Not quite... The end of the ROM based PLA for the 64

[craftsman1234]

After reviewing Skoe's extensive document, it appears my claims about the C64 drawing less power when using the M27C512-90B6 were right but not for the reasons I expected.

It is the M27C512-90B6 itself that takes about 90mA less than an original PLA.

This nice little chip also worked better than original PLAs on a few SX-64 machines as outlined by craftsman1234 on Lemon64 (Supermarket section).

I still have about 60 chips left after acquiring a few more.

Yes, it did work great for the SX and 64 that I am using them in. I need to get 2 more I want to try it in my other SX and have one as a spare. Will it also work in a Vic-20?

[Mike]

Will it also work in a Vic-20?

The VIC-20 has no (single chip) PLA to generate the various chip select and other timing signals, that job is done by several TTL chips.

[craftsman1234]

Will it also work in a Vic-20?

The VIC-20 has no (single chip) PLA to generate the various chip select and other timing signals, that job is done by several TTL chips.

Ah, thanks. Learned something new.

[eslapion]

Will it also work in a Vic-20?

The VIC-20 has no (single chip) PLA to generate the various chip select and other timing signals, that job is done by several TTL chips.

The C64 requires a sort of "manager" for its 16 bit bus (16 bit means 65536 adresses at the most) because it has more than 64k of RAM, ROM and IO registers in all.

That is the function of the PLA; to determine exactly what is accessible to the 6510 and 6567 and at what address it will show up.

In the VIC-20, the machine comes with 5k of RAM, 20k of ROM and you can add up to 35k of RAM or ROM. Then there is another expansion area for IO of about 2k adressable range. I'm not even sure it is possible for the VIC to have something at all possible adresses and even if it is, there is no "overlapping" or "switchable areas" such as in the 64.

Also, in the 64, the PLA plays an important role in the dynamic RAM adressing and refreshing. The VIC-20 and all Commodore made expansions for it use static RAM which does not require external refreshing or multiplexed adressing.

[craftsman1234]

Anyone heard from eslapion? I sent him some money for EPROMS at the end of March but have not heard from him since. He has not posted on here or Lemon since the end of March. I hope he is okay.

[eslapion]

Anyone heard from eslapion? I sent him some money for EPROMS at the end of March but have not heard from him since. He has not posted on here or Lemon since the end of March. I hope he is okay.

I was hospitalized from april 2014 to january 2015.

I am okay and I am back.

[pitcalco]

That is a long hospital stay indeed! I'm glad you are alright, Eslapion.

[eslapion]

After discussing on Lemon64 with other C64 enthusiasts, I consider trying the Atmel AT27C512-45 with diodes and resistors to slow down some specific lines.

[e5frog]

Is there anything wrong with the XC9536 solution, I haven't had any troubles at all finding them on eBay when I wanted to get some.

[eslapion]

@e5frog
The main problem with the XC9536 is its very expensive.

The second one is, it can't be programmed by many low cost programmers.

And last but not least, this is really a 3v chip so hidden under the PCB shown in the photo you posted is a small 3v regulator. As a consequence, this solution draws about as much power as an origianl PLA, that's about 100mA.

Finally, this is really a very fast logic chip that much faster than a real PLA but internally, its been programmed so some specific outputs have been slowed down. For this reason, it remains incompatible with many older revisions of the C64 mainboard.

A ROM based PLA, such as one based on Atmel's AT27C512 or ST's M27C512-90B6 consumes as little as 4mA, it runs on 5V just like the rest of the C64's hardware, the price of the chip is about 1$ per unit and the slower response speed is not a result of a programming trick but a low level property of the technology used and as such all outputs have a similar response speed which makes it compatible with almost every revisions of the 64.

It can also be programmed on just about any standard programmer, including the old Promenade C1.

The parts are easy to find and they don't require minuscule surface mount soldering. A programmed 28 pin DIP PROM can be easily soldered on top of a 28 pin socket with a bit of telephone grade wire or wire-wrapping wire. Its a lot easier to make your own.

[e5frog]

So, you're wrong about it being a 3V (or close) chip, 3V chip is the XC9536XL, this is a 5V compatible chip that needs 5V when programmed - there's no hidden regulator in the image I posted, I have built four of those myself (image doesn't show my build), it's just the one chip.
You usually keep to facts but this time you're just making things up.


A parallel JTAG programmer cost me €11 with shipping in 2012, I believe it was cheaper than an eprom programmer, but that's true a lot of low cost programmers can't program it. A lot of microwave ovens can't program it either as they weren't made to do that...
I have used my JTAG programmer to program that chip in the Nintendo 64 RGB DAC that "viletim" made as well and a few other things.

The first XC9536 chip I found on eBay now cost $4, I don't care enough to compare with what a "proper" eprom costs. Difference is you don't have to experiment with capacitors and nobody seems to have a problem (compared to eproms) with using this chip as a replacement.

http://www.ebay.com/itm/370838187852
I think the 5ns chip were too fast though, so don't buy this particular one. In the comments at nightfallcrew we discussed speeds a little.

