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

Other Computers and Game Systems

Moderator: Moderators

User avatar
eslapion
ultimate expander
Posts: 5458
Joined: Fri Jun 23, 2006 7:50 pm
Location: Canada
Occupation: 8bit addict

Re: The end of the ROM based PLA for the 64

Post by eslapion »

BTW: When doing some research for the develpment (halted) of the Behr-Bonz 64 the XC9536 was considered as a replacement for 74LS/74HCT series of chips.

Even if the chip works at 3.0-3.6v, it is 5V input tolerant and the output is TTL level compatible.

Thank you for the links...

http://www.nightfallcrew.com/18/11/2009 ... modore-64/

http://www.zonadepruebas.com/viewtopic. ... 8c812d64dc
Be normal.
User avatar
e5frog
Vic 20 Nerd
Posts: 551
Joined: Sat Feb 17, 2007 5:46 pm
Website: http://channelf.se
Location: Sweden
Occupation: Service Engineer

Re: The end of the ROM based PLA for the 64

Post by e5frog »

I've got one 7ns and one 15ns, the 7ns isn't built, I've also got extra 7ns circuits, maybe six or seven. But just these two boards right now.

Do you want both?

Too bad they don't share the code, just the JED-file.

XC9536 datasheet:
http://www.datasheetcatalog.com/datashe ... 9536.shtml

Programs at 5V, takes both 3.3V and 5V I/O. Supply volt needs to be at the needed I/O level for output drivers: 4.75-5.25 for 5V operation, 3.0-3.6V for 3.3V operation.

I tried using one to replace the logic and PROM in a Action Replay build, one of my first trials of my own doing that. I could build a schematic which was quite handy instead of making all of it in VHDL - but it didn't work properly - it somewhat work and I'm not sure the schematic I followed was correct. Maybe I'll get hold of a AR clone some day and can check the schematic myself.
My other interest: http://channelf.se
User avatar
eslapion
ultimate expander
Posts: 5458
Joined: Fri Jun 23, 2006 7:50 pm
Location: Canada
Occupation: 8bit addict

Re: The end of the ROM based PLA for the 64

Post by eslapion »

e5frog wrote:I've got one 7ns and one 15ns, the 7ns isn't built, I've also got extra 7ns circuits, maybe six or seven. But just these two boards right now.

Do you want both?
Everyone can benefit more information. If you can check both then by all means please do. The most crucial information is the CAS to CASRAM delay.
There is a more up to date version at Xilinx.
Programs at 5V, takes both 3.3V and 5V I/O. Supply volt needs to be at the needed I/O level for output drivers: 4.75-5.25 for 5V operation, 3.0-3.6V for 3.3V operation.
I disagree. Looking at the XC9536XL datasheet, the specs clearly specifies the chip is 5V input tolerant. In the TTL world, anything above 2V is considered a logic one. All 74LS technology chips, even when powered with 5V will signal a logic one as somewhere between 3 and 4 volts. This is accepted as logic one by all Commodore NMOS chips.

I firmly believe a XC9536XL powered at 3.6V will do the job as a good PLA, the same as a 5V powered XC9536.

I suggest you take a look at John F. Wakerly's Digital Design sections 3.12 (CMOS/TTL interfacing) and 3.13 (Low-voltage CMOS logic and interfacing)
I tried using one to replace the logic and PROM in a Action Replay build, one of my first trials of my own doing that. I could build a schematic which was quite handy instead of making all of it in VHDL - but it didn't work properly - it somewhat work and I'm not sure the schematic I followed was correct. Maybe I'll get hold of a AR clone some day and can check the schematic myself.
Gideon has implemented AR emulation in the 1541U and 1541U2. He must have access to complete information. You should check this with him.

BTW: My 866B can JTAG program all members of the XC95 family of chips.
Be normal.
User avatar
e5frog
Vic 20 Nerd
Posts: 551
Joined: Sat Feb 17, 2007 5:46 pm
Website: http://channelf.se
Location: Sweden
Occupation: Service Engineer

Re: The end of the ROM based PLA for the 64

Post by e5frog »

Not sure the up to date datasheet is important, it's an obsolete chip.

I was referring to the XC9536 and not the XC9536XL so I don't know what you disagree about, I just quoted the "old" datasheet. The XL may work fine, it wasn't something I was looking for it requires some extra components if you can't power it from 5V.

If "John F. Wakerly's Digital Design" is available as electronic material I'd be happy to take a look, whatever it is I'm supposed to look at or re-learn. I do have some digital design literature myself in my native language. English would possibly help in learning the proper terms so I don't seem stupid for making up my own words for stuff.

