Interest for a C64c (250469) PLA replacement?

eslapion

Well-known member
Joined
Jul 31, 2015
Posts
1,557
Country
Canada
Region
Quebec
As some of you may have noticed, over the last 2 years, I have invested time and energy developing replacements for a few custom chips used in Commodore 8 bit computers.

These include:
- PLAnkton which replaces the C64/SX-64 programmable logic array 906114-01 or 251064-01
- TOLB which replaces the 7701/8701 clock generator used on later C64 boards, the C64c and the C128
- GandALF which replaces the 1541 custom controller and array 325572-01

I am presently considering the possibility of making PLA replacements for the C64c board 250469, specifically part no. 251715-01. Board 250469 can use 2 different types of PLA which incorporates most of the glue logic ICs used in the original C64 and they are part no. 251715-01 and 252535-01. Making PLA 252535-01 may prove too difficult because it incorporates the 512x4 bits color RAM and this type of chip is no longer in production.

So far, I did not make this substitute strictly because there doesn't seem to be much interest, not because there are any technical hurdles I can't overcome. This variant of the PLA is based on more modern technology than the one employed on 28 pin PLAs and it doesn't produce any discernable heat so it fails far less frequently.

I would like the people of Amibay to indicate if there is, in fact, sufficient interest to make a substitute for PLA 251715-01 used on board 250469 and how much people would expect to pay for this replacement.

Thank you all for your feedback.

- Francois
 
Last edited:
I am interested in replacements for practically all CBM custom chips. I don't know what the price for this particular PLA should be. 20? Still - if I may suggest, the chip that I'd LOVE to see having a reliable replacement is the 6526 CIA. You also will probably get more interest and more sales as those are known to fail very frequently while no replacements are available. That doesn't mean that the FAT-PLA is not interesting/important. It's more about priorities. Good Luck in any case. And if you can point me to your other designs, I'll be happy to have a look at them too.
 
I am interested in replacements for practically all CBM custom chips. I don't know what the price for this particular PLA should be. 20? Still - if I may suggest, the chip that I'd LOVE to see having a reliable replacement is the 6526 CIA. You also will probably get more interest and more sales as those are known to fail very frequently while no replacements are available. That doesn't mean that the FAT-PLA is not interesting/important. It's more about priorities. Good Luck in any case. And if you can point me to your other designs, I'll be happy to have a look at them too.
Thank you for your post.

I have tried and failed to make a CIA replacement. It requires more than 200 macrocells to be complete. It would cost in excess of 50$US per unit if I ever made and sold a substitute, an unreasonable price tag.

By comparison the 1541 Gate Array Logic, which I replicated in GandALF, requires about 50 macrocells and is implemented in the XC9572XL which carries 72 MCs.
 
I have never heard about someone finding a dead PLA chip to be the cause of a short board not working. Of course they are not invincible, but they do seem much more robust than their old counterparts, so I would imagine that the market for those is quite small.
 
I have never heard about someone finding a dead PLA chip to be the cause of a short board not working. Of course they are not invincible, but they do seem much more robust than their old counterparts, so I would imagine that the market for those is quite small.
From an electrical point of view, they aren't more robust, they just don't operate in 'toaster oven' mode so they don't overheat.

. I don't know if it's worth it at this point... it's obviously not going to sell thousands of units like PLAnkton did.
 
Last edited by a moderator:
I'd probably buy a couple just to have them on hand even if I never have an actual failure. :)

Heather
 
I am interested in having one or two new PLA chips for the short board as I have a couple of C64's with these and you never know when they will die. I have a few spares of the normal PLAnktons for just a situation for my remaining C64s and SX-64. It's just one of those things that people won't be interested until their system develops a fault and the part is required.
 
It sounds nice.
But I don't believe that you will have better results than Frank Buss.
 
Last edited by a moderator:
I have never heard about someone finding a dead PLA chip to be the cause of a short board not working.
You just did. But I probably don't count as regular user. I ran a repair workshop back then and therefore a huge number of broken machines passed through my place. But yes, I agree. This chip was one of the least failure prone.

