ps i’ve got a 28pin nvram board here if you want £13 posted. will just plug into the 28 pin socket.
Pinball info
Register a free account today to become a member! Once signed in, you'll be able to participate on this site by adding your own topics and posts, as well as connect with other members through your own private inbox!
System 11 high scores show 00.
- Thread starter David_Vi
- Start date
I'll take it at least to eliminate it.
I've done it a few more times to double check.
Green is low
Red is high.
When on normally red is on permanently.
When I power cycle I barely see green/low. If I do it's a flicker.
I've power cycled it at least ten times testing and the scores haven't gone yet.
that’s correct the green should be low for a very short time, then it goes high and stays high.
Probably ok and i’d change the nvram as you say to eliminate it. If a new nvram has the same issue, then we have ruled that out. next after that would maybe c30
Probably ok and i’d change the nvram as you say to eliminate it. If a new nvram has the same issue, then we have ruled that out. next after that would maybe c30
- Joined
- Jul 21, 2011
- Messages
- 3,399
- Alias
- .
You're supposed to be eliminating the weird and wonderful things that occur with nvram with some more nvram?
NVRAM normally works just fine. And sorry David, thought the second pic in your before/after was after.
You need a 6264 8kx8 RAM as a plug in replacement, if you want to use a 6116 you need to change W6 to W5 above the RAM.
https://www.mouser.co.uk/ProductDet...w0NFuIszl7QXIPdD_FTHye9u-ny4U2DBoC-RUQAvD_BwE should work e.g. Please check if D1 and D2 are still there as you will otherwise run into trouble with the batteries.
You need a 6264 8kx8 RAM as a plug in replacement, if you want to use a 6116 you need to change W6 to W5 above the RAM.
https://www.mouser.co.uk/ProductDet...w0NFuIszl7QXIPdD_FTHye9u-ny4U2DBoC-RUQAvD_BwE should work e.g. Please check if D1 and D2 are still there as you will otherwise run into trouble with the batteries.
I'll try most things and thought it would be good to try different nvram as most people use nvram with no issue
I don't mind buying ram to try but to buy one from Mouser means £12 shipping
Are there any more local options?
Fubar
Site Supporter
I somehow doubt a bad RAM explains the symptoms. If there was a bad data line the scores wouldn't be zero and other settings would be garbled. If it had a bad sector the whole score (incl initials) would be corrupted. I could be wrong as not familiar with the flash layout.
Might be worth going through the pinmame source code for clues...
Might be worth going through the pinmame source code for clues...
It doesn't make sense to me either, I don't know how it works but for it to lose that one bit of data is weird. But investigating source code is something I have absolutely no clue about.
Why I thought it was a setting flipping - makes no sense for all the highscores to zero out on their own. Don't know how the RAM is mapped though, so maybe it is a data line or such. I do think that there is a read/write test at start to identify such issues though.
As I said, I don't think theres any setting that would make scores 00. All my backup scores are set to different amounts which are factory settings. If i hold the high score reset button in I get the presets come up, no zeros.
Fubar
Site Supporter
Had a quick look and there's nothing of interest. Pinmame just models the nvram as a chunk of address space and the layout is entirely up to the game code. I don't feel like disassembling the game code, sorry
Fubar
Site Supporter
No idea although I would be interested. I've never spent much time reverse-engineering the WPC and earlier stuff although I'm sure others have - would certainly be keen for any pointers.
A quick google reveals this page which seems like a good place to start - there goes the next hour(s) of my life!
Edit: and this seems like quite the selection! https://github.com/tanseydavid/WPCResources
Last edited:
Fubar
Site Supporter
Conveniently IEEExplore just published an article on this 2 weeks ago. Access is restricted but I can get it via work (it's watermarked though so not going to share it publicly, sorry).
Says the APPLE OS was designed by Larry DeMar in 1989 for FunHouse. So definitely WPC-only and I'm guessing sys11 didn't have the same kind of OS layer (probably some copypasta from the previous game)
Edit: the author's father used to work at Williams and has a copy of the original source code lying around - there's an excerpt in the article!
Says the APPLE OS was designed by Larry DeMar in 1989 for FunHouse. So definitely WPC-only and I'm guessing sys11 didn't have the same kind of OS layer (probably some copypasta from the previous game
)
The Williams Advanced Pinball Programming Logic Executive Operating System
Pinball has its roots in tabletop games from the late 1800s and early 1900s, although the 1930s marked the true start of pinball as we know it today. Early pinball was an electromechanical game, with points tallied by hand or through the movements of gears. For a while, it looked pretty far from...
ieeexplore.ieee.org
Edit: the author's father used to work at Williams and has a copy of the original source code lying around - there's an excerpt in the article!
