video game collectors carry blank chips. They're super expensive any more. Very few programmers and basically no modern ones can program those chips. The old programmers require a vintage computer to drive them because they usually plug in to a parallel port or something else that is no longer common and usually the software runs on something obsolete like MS-DOS.
For the Apple-1 you need both the motherboard ROMs, usually the WOX monitorl and usually you need a set fo for the Apple Casette Interface (ACI). For the ACI card I designed a small board to allow use of a 2716 EPROM instead of the two BiPolar PROMs. There is something similar out there for the main board as well that someone else on here designed. That may be the best choice these days.
Oops. You're correct. I didn't notice that this was posted to the Apple-1 group.
PM sent.
Indeed, I'm looking for the apple 1 PROMS with WOZMON burned in them. I've got some PM's so hopefully I can secure a set
Go to Eprompro - see below....
http://www.eprompro.com/
That looks like a pretty good resource.
It's too bad sometimes that there isn't a specialty chip maker who could come up with something like an EEPROM based replacement for the BiPolar PROMs like the MMI 6301 and 6309 (and the many compatibles thereof). Something easily programmed with a modern programmer but pin compatible with those old chips. Theoretically those are really simple chips and any modern fab that makes modern PLD chips should be able to crank them out in sufficient production by running a few in with runs of bigger volume chips. I know there are some "fabless" semicondctor companies that use contract fabbing, so it seems like the mechanics are there if someone could design modern replacements for vintage chips. The issue I'm sure is that if new production was available the price of the vintage chips would plummet and it wouldn't be profitable enough to sell those chips in the volume the market would bear which is mostly hobbiests and a few companies out there that would like to build re-issues of some vintage products (like some guitar effects, keyboards, etc). Or possibly even just companies that wanted to build simple ASICs other than using things like FPGAs (probably not many).
I guess I'm just wishing that modern technology would make this almost like what has happened with things like PCB manufacturing. 40 years ago designing and laying out a professional quality PCB, even something as simple as an Apple II card required resources beyond almost all hobbiests access. Now anyone can download free software like KiCAD and design a board, create Gerbers and send them off to a contract PCB fab and have really nice boards made. It's so simple even someone who is not really a hardware guy like myself can (and has) done it.
It would be awesome if there was chip design software like KiCAD is for PCBs that would let anyone design a TTL, memory, etc., chip from simple building blocks and then there was a way to upload design files to a fab that could make and package small quantities of the chips, package and mark them, combining multiple orders together like places like JLCPCB and PCBWay obviously do. I'm talking simple 1970s-1990s complexity chips... DIP-8 to DIP-40 or simple PLCC chips, etc.
I also know that most of these things are typically done these days using PLDs, primarily FPGAs these days. But that usually requires chip carriers and other messy and non-vintage looking things and some vintage chips that are highy in demand like some used in the Apple-1 and some vintage video game boards are apparently not easily re-created with FPGAs.
Anyway... probably a wild pipe dream, but maybe it will give someone much smarter than me an idea.
I've often read posts on enthuiast forums (or places like Hacker News, which tends to be on a similar level) comparing chip design to PCB design. They both place components together with wires, so what's the difference? But not many posters understand that producing a wafer is rate limited. That is, the materials that make up a PCB can all be produced from stock in minutes. But the steps to fabricate a wafer can take much longer and there are many more steps. The words "semiconductor process" take on a very particular meaning when some of those steps can take months.
There are PCB prototyping technologies like laser milling that can literally turn a board from concept to completed article in minutes, but the comparable techniques for chips—like e-beam pantography—never passed the research stage. So the only feasible technologies for making chips are the ones designed for mass production. Chip fabrication is a numbers game and prices are strongly related to production quantities.
Already pretty close to reality. Sam Zeloof for example who I follow on YouTube has been making his own silicon chips in his garage and I think at this point he has reached where technology was in the 70s. A true testament to the amazing time are all so lucky to be alive in, where one person can do what only could be done by a large company 40 years ago.
Here is one of his videos: https://youtu.be/IS5ycm7VfXg
That site is very cool.
I said I realize a lot of what I was talking about might be a pipe dream. And I certainly realize that producing semiconductors is much more complicated than PCBs. But the advances of technology should have similar effects of making things that were only possible for big and highly funded organizations to be done on a smaller scale by small business, hobbiest groups or even individuals.
And I also understand that commercial production of semiconductors has always been focused on mass production and building chips for pennies so they can make money. The hobbiest market would or could be different in that people are currently willing to pay sometimes up to hundreds of dollars for a relatively simple chip just because it is rare. And when it comes to hobbiests often people are willing to do work without expectation of fair compensation for it. People will make and share things not only with no compensation at all but often at significant expense on their own part in aquiring knowledge, tooling and materials. The rewards are often on the soft side. Creds, good will, community, rather than money.