- - - Updated - - -

I have tried and failed to make a CIA replacement. It requires more than 200 macrocells to be complete. It would cost in excess of 50$US per unit if I ever made and sold a substitute, an unreasonable price tag.

By comparison the 1541 Gate Array Logic, which I replicated in GandALF, requires about 50 macrocells and is implemented in the XC9572XL which carries 72 MCs.

I see. And I find it a true pity.

Yet I still wonder - people implemented the whole (well, close to) C64 and sold for a similar price. Maybe there is still something that could be done differently in order to keep the price at a reasonable level?
 
My impression is the 64-pin chip used in C64C almost never fails. Plus it's a pain in the butt to desolder and remove 64 pin chips that have smaller than standard 0.1" pin spacing. If the replacement clone is made available, I'd get one just in case. Older PLA used for most of C64's life are much easier to reproduce and replace, which is perfect since original PLA blows easily.
 
... people implemented the whole (well, close to) C64 and sold for a similar price. Maybe there is still something that could be done differently in order to keep the price at a reasonable level?
Other than the DTV which was produced in a few hundreds of thousands of units, most 'complete' implementations of the C64 such as the C64 Ultimate (Gideon Zweijtzer) use FPGAs instead of CPLDs.

The FPGA can provide a much greater amount of logic circuitry per $/Euros when compared to a CPLD but it has one very serious drawback, it loses all it's information when turned off and it has to be reloaded from some sort of memory at power on.

Now, that's not too bad if the whole machine has to be reloaded but it's a big problem if the FPGA is actually on a component inside an original C64 which expects all of it's components to be operational immediately at power-on. The PLA and CIAs are among these components. The 1541 Ultimate I/II/II+, being cartridges, can afford a few hundreds milliseconds of inoperative time at power up because they can reset the C64 once the FPGA is loaded with it's critical data.

The FPGA system also makes it easier to implement upgrades after the product has been sold.

- - - Updated - - -

It sounds nice.
But I don't believe that you will have better results than Frank Buss.
Frank Buss did a number of serious mistakes in his design. The most obvious one is the presence of 74_245 buffers which are totally unnecessary.

The XC9572XL on which he based his design is 5V input tolerant and it can also operate at voltages up to 3.6V which makes it generate signals which are fully CMOS/TTL levels compatible.

If I was to make a substitute for the 251715-01 64 pin PLA, it would be exactly the same size as the original chip. Not some extended board which may pose a conflict with double SID boards and others.
 
Last edited:
Might be worth it to create it after all. If only for a few dozens.
 
Last edited:
I have tried and failed to make a CIA replacement. It requires more than 200 macrocells to be complete. It would cost in excess of 50$US per unit if I ever made and sold a substitute, an unreasonable price tag.
You could use XC95216/XC95288. They are available for about $10 and a lot less in quantities over 100.
Or use an FPGA and a loader chip like EPC1064
If you are looking to recover your costs/time, the 6526 is a better project and you'll certainly sell 1000's.
I'm sure everyone here supports making a new 6526 so please give it serious consideration :)
 
You could use XC95216/XC95288. They are available for about $10 and a lot less in quantities over 100.
Or use an FPGA and a loader chip like EPC1064
If you are looking to recover your costs/time, the 6526 is a better project and you'll certainly sell 1000's.
I'm sure everyone here supports making a new 6526 so please give it serious consideration :)
Both of the chips you mention are now out of production so I would only be able to buy surplus chips from eBay or elsewhere. They also don't operate at the proper voltage for true TTL level signaling.

Often, the surplus CPLDs and FPGAs components sold on eBay are rebranded. For example, the XC9536XL I have found on eBay are all rated for 5ns but they also happen to have a very smooth surface because they are actually 10 ns chips on which the surface was sanded and rebranded. You can even see the typeface and logo are smaller than the real chips.

I have a very strict policy of never using counterfeit or re-branded circuits.