Last edited:
It's a straight swap? If so I'll definitely try it. I've searched the part number and can't find any available. Is there a suitable substitution?
View attachment 165293
I noticed the knocker pulses on turn off also, I only saw it as I have the backbox open. Doesn't get enough power to fully knock.
I'll give the logic Probe a go
I can't keep track with this thread too many people saying things, but the ram in use is a ramtron 1608 which is a 6264 (8kx8) ram. The board was rejumpered for this so a 6116 or (2kx8) version will NOT work. Also changing things randomly is just going to end up with a non working game again.
I did say before on pm that i can send you another ramtron chip to swap over if you want to try that. Ie the same type to see if for some reason the first one has got a section which has intermittent save on power off
I'm happy to try another if it could be the current Ram. I don't remember you offering that, sorry, it's been hectic as hell here!
At least the game is reliable now, except the high scores.
Hi all.
I wanted to pitch in at this stage and (hopefully) clear up how SRAM/NVRAM works as it may help the OP with the reported problem. Disclaimer: I’m basing this on how WPC works as I don’t have schematics for System 11. If someone would be good enough to let me see the System 11 schematics, I’d be happy to decipher them and try to explain what is going on in this case.
A standard CPU board is fitted with Static RAM (SRAM). An earlier post suggests this is a 2K x 8 SRAM. This means that it has the capacity to store 2000 bytes (a byte being 8 bits – shown as D0-D7 on the schematics).
The SRAM is used to store information written to it by the CPU during game operation and configuration. This may include high scores, system settings, date/time settings etc. To ensure this information is retained when the machine is powered off, the SRAM is kept alive using power sourced from batteries (VBatt on the schematics). When the machine is powered back on again, VBatt is replaced by power derived from the big transformer in the belly of the pinball machine – known as Vcc (5V). Note that the electronics is designed in a way that ensures the RAM is supplied by VBatt OR Vcc – not both at the same time. This is achieved using diodes (think of them as one-way valves for the power). This ensures that the batteries don’t receive any power from Vcc when the machine is powered on, and the batteries don’t attempt to power all the machine electronics when the mains power is off.
Some people choose to replace the SRAM with NVRAM (NV = Non-Volatile). NVRAM works in the same way as SRAM in that it just stores data written to it by the CPU. The main difference is that NVRAM retains the data written to it even when power is removed. This means that you don’t need VBatt to keep the stored data alive when you turn off the machine. As such, the batteries can be removed (they can be left in place if required – they won’t do any harm).
To summarise: -
The time/date used by the RTC is stored in RAM.
Let’s assume you still want to fit NVRAM (or a new SRAM) and are concerned about which parts to fit. The short story is that you can fit pretty much what you like – but it will require soldering skills and an understanding of how RAM is addressed.
If we assume we have a 2K RAM, we will need 11 address bits to be able to access all the RAM locations. These can be found on schematics as A0-10 (we always count from 0 in electronics). If you raise 2 to the power of 11, you get 2048 – or 2K.
Using this model, you can see that extra address bits/lines means we can address/store more data – 12 address bits gives us 4096 bytes, 13 gives us 8192 bytes. You can see that this means 64 bit PCs can address a staggering amount of memory.
Note that it is increasingly difficult to obtain small RAM chips nowadays. You can fit larger RAM chips (SRAM or NVRAM) as long as you can connect the address pins/lines on the new chip to the CPU board. All you need to do is ensure that the address lines, data lines, write enable (WE), output enable (OE) and chip enable (CE) lines match and you should be fine – the system will read data back from the locations it left the data in. The only thing you really need to do is make sure the access speed of the replacement RAM is at least as fast as the original SRAM. Back in the day, SRAM speeds were in the order or 70 something nanoseconds (70ns) which is pretty slow by today’s standards. I would expect and modern RAM to be faster than this – but it is worth checking.
Looking at the original post, the reported problem is that the previous scores have been lost and the system restored to factory defaults after being powered off. Being honest, I’m unsure as to why the system would remember the player initials but not the score if the problem was related to SRAM or NVRAM.
BTW, I’m also concerned that the OP reports that nearly every transistor in the switch matrix column drivers has failed – this suggests a deeper, perhaps more generalised problem (e.g., power supplies?).
Again, if someone could let me look at the schematics, I can work out the differences between WPC and system 11 and have a better guess at what might be causing the problem.
