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

[eslapion]

If I am reading this correctly, would it be fair to say that different revisions of C64 mainboard PCB's have different physical layouts which may in turn have different impedance loads?

It doesn't vary much but the answer is yes. However, keep in mind that traces for digital signals on various C64 boards pretty much all have the same width and the same copper thickness so the variation isn't very big.

The idea is, compared to today's computer motherboards or even those of early 90's Amiga and PC computers, the traces in the C64 are considerably larger and were built with much less attention to the parasitic capacitance and inductance. Consequently a C64 board made by Commodore has a much lower bandwidth limit than anything built today. At the same time, these bulky traces do act as RF noise filters.

I noticed that the dodgy Winbond W27C512 EEPROM PLA substitute in one of the three 250407 variants I tested actually seemed to work ok. However all the other boards didn't go so well.

Watch for the presence or absence of R42 and C204.

I noticed that on the 2 Ku style mainboards I have they have added in extra components at the factory that are different from each other when viewed side by side.....
Were they trying to custom match the impedace load to PLA chip used?
Maybe the very early PLA chips exhibited some undesirable behaviour.
The PLA's currently in these KU boards are not the original ones but later 1986 dated PLA's.

You're not specifying what the components are but I suspect this still revolves around R42 and C204.

The very first PLAs installed in C64 had a latency of 38ns, give or take. As soon as Commodore starting making their own in-house MOS 906114-01, the latency dropped and proper access to DRAM required a slowing down of the CASRAM signal. The was done with the above 2 parts.

[banman]

Hi eslapion,


I think you're right about the RC filter combination added in the 250407 'hipric' board.



I just went through the 3 various 250407 boards I have here.

Only the 250407 'hipric' board appears to work correctly with the dodgy Winbond PLA chip.

I think I figured out where R42 is positioned. I just can't figure out where & C204...

Here's a picture of the board. Do you have any ideas where to look?

I also noticed that this board looks like it has had some components added on ad-hoc...

[eslapion]

I think I figured out where R42 is positioned. I just can't figure out where & C204...
...
Here's a picture of the board. Do you have any ideas where to look?

According to the schematics, the presence/absence of C203 and C204 depend on the type of PLA installed as in 1983, Commodore had the old 82S100 based ones and the newer ones (aka 7700-01) made by their own fab (the ones with the bad chemicals which would self destroy) and a 3rd party model called MB112A101.

See: http://www.zimmers.net/anonftp/pub/cbm/ ... 8-1of2.gif
Look at the table in the lower right corner which indicates the presence/absence of C203/C204 depending on the type of PLA installed.

PLAnkton being based on the response speed of the slowest type, it requires neither C203 or C204 but will work fine with any of the 2 installed. Only one of the 2 can be present.

[banman]

Hi eslapion,

Many thanks.
You drawing my attention to the lower right hand side of this document resolved why I didn't see the capacitor.

http://www.zimmers.net/anonftp/pub/cbm/ ... 8-1of2.gif

On my board it's marked as C 203 and is unpopulated. From the document you supplied to me my board is a PAL version that had a 6569R3 PAL VIC II chip

It is very important for me to state that even though this board appears to work well with this dodgy Winbond W27c512 PLA chip. When I observe the Winbond PLA at pin10 !ROMH on an oscilloscope I see the glitch.
I have seen the voltage on the glitch will go down to as low as 1.88v.

[banman]

Hi eslapion,

Here is were the 2 components are positioned on the board.

[eslapion]

@banman
You're studying this board in particular because it happens to work fine with a PLA substitute which is truly borderline and the glitch you noted clearly indicates this is not something you should use.

I'm wondering exactly what you're looking for.

[banman]

Hi eslapion,

Could this be why some think that the Winbond W27C512 EEPROM is a possible substitute PLA?

I noted a few forums where there was/is a great deal of discussion about using this Dodgy Winbond EEPROM 27C512 as PLA substitute.

Considering that I can observe a glitch on !ROMH & !ROML while the computer appears to function normally. Would it be fair to hypothesize that the other pins on the Dodgy Winbond PLA substitute would be generating the same characteristics as what I see on ROMH & !ROML?