Even if the XC95216 and XC95288 (non XL) were still in production, their physical package (PQFP and MBGA) are way too large to even think making them a usable size for a substitute of a PDIP 40. The smallest XC95216 is 28x28mm while the PCB for a sub 40 PDIP is 18mm wide.

The presently available XC95288XL's smallest package is 144 pins TQFP which is 20x20mm and the pins extend an extra 1mm on each side. On PLAnkton and GandALF, the TQFP 44 ICs are 10x10mm with pins extending 12mm wide.

Using an FPGA with a loader chip often results in a C64/C128 just crashing as this chip is responsible for part of the memory management on these computers. It must perform the instant the CPU reset is finished.
 
Using an FPGA with a loader chip often results in a C64/C128 just crashing as this chip is responsible for part of the memory management on these computers. It must perform the instant the CPU reset is finished.

Would it be possible for this board to just override and hold the reset signal active until it is ready?
 
Last edited:
Using an FPGA with a loader chip often results in a C64/C128 just crashing as this chip is responsible for part of the memory management on these computers. It must perform the instant the CPU reset is finished.

Would it be possible for this board to just override and hold the reset signal active until it is ready?



pretty sure the reset line is intergral to the c64c board and goes to all parts that require it at the same time on the board,you would have to bypass the reset line where it begins i would imagine for it to wait for the pla by holding it low for longer

EDIT: which would make its own problems i think
 
Last edited:
Using an FPGA with a loader chip often results in a C64/C128 just crashing as this chip is responsible for part of the memory management on these computers. It must perform the instant the CPU reset is finished.

Would it be possible for this board to just override and hold the reset signal active until it is ready?

pretty sure the reset line is intergral to the c64c board and goes to all parts that require it at the same time on the board,you would have to bypass the reset line where it begins i would imagine for it to wait for the pla by holding it low for longer

EDIT: which would make its own problems i think
My comment about not using FPGAs was in response to fordav1 concerning the making of a substitute for the CIA. The CIA is used in the C64, C64c, C128, C128D, the 1571 and 1581 drives too. In almost all cases the CIA must perform immediately after power up and this has nothing to do with PLAs.

In the 1571 in particular, the CIA manages the modes of operation including the CPU speed.
 
Last edited:
Yes, the PLA is not even connected to the reset signal, however after looking at the schematics, I still think that an FPGA based CIA should be able to just hold RES low until the FPGA is done loading from flash and ready to go. The RES signal on CIA pin 34 is directly connected to the CPU RST pin, and the reset circuit is an open collector output so any connected device is allowed to pull it down.
 
Both of the chips you mention are now out of production so I would only be able to buy surplus chips from eBay or elsewhere.
Often, the surplus CPLDs and FPGAs components sold on eBay are rebranded. For example, the XC9536XL I have found on eBay are all rated for 5ns but they also happen to have a very smooth surface because they are actually 10 ns chips on which the surface was sanded and rebranded. You can even see the typeface and logo are smaller than the real chips.
I could buy about 50000 of those chips right now for probably $5-$8 each. Yes they are bigger, but that might be the only way to go.
As far as counterfeit etc, I've purchased several hundred ICs from this company and everything works fine. I bought almost all of the parts from them to build a GBA1000 and it is working fine. They have a return policy also so if anything is found to not work you can return it for a refund or exchange, although in 10 years of buying from them I have not found any dodgy chips yet. The company is utsource and parts for the kipper2k GBA1000 build + PicassoII + 68060 turbo board were also bought from them by "Tuxbar81". This company is very serious about supplying quality parts in small or large quantities.
If you do some research I'm sure you can find a chip that will do the job. It seems like you are just making up excuses why it can't be done, when you should be positive about it since it is a win-win situation.... you recover your investment in time many times over by selling 1000's of chips and C64 owners get chips that are getting harder and harder to find. As far as I can see there are no reasons to not do it. If a 6526 is re-made, I'm sure everyone who owns a C64 would buy several of them (the 6526 related comments above also back that up) and that's a guarantee of a successful project.
As I said above, please think seriously about making a replacement 6526. The entire Commodore community will thank you :)
 
Last edited:
Back
Top Bottom