I wanted to pitch in at this stage and (hopefully) clear up how SRAM/NVRAM works as it may help the OP with the reported problem. Disclaimer: I’m basing this on how WPC works as I don’t have schematics for System 11. If someone would be good enough to let me see the System 11 schematics, I’d be happy to decipher them and try to explain what is going on in this case.
A standard CPU board is fitted with Static RAM (SRAM). An earlier post suggests this is a 2K x 8 SRAM. This means that it has the capacity to store 2000 bytes (a byte being 8 bits – shown as D0-D7 on the schematics).
The SRAM is used to store information written to it by the CPU during game operation and configuration. This may include high scores, system settings, date/time settings etc. To ensure this information is retained when the machine is powered off, the SRAM is kept alive using power sourced from batteries (VBatt on the schematics). When the machine is powered back on again, VBatt is replaced by power derived from the big transformer in the belly of the pinball machine – known as Vcc (5V). Note that the electronics is designed in a way that ensures the RAM is supplied by VBatt OR Vcc – not both at the same time. This is achieved using diodes (think of them as one-way valves for the power). This ensures that the batteries don’t receive any power from Vcc when the machine is powered on, and the batteries don’t attempt to power all the machine electronics when the mains power is off.
- When the mains power is off, VBatt powers the SRAM through a diode (lets call it diode 1) – whilst another diode (diode 2 - fitted in the ‘other direction’) prevents the power from VBatt making it to the rest of the electronics in the machine (which would flatten the batteries very quickly).
- When the mains power is on, Vcc powers the SRAM through diode 2, whilst diode 1 prevents this power making it to the batteries (standard alkaline batteries are not designed to be charged from an external source).
Some people choose to replace the SRAM with NVRAM (NV = Non-Volatile). NVRAM works in the same way as SRAM in that it just stores data written to it by the CPU. The main difference is that NVRAM retains the data written to it even when power is removed. This means that you don’t need VBatt to keep the stored data alive when you turn off the machine. As such, the batteries can be removed (they can be left in place if required – they won’t do any harm).
To summarise: -
- SRAM needs power at all times (from VBatt or Vcc) or it will lose its stored data.
- NVRAM only needs power when something is trying to write data to it or read data from it (from Vcc).
The time/date used by the RTC is stored in RAM.
- If you use SRAM (with VBatt and Vcc power supplies), the time/date is always retained as the RTC circuit and SRAM are never powered down.
- If you use NVRAM (with VBatt and Vcc power supplies), the date/time is again retained as the RTC circuit and SRAM are never powered down.
- If you use NVRAM with Vcc power supply only (i.e. batteries removed), the date/time will stop counting when the machine is powered down as the 32KHz clock will stop running.
- In the above NVRAM cases, the other data (system settings, configuration etc) are always retained – even after turning off the power.
Let’s assume you still want to fit NVRAM (or a new SRAM) and are concerned about which parts to fit. The short story is that you can fit pretty much what you like – but it will require soldering skills and an understanding of how RAM is addressed.
If we assume we have a 2K RAM, we will need 11 address bits to be able to access all the RAM locations. These can be found on schematics as A0-10 (we always count from 0 in electronics). If you raise 2 to the power of 11, you get 2048 – or 2K.
Using this model, you can see that extra address bits/lines means we can address/store more data – 12 address bits gives us 4096 bytes, 13 gives us 8192 bytes. You can see that this means 64 bit PCs can address a staggering amount of memory.
Note that it is increasingly difficult to obtain small RAM chips nowadays. You can fit larger RAM chips (SRAM or NVRAM) as long as you can connect the address pins/lines on the new chip to the CPU board. All you need to do is ensure that the address lines, data lines, write enable (WE), output enable (OE) and chip enable (CE) lines match and you should be fine – the system will read data back from the locations it left the data in. The only thing you really need to do is make sure the access speed of the replacement RAM is at least as fast as the original SRAM. Back in the day, SRAM speeds were in the order or 70 something nanoseconds (70ns) which is pretty slow by today’s standards. I would expect and modern RAM to be faster than this – but it is worth checking.
Looking at the original post, the reported problem is that the previous scores have been lost and the system restored to factory defaults after being powered off. Being honest, I’m unsure as to why the system would remember the player initials but not the score if the problem was related to SRAM or NVRAM.
BTW, I’m also concerned that the OP reports that nearly every transistor in the switch matrix column drivers has failed – this suggests a deeper, perhaps more generalised problem (e.g., power supplies?).