Here's a better price for 50 pcs:
http://www.ebay.com/itm/380932198267
So if you made 50 of these to sell, including the cost of the programmer - perhaps not overly expensive.

How old motherboard do I need to try? I'm not sure I'm up for desoldering PLA:s just to show that it works, I really don't care to prove it as I'm not selling the product.
If programming slows down needed outputs -what's the problem with that, isn't that what you're supposed to do with a programmable chip - adapt it to your needs?
Just like people to know there's another type of replacement than an eprom - that you can also make yourself. If you just need one then a SuperPLA might be cheaper solution - or why not sell the parts again after you're done?


EDIT:
If you happen to have an eprom programmer and need one M27C512-90B6, the cheapest one I found on eBay was $7 with shipping (about $3 without). But why not buy a ready made one from Italy, only €16 plus shipping - that's right you can't just plug the "$1" eprom in, it needs an adapter as well...
http://www.ebay.com/itm/171756619191

Did anyone want to specifically put the cheapest things they can find in their C64, not the best working? You can still buy 10pcs of M27C512-90B6 for $20.83 so why start experimenting with another eprom and diodes and resistors...?

[e5frog]

Found a poor image of one of mine, it has a capacitor as well:


Here's my "working" eprom PLA, some atmel '010 IIRC, at least it worked with what I tried at the time - a loong time ago... it started working after adding the combination of capacitors I happened to have at hand.

[eslapion]

All the XC9536 I found on Digikey's web site were rated for 3V-3.6V power. If there are ones compatible with 5V, I wouldn't know where to find them.

You mention these chips are rated for 5ns. The slowest ones that were available at Digikey were rated for 10ns but I could see information about obsolete models rated for 15ns. Well, the Atmel AT27C512-45, the only revision still available at Mouser and digikey, is actually TOO FAST at 45ns for doing a good job as a PLA on all revisions of the C64.

BTW, your ROM based PLA looks like a ugly hack compared to this:
http://www.ebay.ca/itm/Commodore-C64-PL ... 262wt_1362

This guys is selling MY solution... emulation is the sincerest form of flattery!

[e5frog]

I believe the XC9536 is obsolete, DigiKey has it listed but they only have the XL version in stock - which is "3V".

The link I first posted to eBay are 5ns chips, it was tested by a poster at the nighfallcrew site and didn't work, the image with the non coated version has a 7ns, I have used 10ns and 15ns chips myself. Not that the speed rating of this CPLD has much to do with the one of an eprom. I don't think anyone answered about the recommended speed, the one that the program was made for.

Just as you can still buy other obsolete old chips NOS on eBay the XC9536 (without being followed by XL) is not hard to find.

Don't know how you search on DigiKey, the 5V 5ns is right here (not available):
http://www.digikey.com/product-detail/e ... -ND/406931

I think it's a nice little module compared to the eprom adapter - but in the end it's what works best that's most important to me, not that I intend to find out - I use a C64C with the narrow board most of the time anyway.

If you have time to make measurements on this build as well it would be nice for the community to know if it's a good replacement or not.

[eslapion]

I believe the XC9536 is obsolete, DigiKey has it listed but they only have the XL version in stock - which is "3V".

That's not the real issue which causes a potential problem anyways with this solution. If the chip is 5V input tolerant then, using a pair of diodes in series at the Vcc line, it might work as this would power the chip with 3.6v (5 - 1.4 = 3.6).

The real issue is, if you program a XC9536 rated for 15ns with cascaded internal gates to slow down the CASRAM response speed then if you stumble upon a chip rated for 5ns, 7.5 ns or 10ns then there is no telling exactly what the exact response speed will be.

The link I first posted to eBay are 5ns chips, it was tested by a poster at the nighfallcrew site and didn't work, the image with the non coated version has a 7ns, I have used 10ns and 15ns chips myself. Not that the speed rating of this CPLD has much to do with the one of an eprom. I don't think anyone answered about the recommended speed, the one that the program was made for.

Guess why...

Just as you can still buy other obsolete old chips NOS on eBay the XC9536 (without being followed by XL) is not hard to find.

Sure... in that case you should always assume you need exactly the same type of chip the original programmer used. Any "read me" in there to guide you with that?

Don't know how you search on DigiKey, the 5V 5ns is right here (not available):
http://www.digikey.com/product-detail/e ... -ND/406931

Quantity available: 0

I searched for "In stock"... not for obsolete items.

I think it's a nice little module compared to the eprom adapter - but in the end it's what works best that's most important to me, not that I intend to find out - I use a C64C with the narrow board most of the time anyway.

You were concerned with finding the best solution that's adequate FOR YOU. The people discussing the ROM based PLA solution on Lemon64 and me as well, are concerned with finding a solution that's the easiest and the most available for ALL C64 users out there or the widest possible audience.

So far the M27C512-90B6 remains the winner.

Making a custom TQFP-44 to DIP-28 adapter does not seem like the easiest solution at all.

If you have time to make measurements on this build as well it would be nice for the community to know if it's a good replacement or not.

Sure. You got a sample for me to test? Perhaps a source of information for the programming and adapter connections?