7400 series logic availability

[wbooth.clearscene]

Hi,

I have a design that calls for a 74189.
I tried to source the components in LS TTL, but half the chips were not available, and the 74LS189 was really expensive (from a dodgy eBay seller)
So, I migrated to HC CMOS. I've found all the required chips except for the 74HC189, which I cannot find for love nor money.
Does anyone know where I could source 4 (or ideally 6) x 74HC189 (or 74HC219) ?

Cheers.

[tggzzz]

So, you think your solution is a 74*189.

You will get a better response if you detail your problem, and why you think a 74*189 is a solution.

[tszaboo]

74189 64-bit (16x4) RAM, inverting three-state outputs
Not all 7400 logic is made equal. While 74HC02 will be probably made in the 22 century, most of the 7400 series is obsolete now, with no manufacturers.

[wbooth.clearscene]

This is a pet project to build an 8 bit computer from 7400 series logic. 

[tggzzz]

OK.

I predict you will have a flatter forehead and dents in various walls before you have finished.

Or maybe I could sell you some of my random pieces of old TTL at excessively inflated prices

[Ian.M]

So you want to build a SSI/MSI logic 8 bit processor.  That's a concept that became obsolete in the late 70's so its not surprising that parts are discontinued and difficult to find forty years later.   

Expect to spend a lot of time and money sourcing N.O.S. parts from dodgy sellers, then testing and qualifying them to determine if they are genuine, functional, and still meet original specs.

Jameco still claim to have stock of 74189, and they are more reputable than most:
http://www.jameco.com/z/74189-Major-Brands-Tri-State-64-Bit-RAM-DIP-16_49883.html.  However its going to be a PITA interfacing original TTL to 74HC CMOS logic - you'll probably need buffers to drive the TTL and pullups or 74HCT buffers to fix up its output levels.   

You've already compromised on CMOS, so why not go the whole hog and implement the processor in a FPGA?

[tszaboo]

This is a pet project to build an 8 bit computer from 7400 series logic. 

Getting a girlfriend/wife is a very effective way to solve these problems.

[rob77]

no-one is making the 74189.. you can tell that by  the fact the datasheets are scanned bitmaps

go back to drawing board and re-design your circuit to use a different mamory.

if you're just building some retro stuff , then why not mimic that 74189 memory chip with a small FPGA or if it's slow then even with a microcontroller ?

[grumpydoc]

Unless you really need the split data in/data out just use the smallest SRAM that you can find - likely to be 32k x 8 but you can just ignoreany data or address lines that you don't need (tie to ground).

Some of the 74 series never had good availability

[Cerebus]

go back to drawing board and re-design your circuit to use a different mamory.

Left or right? 

[Cerebus]

Unless you really need the split data in/data out just use the smallest SRAM that you can find - likely to be 32k x 8 but you can just ignoreany data or address lines that you don't need (tie to ground).

I don't know the specifics of the design that the OP is using, but it's quite common for register files to simultaneously read and write in the same CPU clock cycle on different clock edges and so the separate in and out lines may be necessary. I can't remember if the memory in question supports this, but quite a few of the TTL register file RAMs were dual-ported too, so you could also read and write different locations in the same cycle.

[grumpydoc]

Unless you really need the split data in/data out just use the smallest SRAM that you can find - likely to be 32k x 8 but you can just ignoreany data or address lines that you don't need (tie to ground).

I don't know the specifics of the design that the OP is using, but it's quite common for register files to simultaneously read and write in the same CPU clock cycle on different clock edges and so the separate in and out lines may be necessary. I can't remember if the memory in question supports this, but quite a few of the TTL register file RAMs were dual-ported too, so you could also read and write different locations in the same cycle.

Mouser stock the 74LS670 4x4 register file which might help

[james_s]

I bought some 74189's from an ebay seller several years ago to repair a classic arcade game, actually I think it was more like 10 years ago. Anyway they took a while to arrive but worked just fine. You might try Jameco, they tend to have a good selection of random old ICs.

[grumpydoc]

I bought some 74189's from an ebay seller several years ago to repair a classic arcade game, actually I think it was more like 10 years ago. Anyway they took a while to arrive but worked just fine. You might try Jameco, they tend to have a good selection of random old ICs.

Littlediode have some 74F189's, though the £13.49 each price might make your eyes water a bit.

[james_s]

64 bit RAM is simple enough that it could probably be implemented with flipflops in a cheap CPLD.

[tszaboo]

64 bit RAM is simple enough that it could probably be implemented with flipflops in a cheap CPLD.