If this glitch is operating on the other pins would this potentially cause chips to be called up on the mainboard's bus simultaneously? Is this what is referred to as Bus Contention?

[eslapion]

Could this be why some think that the Winbond W27C512 EEPROM is a possible substitute PLA?

I noted a few forums where there was/is a great deal of discussion about using this Dodgy Winbond EEPROM 27C512 as PLA substitute.

To answer that question implies you can assume

Considering that I can observe a glitch on !ROMH & !ROML while the computer appears to function normally. Would it be fair to hypothesize that the other pins on the Dodgy Winbond PLA substitute would be generating the same characteristics as what I see on ROMH & !ROML?[/quote]
To answer these questions implies you know for sure if you buy 1 or 100 of these Winbond ICs they will all behave exactly the same.

When I bought ST's M27C512-90B6 from Digikey or Mouser back in 2008-2011, I knew for sure all units I bought were faithful clones of each other. The units levoman has ordered from Ray Carlsen are genuine M27C512-90B6 made by ST since they were originally purchased from Mouser which is an official distributor.

Now, I just looked at Mouser's web site and Digikey's web site and none of them sell anything from Winbond and this begs the question: Where exactly did you get yours ? Right now, I assume that's from some Chinese reseller listed on eBay and this means there's a great deal of chance what you'll get is various surplus chips from all sorts of sources and all sorts of brands and they were rebranded in some shop in China. No 2 chips will have the same behavior.

If you look at Winbond's web site, they sell all sorts of RAM ICs and flash ROMS but they absolutely don't sell any (E)PROMs of 64k x8.

How can you be sure you will get consistency in the product you buy ? You simply can't.

If this glitch is operating on the other pins would this potentially cause chips to be called up on the mainboard's bus simultaneously? Is this what is referred to as Bus Contention?

Yes. The PLA in the C64 acts some sort of 8 way traffic light telling what chip can signal on the bus at any given time. The 'green light' is one of the 8 outputs of the PLA going low so if glitches occur on more than one of these 8 outputs then more than one IC is going to signal on the bus and this is a bus contention.

The 8 outputs of the C64 PLA are as follows:

- CASRAM
- BASIC
- KERNAL
- CHAROM
- GR/W (this is the only output that can go low at the same time as another and only if the 'other' is IO)
- IO (see above)
- ROML
- ROMH

Only one of these can go low at a time with the exception of IO and GR/W which will both go low when the CPU writes to color RAM.

[banman]

Hi eslapion,


Very interesting....


Yes, these Winbond branded chips are in my opinion non genuine knockoffs from who knows where...

Is it possible to check for simultaneous device activity by observing the outputs on say any 2 outputs of the PLA ? Which are the best candidates to study?

[eslapion]

Yes, these Winbond branded chips are in my opinion non genuine knockoffs from who knows where...

Then I go back to my previous suggestion: Check the AT27C512R from Microchip, the AT27LV512A also from Microchip and the AT49F512 Flash ROM.

All 3 are available at Digikey and all are from official sources.

Is it possible to check for simultaneous device activity by observing the outputs on say any 2 outputs of the PLA ? Which are the best candidates to study?

Well, usually, when a memory chip generates a glitch caused by a change of input values in the middle of an access cycle, pretty much all outputs will be affected. I don't think there's such a thing as a 'best candidate'. The only reason I picked up ROMH and ROML for the captures I posted on Melon64 is because they are available on the cartridge port so attaching a probe on the big connector legs is very easy.

If you have a logic analyzer you can have the luxury of checking all 8 outputs simultaneously but be aware what you're looking for requires a capture rate of 100MS/s or more.

[banman]

Hi eslapion,

I managed to observe a simultaneous glitch on pin10 (!ROMH) & pin 11 (!ROML) using the dodgy Winbond PLA substitute PLA on a 250407 'hipric' mainboard.

What is the ! for in !ROMH & !ROML?

I understand my equipment isn't the greatest and my skills even more so.

I notice both traces dip at the same time. I wasn't getting the best readings. I will try harder to get a clearer reading.
If these two pins are signalling towards zero volts simultaneously would it be fair to say that hypothetically this could potentially cause chips to be called up on the mainboard's bus together - a Bus Contention? What are the problems with a Bus Contention?