I don't have a very fancy oscilloscope like you do, so the question was if you can make some test if I ship (both versions) to you, PM with an address if you have the time to take a look at it, would be interesting to get some cold hard facts.
My other interest: http://channelf.se
groepaz
Vic 20 Scientist
Posts: 1180
Joined: Wed Aug 25, 2010 5:30 pm

Re: The end of the ROM based PLA for the 64

Post by groepaz »

Gideon has implemented AR emulation in the 1541U and 1541U2. He must have access to complete information. You should check this with him.
Isnt it common knowledge? A quick look at the VICE source or even http://rr.pokefinder.org/wiki/Action_Replay should tell you all you need (warning: the schematics that are floating around on the net are NOT correct)
I'm just a Software Guy who has no Idea how the Hardware works. Don't listen to me.
User avatar
e5frog
Vic 20 Nerd
Posts: 551
Joined: Sat Feb 17, 2007 5:46 pm
Website: http://channelf.se
Location: Sweden
Occupation: Service Engineer

Re: The end of the ROM based PLA for the 64

Post by e5frog »

I'd like to make my own schematic from a clone cart.
The implementations out there usually has some little extra things that works for Retro Replay and the other compatible things, code is done in modules, it's not that easy just to lift out the needed code for just the simple ordinary AR6. I'm sure code for VICE has all details if you have the proper background knowledge. I have looked at it, didn't really help me build the schematic in the editor for the CPLD.
A proper schematic would help but then there's still the work to be made to adapt it to a programmable circuit like the XC9536 (and not XC9536XL) that I used in this case. Something I'm not very skilled at, I also imagine that when using real logic chips to build it there's a matter of various timings that gets off when stuffed into something like that.

It should be possible I think, don't have the skills yet but it's a matter of money as well.

Bought a logic probe to be able to have a look at an original cart, haven't gotten one yet, one of those clones would be more interesting. Maybe I'll be able to borrow one one day.
Figuring out what's wrong with what I have already done and fixing that would be another approach. As we have a cloned TFCIII it would be nice to have a new AR6 - that's just that and not a bunch of other things with it.
My other interest: http://channelf.se
groepaz
Vic 20 Scientist
Posts: 1180
Joined: Wed Aug 25, 2010 5:30 pm

Re: The end of the ROM based PLA for the 64

Post by groepaz »

you may want to look at an AR5 instead - which is exactly the same hardware, but uses TTL logic instead of the "custom chip"... and its not timing critical at all, just plain state driven logic.
I'm just a Software Guy who has no Idea how the Hardware works. Don't listen to me.
User avatar
e5frog
Vic 20 Nerd
Posts: 551
Joined: Sat Feb 17, 2007 5:46 pm
Website: http://channelf.se
Location: Sweden
Occupation: Service Engineer

Re: The end of the ROM based PLA for the 64

Post by e5frog »

Yes, that's also in my eBay search, haven't had the money available when there has been a chance to get one.
I believe I read somewhere that 4.1 and up uses the same type of hardware, the way it works at least.

I looked at this schematic and the one from Jerzy Sobola (I think he's called) even had a chat with Jerzy about it, but it was a long time ago.
http://ar.c64.org/wiki/File:Action_Replay_MK5_6.gif

Difficult things was how to handle diodes in the description for the XC9536 (most were gates) and if the table for the PROM was correct (the outdated one with the table).
The schematic above also shorts through the 74LS74 when reset is held. Next was the pull up resistors - can be skipped?
Double gates for the chip select signal - needed as a delay?

I tried various versions, none of them worked but if I get hold of one of the logic chip versions I'll tear it down and make my own schematic.
Last edited by e5frog on Fri May 01, 2015 10:17 am, edited 1 time in total.
My other interest: http://channelf.se
groepaz
Vic 20 Scientist
Posts: 1180
Joined: Wed Aug 25, 2010 5:30 pm

Re: The end of the ROM based PLA for the 64

Post by groepaz »

that schematic is the wrong one that is floating around (i dont recall what exactly the problem was, some little detail)
I'm just a Software Guy who has no Idea how the Hardware works. Don't listen to me.
User avatar
e5frog
Vic 20 Nerd
Posts: 551
Joined: Sat Feb 17, 2007 5:46 pm
Website: http://channelf.se
Location: Sweden
Occupation: Service Engineer

Re: The end of the ROM based PLA for the 64

Post by e5frog »

Kind of crappy that a site dedicated to AR has a faulty schematic on display.

There's Jerzy's version as well, they guy who published drawings with errors for the diagnostic harness - IIRC.
Image

I compared the two when I was "in the zone" with it and asked Jerzy about the differencies, but he couldn't really help at the time.
Still, even if I have a 100% proper schematic I'm not sure I'll be able to make it properly in the Xilinx software and make it work in a CPLD, I planned to stuff all logic in there like the AR6 and not just the PROM - I was considering trying just PROM replacement as well in a coming version but - preferably after confirming a schematic with real hardware - then there was some other projects in between. Still have three nice Oshpark purple and gold PCB:s, would be nice if they could be made to work.

My experimental board, tried putting the diodes outside - worked equally as the programmed solution.
Image


Anyway - it's getting far away from the subject.


