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

[eslapion]

We are speaking about a circuit that has to find all spikes onto 2 lines, without reasons the lines have to not change state, otherwise could be generated a problem.

No we were not. We were discussing the use of the Super Zaxxon cartridge (or clone thereof) to detect glitches in the signal driving a flip-flop.

The 74LS279 circuit checks for 2 signals that are supposed to stay high continuously. The Super Zaxxon cart uses a flip-flop (74LS74) to toggle access to the lower/higher ranges of address of an EPROM and therefore requires a signal that has transitions in order to operate properly. It checks only one signal but it's more demanding. The signal must both be glitch free and switch correctly.

So:
you are telling that IF a potential problematic spike is generated (spike = 2 fast state changing= slaw rates presence) my solution will find all of them, also the fast one.
you are also telling that without a spike (=without states changing =without presences of slew rates) my solution has not problem.

No!
If you use a 74F to check on signals generated by NMOS/HMOS/TTL-LS and destined for the same type of technology (low slew rate), the 74F will find both real glitches AND ordinary noise. It's NOT the right type of chip for the task.

And that's because (see figure 3)... https://www.microsemi.com/document-port ... t-app-note

Please stop misinforming people and stop wasting my time.

[levoman]

Have had a busy few days . I have some eproms layng around that i have been testing , and the results look promising the one's that work have not yet set off the latch test circuit .

I have created a pcb to adapt a 27c512 to the PLA socket. I know plenty of people have done one , but i can't seem to find one that has been shared!

https://www.pcbway.com/project/sharepro ... apter.html

will post pictures when i receive them next week. The cartridge pcb's should be here at the same time so i will hopefully be able to have one up and running . Again i will post pictures to show it working and let people know the eproms , if they work with it!

[banman]

Hi levoman,


Nice work on your PLA replacement PCB!


I used Winbond W27C512 EEPROM, a Commodore Factory PLA and a 27C512 E/PROM to ensure that my latch circuit had the appropriate 74LS279 Quad Set-Reset Latch.
The Winbond W27C512 EEPROM I had was absolutely not suitable as a E/PROM based PLA solution.

As a side note I also found that using a modest logic probe on !ROMH and on !ROML also gave me valid test results. I imagine that the logic probe has the correct TTL trigger threshold levels.. Someone with far greater understanding can no doubt elaborate on this topic....


I got my boards made up from this site:

https://oshpark.com/shared_projects/mdayhhyk

you can download the gerbers and take them to any PCB manufacturer you prefer. I also used PCBWay and found their boards better quality and cheaper.



I didn't put the delay timing capacitor in.


Burned a PLA.BIN with a TL866 minipro from here:

http://vic20.de/html/eprom_pla_8296_und_c64.html


It all appears to be working fine so far. I had originally used a Winbond 27C512 45ns eeprom. These worked with lots of screen artifacts on quite a few programs/games.


I am interested to know the EPROMS you are using that are giving you promising test results?

[levoman]

Remember these have NOT been tested with Zaxxon yet.

When I found the information for a ROM substitute for a PLA I dug into my box of memory to find any 512 memories that i could and then programmed them all for PLA use. The first one that i tried was the 120ns one . A complete failure! then i tried the 150ns amd one and it worked , not only worked but I loaded tape games and played them for a while and the latch circuit NEVER set.
I tested the others , one by one and found a few that seem to work. Here are the one's that DO NOT work.

and the one's that DO seem to work!

I have two of the AMD one's on the left and both seem to work fine!

Strange that they all seem to be 150ns. Also strange that the one's that do not are all national semiconductor!!!

[levoman]

a question for eslapion.

I am thinking of putting a buffer on the inputs so that i can run a PLA and a ROM in tandem and compare the outputs of the two on my analogue scope. What loading would you recommend for the outputs of the chip that is not connected?

i.e. two lots of buffers one for PLA and one for ROM connected to the input lines and leave the pla outputs connected to the c64 but load the ROM outputs to get a simulation of being connected. The buffers will be ls541's as i have plenty of them.

Any input would be appreciated.

[banman]

Hi levoman,

Holy cow!

I've actually got chips like those! I never tested them.I am sure I've got photos of them on Melon64 thread.

https://www.melon64.com/forum/viewtopic ... 1&start=50


Fairly sure I have AMD, Texas instruments and Intel ones too.

I only investigated the one that was the closest match to eslapion's ST unit. The ST one was an original from the electronics shop dating to the early 90's or 80's.
That's why I only focused my attention on the ST line of E/PROMS.

That's very exciting!

I'm going to test the others and get back to you......

[vicist]

Maybe watch this from Adrian's Digital Basement.

29 minutes of in-depth analysis of how to make a pretty darn good (and cheap) replacement pla that works with most revision boards AND with the Super Zaxxon cartridge.

It's worth a look.

[levoman]

Yes i have already got boards and chips coming from china , just awaiting the arrival!

[banman]

Hi levoman,

I have been doing some more testing of other chips.

I thought I had more 27C512 EPROMS. unfortunately a majority of them are 256kb and lower capacity chips.



I tested the following chips on a C64 Mainboard KU 141194HB with original 200ns memory.

These original chips don't work. However they do not trigger the Glitch testing circuit.

I think the main reason is the speed is 200ns and is running close to the speed of the C64's memory I was using.

C64 PLA ATMEL AT27C512-20DC 8722 UV EPROM Failure



Maybe a faster rated version could be suitable.