[eslapion]

What is the ! for in !ROMH & !ROML?

They can also be called /ROML and /ROMH and the prefix, whether it's ! or / means this signal is active when low or logically true when low. Sometimes this is also noted with a horizontal bar above the designation. You see that often in the datasheets for RAM, ROM and EPROM ICs.

See: https://en.wikipedia.org/wiki/Logic_level#Active_state

I notice both traces dip at the same time. I wasn't getting the best readings. I will try harder to get a clearer reading.
If these two pins are signalling towards zero volts simultaneously would it be fair to say that hypothetically this could potentially cause chips to be called up on the mainboard's bus together - a Bus Contention?

Correct. The convention used in Commodore 8 bit computers, TTL levels, indicate a low logic level is signaled when below 1.5V. If more than one outputs of the PLA go below 1.5V then more than one device was flagged active on the bus and a bus contention may have occurred if this flagging lasted long enough for the devices to become active on the bus.

Since older memory chips require a longer period of flagging before they start signaling on the bus then simultaneous flagging of multiple devices lasting 50ns or less are not likely to result in bus contentions if you have all the original (slow and old) chips in your C64. However, if you installed a brand new JiffyDOS module which happens to use fast 45ns EPROMS then there is a greater chance of having a bus contention between it and something else.

Looking at your scope captures, the lowest voltage captured was 2.72 volts so technically these glitches did not cause flagging of more than one device to be active on the bus and no bus contentions occurred.

What are the problems with a Bus Contention?

The multiple devices which have been flagged active enter a sort of tug of war on the bus. These can get damaged or destroyed. If these contentions are short in duration, they may simply consume more power and dissipate more heat.

[banman]

Hi eslapion,

Thank you for the clarification on the meaning of. ! & / In regards to !ROMH & !ROML. I have seen it in documents like you described but didn't get the connection.

I think in the last photo of my previous post the low voltage on probe A (!ROMH) got to 1.9V.


It was a bit tricky Tring to get the best photos to post.


Shortly after I managed to get a video of probe A triggering at 1.5V. I muffed it up though. I didn't set probe B to 500mV as it was for probe A . I think probe B was at 5V.

The oscilloscope was going as hard as it could. I am sure there must be glitches that aren't being detected due to the limitations of this machine.
Don't get me wrong the oscilloscope gas been a great tool. It has been an incredible opportunity to see something and meet interesting people who have given me a glimpse of a dimly perceived thing I couldn't explain.

Here's a video of what I observed.....

https://youtu.be/w6mrSTv7P0s

[eslapion]

I think in the last photo of my previous post the low voltage on probe A (!ROMH) got to 1.9V.

Correct.

It was a bit tricky Tring to get the best photos to post.


Shortly after I managed to get a video of probe A triggering at 1.5V. I muffed it up though. I didn't set probe B to 500mV as it was for probe A . I think probe B was at 5V.

The oscilloscope was going as hard as it could. I am sure there must be glitches that aren't being detected due to the limitations of this machine.
Don't get me wrong the oscilloscope gas been a great tool. It has been an incredible opportunity to see something and meet interesting people who have given me a glimpse of a dimly perceived thing I couldn't explain.

The setting on that last photo was 100ns per division. Could you have increased resolution to 40ns/div ?

[banman]

Hi eslapion,


Thanks, I'll try your suggestion of 40ns/div.





For what it's worth....

I managed to lift a still of the video I just posted. I am sure this is the absolute limit of the oscilloscope. I would say because it's going so hard accuracy may not be reliable. I notice it gets very flakey at the fastest speeds. I noticed the text update on the screen wouldn't keep up with the physical trigger LED.
I would also say a part of it lays with me not operating the oscilloscope correctly.
Must be another way to approach this....

I would say levoman's oscilloscope would have a better resolution than the one I'm using. Maybe he's got a dodgy Winbond PLA substitute he's willing to test or something else like it.

Probe A (!ROMH) is triggering at 1.5V with a sweep time of 10ns.
Damn, I was too busy trying to record the event on camera and keep my eye on the trigger light I didn't set probe B to 500mV. So I missed getting a look at !ROML on pin 11.