The link to the guy doing semiconductor manufacturing in his garage is actually very encouraging to me. It seems like someone inspired by what he is working on tht was focussed on re-creating vintage chips instead of new designs might be able to produce small quantities of certain chips. I don't think that the kind of production using MOSFETs he is doing would be applicable to BiPolar PROMs, but maybe some of the other chips. One thing I had thought of for replacement of PROMs that might be possible is instead of making them programmable, to make something like mask ROMs which the program was in the etching. That way someone could make chips that would work as replacements for 6301 or 6309, but the programming would be baked in when they are made instead of burned in with fused elements.
The main chips that are pain points for the Apple-1 besides the PROMs are Signetics 2504, 2513, 2519 and DS0025. I don't really know a lot about those chips or how complex they are. Looking at the data sheets it seems like the transistor counts should be within the realm of possibility with the processes that guy is using. ut I could be completely wrong.
The advances in technology have exactly the opposite effect from what you suggested: new generations of fab technology currently have a price tag above $12 billion.
The only way that technology enables wildcat labs is by surplusing obsolete control equipment. The costs of the reagents is still the same or higher, the cost of environmental cleanup is the same or higher. I certainly would not want a garage semiconductor lab in my neighborhood (arsine, phosphine, diborane, what fun).
In post #9, robespierre wrote:
"The words "semiconductor process" take on a very particular meaning when some of those steps can take months."
Uncle Bernie comments:
None of the steps in a wafer fab takes months. Some take many hours, though. Some take mere minutes. So the trick is to stage the wafer lots such that ALL "tools" (the machines who do the "magic") are fully occupied all the time. So some lower priority wafer lots have to wait for days until they get the next process step. One company I worked for as an IC designer had "rocket lots" which would get preference and there you could get 350nm CMOS wafers in four weeks after masks having being made and moved into the fab. The normal process spin took six months. Need two unplanned spins due to bugs in the design, lose one year and lose the project.
Modern deep submicron processes may have 150 process steps (or even more, after all, it's modular processes now, you can combine process modules such as number of metals, thin film, high voltage transistors, EEPROM ... A typical rule of thumb for such a process is three process steps per day, so expect 50 fab days to get your wafers out.
But this is state of the art stuff. Simpler processes like 0.6um 1P2M CMOS could be run in 2 weeks to the finish line. But the problem is, these vintage processes are hard to access. The big semiconductor corporations still make a lot of their bread-and-butter ICs on these vintage processes, but foundries usually don't offer them anymore, as there is no profit to be made with them.
In post #12, softwarejanitor wrote:
"The main chips that are pain points for the Apple-1 besides the PROMs are Signetics 2504, 2513, 2519 and DS0025. I don't really know a lot about those chips or how complex they are. Looking at the data sheets it seems like the transistor counts should be within the realm of possibility with the processes that guy is using."
Uncle Bernie comments:
I've seen this garage fab guy on youtube and applaud his work, it's really amazing, but don't expect to be able to make any MOS transistors that way which would last. "Mobile ions" are the problem, Sodium being the worst, and even if you have some MOS transistors that work initially, they will die one by one after a supply voltage is applied and the mobile ions start moving due to the electric fields. But I can see that a simple bipolar process could be done in a garage. Although the yield would be terrible. You probably could make a DS0025 with it, though.
ON APPLE-1 IC COMPLEXITY
None of the "critical" ICs in the Apple-1 (2504, 2513, 2519, DS0025) are complex. I probably could design the circuit and the masks for all of them in maybe 2 weeks. But there would be no foundry available to make them. Here are some details which may be of interest to builders:
The DS0025 actually began life in the 1960s as a hybrid called MH0025 in a metal can (only 4 NPNs and 4 diodes on a ceramic substrate with the thick film resistors) which then in the end of the 1960s or early 1970s was integrated as a single chip solution. The inner circuit is in the 1979 "INTERFACE DATABOOK" from National Semiconductors.
HOW TO USE DS0026 in lieu of DS0025
If you put my complete "reliability mod" in your Apple-1 build you can use the later DS0026 which still is available (most "DS0025" the Chinese counterfeiters sell are re-labeled DS0026, unless the fraud is worse and they are modern CMOS parts).
If you use DS0026, you may get away with less effort than the complete "reliability mod" if you put 39 Ohms series resistors in the PHI3, PHI4 output circuit of the DS0026.
OTHER DS0025/DS0026 SUBSTITUTES
If DS0026 would ever run out, there are plenty of industry standard CMOS based power MOSFET gate driver ICs which have the same footprint, and could drop right in, but in the Apple-1 they would run a tad above their abs,max power supply voltage. So you might need to reduce the -12V rail to -9V to be safe: swap the 7912 for a 7909. I experimented with that and the Apple-1 ran fine. The series resistors always were needed, though. Otherwise the switching action kicking into the power supply rais would crash any running Apple-1 program.