These original 150ns speed EPROMS do seem to work very well. They don't trigger the Glitch testing circuit. I will say the slow EPROM speed will have an effect on later C64 mainboards running 150ns memory.

C64 PLA original TMS JL 27C512-15 Blue9410000 Korea & Original ST27C512-15F1 B88AI 9331C Singapore and Original TMS 27C512-15JL Aleq 9127 UV EPROM Success (requires capacitor) 150ns speed



In this last test I used Knockoff EPROMS recently purchased online.

C64 PLA Project Knockoff ST27C512-90F1 B8810 0033K Singapore & Knockoff ST27C512-70XF1 5880I 9849E Singapore UV EPROM Success (requires capacitor)

It is interesting to note that in these observations I needed to run a capacitor across the CASRAM line (I think this is the way to word it, please correct me if I'm incorrect).

I have only seen the Knockoff ST27C512 EPROM's stamped CHINA and Korea work effectively without the CASRAM capacitor. I have posted these observations earlier in this thread.

I think hypothetically one might be able to find contemporary or Knockoff chips online matching these EPROMs.

This is the board I have been using.

[eslapion]

a question for eslapion.

I am thinking of putting a buffer on the inputs so that i can run a PLA and a ROM in tandem and compare the outputs of the two on my analogue scope. What loading would you recommend for the outputs of the chip that is not connected?

i.e. two lots of buffers one for PLA and one for ROM connected to the input lines and leave the pla outputs connected to the c64 but load the ROM outputs to get a simulation of being connected. The buffers will be ls541's as i have plenty of them.

Any input would be appreciated.

I'll give you my input on the outputs of TTL-LS technology ICs...

If you want to run a ROM as a logic circuit in tandem with a genuine PLA and want to load the outputs as if they were attached to standard NMOS/TTL-LS chips then know the inputs of these chips is usually equivalent to a 10kOhms pull-up resistor. I suppose you'd want to load the outputs of the device under test with something equivalent.

The EPROMs usually are using CMOS technology which has a pretty high input impedance so I'm quite sure the presence of buffers isn't required. In fact, it could add extra load on the lines and cause a delayed response.

Maybe watch this from Adrian's Digital Basement.

29 minutes of in-depth analysis of how to make a pretty darn good (and cheap) replacement pla that works with most revision boards AND with the Super Zaxxon cartridge.

It's a pair of GAL20V8 and all the logic equations are there but it's a bit too fast. It may not work well with early 250407 boards which don't have the CASRAM filter or boards 326298.

[levoman]

ok so forget the buffers and use a 10k pullup on the outputs.
i will give that a try and see what we get.

Thank You.

[levoman]

They have created a version of the software that adds an extra element to the casram signal to generate the delay.

I will compare them with a pla when they arrive.

Just bought a 100mhz dual channel scope to give better resolution than my 20mhz one. Should be with me next week.

banman , I notice that you are using one of the KU boards that gave people so much trouble at the start of creating PLA replacements , have you done the mod they speak of ? Or is that for a different board ? Or has it already been done on yours ?

[banman]

Hi levoman,



Thanks for sharing your test results on the non ST based EPROMs. You might want to reference them on your PCBWAY project.
I don't think I would have ever tried to test these other brands I had here.....


Regarding this pictured KU style board.
I personally have done nothing but replace 2 leaked capacitors near Vic II chip inside that shielded cage and the SID chip with a SwinSid b nano. Some of the factory socketed chips have been replaced with 1983 and beyond dated ones. Not by me though.
Interesting to note that the PCB on the Mitsumi keyboard has VIC20 stamped on it

Notice the additions to the front side of the board of a jumpered brown ceramic disk capacitor (above the memory chips near u13) and a resistor near u28.
On the back side there are little points made from what I can tell are very precise small drill marks that cut traces (I would say 1-2mm dia. I missed seeing them at first when I got the machines). As far as I can tell they look like they were done on the production line.

Interestingly I have another KU style board, these components I mentioned above aren't there. However there are other components added close to the outside of the Vic II chip shielded cage. These too look very much like they were done on the production line.
The only changes to that board is I socketed the ram and associated ram logic chips and a replacement NOS 6510 cpu.
I was mislead by the dead test cartridge thinking it was faulty memory. In fact the 6510 CPU was only partially working . It wasn't able to boot however it was allowing the dead test cartridge to run the sequence of flashes.


I heard stories that no 2 KU style mainboards that came off the factory floor were the same. They had loads of issues getting them to pass QA. They were bodged as they went thru the production line.
It cost them too much time and money so they quickly moved onto the next board revision.

I observed the E/PROM chips under test work without any issue in both of these KU boards. The CASRAM capacitor needs to be in place on the EPROM as discussed in the above posts for it to work correctly.
Again only the China and Korea ST M27c512 labeled OTP chips I mentioned way way back in this discussion work fine without the addition of a capacitor across the CASRAM line.

I used a 250407 "Hi Pric " board and found that even the infamous Winbond w27c512 electrically erasable EEPROM performed very well . These are plentiful boards. I think this is why some say these EEPROM chips are a suitable E/PROM based PLA substitute replacement.

[banman]

What I might do is break out the oscilloscope and take some readings on this unmodified KU style mainboard.
My oscilloscope is an old steamer...

I'll post some observations......

[levoman]

my current scope is a Hameg 20mhz model the one i have coming is a tektronix 100mhz model!!

should do better as i have tried comparing two PLA's to check the loading but i am getting a lot of ghosted traces.

It does show that the loading is not quite right. Could be the capacitive load from the veroboard?