Should I ship to you or can you direct me to make any special measurements on the XC9536 PLA using my old analog two channel oscilloscope. What software to run or how to properly measure.
My other interest: http://channelf.se
User avatar
e5frog
Vic 20 Nerd
Posts: 551
Joined: Sat Feb 17, 2007 5:46 pm
Website: http://channelf.se
Location: Sweden
Occupation: Service Engineer

Re: The end of the ROM based PLA for the 64

Post by e5frog »

That's right, one problem with the first mentioned schematic is this:
Image

Upon pressing reset, VCC and GND meet in the node of the X.
My other interest: http://channelf.se
User avatar
eslapion
ultimate expander
Posts: 5458
Joined: Fri Jun 23, 2006 7:50 pm
Location: Canada
Occupation: 8bit addict

Re: The end of the ROM based PLA for the 64

Post by eslapion »

@e5frog
PM sent!
Be normal.
User avatar
eslapion
ultimate expander
Posts: 5458
Joined: Fri Jun 23, 2006 7:50 pm
Location: Canada
Occupation: 8bit addict

Re: The end of the ROM based PLA for the 64

Post by eslapion »

Tested the M27C512-90B6 on two different C64 with board 250407 and ceramic 6567R56A. Both machines have chips dated early 1983.

One of the two boards has 250ns DRAM, the other has 150ns DRAM:
Board with 250ns DRAM works well except if I set my 1541U to emulate Fastload. Then I get a startup screen with no text.

Board with 150ns DRAM works well in every aspect I know of.
Be normal.
User avatar
eslapion
ultimate expander
Posts: 5458
Joined: Fri Jun 23, 2006 7:50 pm
Location: Canada
Occupation: 8bit addict

Re: The end of the ROM based PLA for the 64

Post by eslapion »

In the specific case of the M27C512-90B6, there seems to be compatibility problems with older KU boards as well as 250407 rev a boards. The only difference that's notable between 250407 rev a and rev b is the presence of C204 on the rev b which, just like on board 250466 causes RC filtering of !CASRAM to delay the reading by DRAM of the multiplexed addresses.

After a short exchange with Commie on Lemon64, I implemented the same type of RC filter Commodore engineers installed on boards 250407 rev b and all subsequent revisions. I made sure this RC filter has the same time constant as the one on board 250466 (about 12ns) and plugged this modified sub on my own board 250407 rev a which had problems with an emulated or real Fastload cartridge. Sure enough, the presence of the RC filter resolved the problem.

Image

The RC filter added consists of a 56 Ohms resistor and a 220pf capacitor.

After an "invitation" from Jens Schönfeld, I decided not just to take a quick look at the threshold voltages on input of the Commodore PLAs and the ST PROM but to also take a close look a the exact values and the resulting hysteresis which makes them compatible with devices that cause slow transition signals such as the Epyx Fastload.

The SuperPLA V3 is incompatible with the Epyx Fastload cart while the ST PROM has no problem.

The input is a triangle wave adjusted (DC offset) to have a max value of approx 4.5v and a min value of 0.5v

Cursor 1 is the negative going threshold voltage
Cursor 2 is the positive going threshold voltage
Indicated delta is the amount of hysteresis.

As a guideline, a 74LS14 TTL Schmitt trigger has about 800mV of hysteresis.

The output signal clearly indicates the threshold points where DUT perceives a 0 or a 1.

DUT #1: 82S100
Image
Results on the scope:
Image

DUT#2: 8700R2
Image
Results on the scope:
Image

DUT#3: The M27C512-90B6
Image
Results on the scope:
Image

And this lovely little chip does all the work while consuming only this much power...
Image
It has the best noise resilience and consumes only one tenth the power. Who could possibly ask for better?

Bring on the "abusive" carts...

Test configuration is as follows:
!OE pulled low - all inputs high except pin 9 which receives the triangle wave - digital output is on pin 18.

Incidentally, these snapshots prove Skoe was right when he said that the voltage required to signal a logic one should be considered to be 1.5volts, not 2 volts as I did originally.

My square shaped crystal ball has spoken again!
Last edited by eslapion on Thu Jun 04, 2015 3:31 pm, edited 2 times in total.
Be normal.
User avatar
eslapion
ultimate expander
Posts: 5458
Joined: Fri Jun 23, 2006 7:50 pm
Location: Canada
Occupation: 8bit addict

Re: The end of the ROM based PLA for the 64

Post by eslapion »

As soon as I posted this information on Lemon64, Mr. Jens Schönfeld and his usually devout follower Groepaz stumbled in there and started posting extremely negative (borderline to defamatory) messages, for example, to the effect that this was a hack of some sort. Never mind the fact that this RC filter is present on all C64 boards starting with 250407 rev b. The one on board 250466 is the most intense since the load on this specific line is alleviated by the removal of 6 DRAM chips.

Of course, I was the only one subjected to this type of bullying but I was not the first one to observe the RC filter on the !CASRAM line on all post 1983 C64 boards. This post is from Skoe and dated October 2012:
CBM made this with their CASRAM RC delay already :/
Be normal.
Post Reply