THE 2504, 2513, 2519 "UNOBTANIUM" ICs
The 2504, 2519 are trivial dynamic shift register ICs which could be designed in a few days, circuit and layout. More modern CMOS processes are not suited for 1970s style dynamic logic, though, but static shift register cells are easy enough.
The 2513 is a ROM, also trivial to design. The tedious part is to put the bit pattern in. Open source IC design CAD tools like MAGIC don't automate this. If you work with Cadence and a good foundry you get ROM and RAM generators in your design kit. But the Cadence CAD suite is out of the financial reach of individuals.
PROMs
Fuse link PROMs could be designed using 1k/sq SiCr thin film resistors, but these need a process option with MOSFETs able to run at 12...15 Volts. For the NMOS you could design your own high voltage MOSFETs, if the design kit allows to draw thinox without invoking any diffusion implants, but for the high voltage PMOS you would need a BiCMOS process which has a suitable pbase implant normally used for the NPNs. These tricks usually work down to the 0.6um node. (Done that, these PROMs work fine, and, ironically, use a smaller area than the EEPROMs offered at that node, unless it's a specialized EEPROM technology).
Again, access to suitable process is the key, a key we don't have. Thin film resistors on a foundry process are rare. TSMC started to offer them just a few years ago, but I don't know how good they are and if they don't corrode away after a few years.
Fuse link PROMs also can be made using polysilicon filaments, but depending on how the particular CMOS process builds the gate stack they may be unreliable, have poor programming yield, or growback issues. "It only works if it works".
In post #13, robespierre wrote:
"I certainly would not want a garage semiconductor lab in my neighborhood (arsine, phosphine, diborane, what fun)."
Uncle Bernie comments:
The amounts needed for the toxic process chemicals you named are microscopically small. Unless large quantities are recklessly stored on site, they pose no serious risk. The elephant in the room are the nasty solvents to clean the wafers and to strip the photo resists. And the HF etch bath also is nasty stuff. Older fabs used to vent the fumes from the wet etch aisle into the atmosphere, this is why you could tell the direction to the wafer fab from the side of the rust on the street light poles. Horrible environmental pigs.
BEWARE OLD SEMICONDUCTOR FAB LEGACY
Old, spooky, long abandoned wafer fab sites often are a huge environmental problem. The first thing they did put in when building a new wafer fab is a huge underground toxic waste tank. Then the concrete slab for the building was poured. And then the "Semiconductor Corp" building was erected, complete with the cleanroom for the wafer fab. When the toxic waste tank was full, the wafer fab was technically obsolete anyways and the site was abandoned. At this point, huge profits had been made, and the taxpayer had to pay for the cleanup which - often - never happened. Many are "Superfund" sites now.
The old MOS Technology site in Norristown, PA, is one of those. Their toxic waste tank developed cracks / leaks soon after the fab was started in the late 1960s, which conveniently allowed them (and later, Commodore Semiconductor Group, "CSG") to run the on site wafer fab longer than initially projected, because the toxic waste tank never seemed to fill up, due to the leaks. The ground water table was poisoned, though.
A problem which also plagues Silicon Valley until today. The root cause for this is greed - the amount of toxic waste from a wafer fab is so small that it could have been packed in a barrel and when full hauled off site to an incinerator of the type they use to destroy chemical weapons (high enough temperature needed). But - the toxic waste tank was cheaper, and the stock holders laughed all the way to the bank. In Capitalism, it's always the same pattern: privatize the profits, and socialize the losses. Communism is no solution either, as it is much worse: here are no profits, only losses, only crappy products, and everything is turned into a toxic waste dump anyways.
But that toxic waste issue from semiconductors is nothing compared to what is going on in nuclear sites like Hanford, WA. I never ate salmon caught in the Columbia river for a reason. Again, the Communists were worse environmental pigs: their Majak nuclear plant at Chelyabinsk-40 (now Ozyorsk) used to dump their nuclear waste into open storage basins linked to the nearby river. Look up the "Kyshtym disaster" for more creepy details. Compared to the nuclear industry, the semiconductor industry is environmentally friendly, more like a candy factory - think "Willi Wonka".
- Uncle Bernie
I remember reading an article in IEEE Spectrum that referred to a 30-day processing period.
That seems to be what is quoted in this dissertation from 2007.
Are you referring to this line in the article?
"The fabrication of a modern IC in the CMOS process involves hundreds of sequential steps and can last up to 30 days of processing time"
That is just referring to the total time being up to 30 days, not any of the individual steps. If they used 'which' instead of 'and' that could be open to interpretation of a different meaning.