He can build the entire thing in a Spartan 3 FPGA. That would result in some actual usable knowledge.

[kripton2035]

just got a box of really old things from a dead electronician shop ... a full storage unit of 74xxx but sorry no 74189 !

[westfw]

I see quite a few 74xx189 on eBay for much less than the price you mentioned.  http://www.ebay.com/sch/Business-Industrial/12576/i.html?_from=R40&_nkw=74+189

It was never a cheap chip; I found them in the online Popular Electronics archives for about $9 each (in ~1970s dollars!)

You didn't expect building a computer from SSI/MSI logic to be CHEAP, did you?  There's a reason that everyone jumped on the microprocessor bandwagon!

[chris_leyson]

You could dig quite a few SN10144 256x1 static RAM chips out of an old TEK 7D01 logic analyser plus a fair amount of ECL.

[alanb]

I have been on the same quest for 74LS189. I have recently bought 2 pieces from Cricklewood Electronics Ltd for £9.00 each + Vat and postage. The cost has prevented me from buying more. I have been unable to find any from a reputable source for less money.

[Cerebus]

64 bit RAM is simple enough that it could probably be implemented with flipflops in a cheap CPLD.

He can build the entire thing in a Spartan 3 FPGA. That would result in some actual usable knowledge.

He could just be doing this for fun, you know? Because it's a challenge? Perhaps he prefers soldering to coding in an HDL? Not everything in life has to have a practical or instructive outcome, otherwise the gods wouldn't have invented beer or Rugby football (or, for our antipodean friends, Aussie 'no rules' football).

I'm sure Sir Edmund Hillary could have got to the top of Everest more quickly and more easily using a helicopter. But I suspect that had you suggested it to him you would have received short shrift and possibly even a black eye. Had he done so, no doubt he would have gained some actual usable knowledge of flying helicopters, but we wouldn't be calling him Sir Edmund. Note that, even though it has been done many times before, many people still risk their lives to climb Everest (approximately 1 in 6 who make the attempt die*).

Finally, if one was going to use an FPGA, something a tad more recent that the Spartan 3 might be in order, or were you trying to change the challenge from using vintage SSI/MSI logic to using a vintage FPGA?

*The risks of constructing a CPU out of 7400 series logic are less. Probably limited to the odd burnt finger and occasional agony as you step onto a dropped 14 pin DIL package with bare feet.

[retrolefty]

Depending on the features and functions you want, as reference, I worked field service on TTL minicomputers in the 70s. There were many different manufactures, DEC, Data General, HP, Varian Data Machines, etc. They generally had a processor board around one foot square or larger, so very high chip counts. The instruction decode section is the hardest part depending on the complexity of the machine instruction set desired. Some used microprogramming (via high speed ROM) to implement decoding. Pretty large +5vdc high current power will be required.

 It's a very ambitious projects, but others have done the same.

Good luck.

[brucehoult]

I'm sure Sir Edmund Hillary could have got to the top of Everest more quickly and more easily using a helicopter. But I suspect that had you suggested it to him you would have received short shrift and possibly even a black eye. Had he done so, no doubt he would have gained some actual usable knowledge of flying helicopters, but we wouldn't be calling him Sir Edmund.

I'll just quietly note that he'd have to have waited a while. The first helicopter landing on Everest wasn't until 52 years later, in 2005.

He didn't turn up his nose at driving a regular Massey Ferguson farm tractor to the South Pole.

And he flew to the north pole in a small plane (Twin Otter) with Steve Fossett and Neil Armstrong.

[tggzzz]

Depending on the features and functions you want, as reference, I worked field service on TTL minicomputers in the 70s. There were many different manufactures, DEC, Data General, HP, Varian Data Machines, etc. They generally had a processor board around one foot square or larger, so very high chip counts. The instruction decode section is the hardest part depending on the complexity of the machine instruction set desired. Some used microprogramming (via high speed ROM) to implement decoding. Pretty large +5vdc high current power will be required.

It's a very ambitious projects, but others have done the same.

All those can be minimised by using a bit-serial implementation, at the expense of speed and with use of imagination.

Or he might be able to learn from the ACC's Weeny Bitter computer, see http://www.smrcc.org.uk/members/g4ugm/acc.htm Vol 3 issue 4 and following.

[Tom45]

Building an 8 bit computer using 74xxx chips is a walk in the park compared to the MegaProcessor:
 
  http://megaprocessor.com/index.html

In either case these projects are a personal journey and who are we to say they shouldn't be attempting it.