Justin.
There are ways to recreate old chips that can be done without setting up everything in your garage. These are either smaller foundries or R&D labs that charge per hour or day.
Litography for example isn't that expensive anymore. I can go into the local lab, do spinoff/baking of photoresist at $50 per hour, then exposure at $150. I don't even need a mask for it as its a maskless aligner.
Then you have polysilicon deposition, dry etch, Silicon nidride deposition, metallization. All quite established processes. Calibrations are needed, which is probably were most of the cost would be.
All-in-all I would say that a 4inch wafer would set you back around 15-20K USD, but if doing 25 wafers it would probably go down to around 5K USD per wafer.
This is all without designing the actual layout.
Depending on how many usable chips can be cut from a wafer, and what chips you are taking about, it could be an interesting project for someone well heeled enough to fund it and with connections to people who could do the work if they don't have the expertise themselves.
The other thing then would be what it would cost to cut and package the chips into a usable form like a DIP package.
There are a number of highly sought after vintage chips that would be cool to reproduce. I think most of the ones that are expensive are also fairly complex though so designing them would be something few could do. And I'm not sure what yeild would be, at least initially. I'm pretty sure someone like Uncle Bernie could do the design work, but I don't know how much foundry floor experience he has actually doing the fabbing, that's a while other set of skills. I know someone who studied that kind of thing at A&M, but he graduated back in the 1990s so even though they did actually fab chips in his classes from what I understand, that was 30=ish years ago. And he's worked in software since then as far as I know so he might not even be interested. Probably not and probably too busy anyway.
The other thing is if a new source for some of these rare chips came online the value would probably plummet. A lot of the high $$$ is because of greedy hoarders as Uncle Bernie has talked about before.
But anyway... super cool idea none the less.
Kakemoms wrote:
" There are ways to recreate old chips that can be done without setting up everything in your garage. These are either smaller foundries or R&D labs that charge per hour or day. "
Uncle Bernie comments:
This is wishful thinking. "Smaller foundries" don't exist anymore, there once were companies like Orbit and others, who tried that, but it was not viable. You can't run a wafer fab, any wafer fab, without a base load, meaning X wafer starts per week. And "X' is in the 1000's. The reason is that all the "tools" must run 24/7. Just consider what happens if you turn the diffusion furnace off. You can't shut down a fab for a longer time and then hope to be able to turn everthing on and keep going. Not gonna happen.
Some universities have simple wafer fab where they can teach students how to run a typical semiconductor fabrication process. Just to teach them the basics, using long obsolete equipment. They can make wafers. But the yields are abysmal (been there, used such services, never again).
It's called "foundry" and so people get misleaded into thinking it's like the shop of a blacksmith, you walk in, fire up the furnace, and then start forging metal by means of hammers (or forging / shaping silicon).
It's not like that. Making semiconductors (especially MOS) is truly high tech and always has been. You need the pure materials and chemicals, you need the very complex tools and a team familiar with them to maintain them, you need a lead process technologist (who knows his fab and the processes running there), you need device designers who will design the devices (*not* trivial ! Need a Ph.D. in semiconductor physics and lots of experience) and tell the lead process technologist about film thicknesses, ion implantation parameters, etc.) and then you need the operators who push the buttons on the tools and move the wafers around or dunk them into etching baths (in cutting edge fabs, this manual work is automated, no human ever moves a wafer). All this is a huge organization involving 100's of people. The IC designers themselves only design the electronic circuits and the mask layouts. No IC designer could design a competitive device from scratch, or the process steps to make them, or build or run a fab, or maintain the "tools" in the fab.
It is called "High Tech" for a reason. If you factor in the whole supply chain from the silicon ingots to the "tools" for lithography, ion implanation, plasma etching, and the backend with the testers, bonders, and encapsulation, it's 10,000's of people working in that industry just to make it happen that a stream of ICs in tubes come out in the end. It is impossible to do that in a smaller nation. Not enough brains / engineers /technicians to be had.
Forget about the SciFi tales of crazy scientists building electronic marvels in their lab from the ground up. Not gonna happen. And the problem today is that the entry costs of anything involving cutting edge process technologies are so huge that nobody can start a semiconductor company anymore. Instead, the big players hoover up the smaller ones. Until only a handful of big players are left. But a small, innovative startup has zero chance to compete in this field. Heck, I have innovative analog circuits which are 10 x higher performant than anything the industry has, and could not find venture capital for my own semiconductor startup. And these naysayers are right - there is simply no way to make such "marvels" in small numbers and keep the doors of a business open. 99% of the market is covered by inferior crap. No need for top performance. Or do you need an ADC accurate to 1 ppm ? Most people won't. The $10 multimeter from Harbor Freight will do.
- Uncle Bernie