Again, if someone could let me look at the schematics, I can work out the differences between WPC and system 11 and have a better guess at what might be causing the problem.
I've not had it restore to factory settings since I had all the board issues. The transistors were old and looked damaged, probably the moisture damage.
Nothing has failed in over a week whereas before I had them failing on a daily basis.
The only issue i've had now is the high scores being 00 but retaining all other info.
The schematics for sys11 are in the operation manuals for each pin, I don't think there's a separate schematics book like with later WPC systems.
Nothing has failed in over a week whereas before I had them failing on a daily basis.
The only issue i've had now is the high scores being 00 but retaining all other info.
The schematics for sys11 are in the operation manuals for each pin, I don't think there's a separate schematics book like with later WPC systems.
OK, thanks.
Any chance of scanning the schematic and sending it to me?
Any chance of scanning the schematic and sending it to me?
schematics are on ipdb - diner
Hi Dave (et al),
Thanks for pointing me at the schematics. I’ve had a good look at them now and I’ve also re-read all the posts in this thread. I’m still a little unsure about why your scores are resetting to 0 but I have a couple of ideas.
Looking at the schematics, it is obvious that this System 11 design is the basis of the WPC system that followed.
I can also see why they moved to an ASIC in WPC. The System 11 design made extensive use of PIA devices (Peripheral Interface Adapters) for IO expansion and connectivity – which is much easier to achieve using a single ASIC. I’m currently doing a similar thing by embedding my WPC CPU, Fliptronics 2 and DMD driver board into a single CPLD device – but that’s another, much more complex story.
I can also see that System 11 uses a similar system for powering the SRAM/NVRAM. The RAM is continually powered either from the main 5V line (+5VDC on the schematic) or a battery derived voltage (CMOS B+ on the schematic). I note that there doesn’t seem to be a real time clock (RTC) circuit so I’m assuming your pinball machine doesn’t know or store the time/date.
Anyway, back to your problem.
The first thing I noticed is that the NVRAM you have fitted to position U25 is an old Ramtron/Cypress FM1608-120-P6. This device is known as FRAM – Ferroelectric NVRAM. This just means it uses a slightly different internal technology to function as NVRAM. I had a look at the device datasheet. This is a relatively slow device – taking around 180ns for each read or write cycle. A (modern) SRAM would only take around 30ns.
Is this the cause of your problem? Unlikely but it is possible the slow access speeds of your chip could potentially cause problems if they are significantly slower than the SRAM that the Williams designers had in mind.
I also noticed (from the photos that you posted) that you don’t have onboard batteries (you may have them somewhere else and connect them to the board using wires – but I can’t tell from the photos).
Is this the cause of your problem? Highly unlikely as you are using a form of NVRAM. The data should be retained when you turn off the machine even though the batteries are not fitted (which is the whole point of fitting NVRAM in the first place). Please note my previous comments – you can leave batteries fitted even if you have NVRAM – it won’t do any harm even though they aren’t needed.
This is going to be a bit tricky to diagnose I’m afraid – particularly as I don’t know how the RAM is mapped to the CPU. By that I mean that I don’t know where (within the RAM) that the high scores and player initials are stored. I’ll stick a couple of potential causes on here – none of which I’m particularly confident about – but they are the only things I can think of.
It is possible that you have a faulty area of RAM – but this faulty area would have to be the area where the high scores are stored. The reason I’m not happy with this theory is that it would mean that the high scores and player initials would have to be stored in different areas of RAM. The RAM is divided into 8 separate 1K blocks – so it is possible (however unlikely) that the RAM could have a faulty block.
It is worth bearing in mind that FRAM (as do all components) has a limited lifetime – but FRAM/NVRAM lifetime is affected by the number of times you write to a RAM location. This is known as Write Endurance. Admittedly, this number is big (around 100 million cycles) but modern software that writes to NVRAM take this into account and is written to spread the load across the device. For example, a memory location in the disk used in a PC (solid state disk – SSD) would have a significantly shorter lifetime if it continually wrote to (and read from) the same locations. This is why these devices have clever ‘wear levelling’ algorithms in them that spread the load and make sure that (as far as possible) all memory locations are used an equal amount of time. The Williams engineers probably use the SRAM to store all sorts of information many times – and they almost certainly didn’t consider NVRAM and wear levelling when they wrote their software. So, there is the possibility that the NVRAM has locations that have simply worn out. Again, this is unlikely but possible – and I can’t say for sure as I don’t know how the system partitions/allocates/uses the available address space.
Another possibility is a faulty address line (either on the CPU board or within the FRAM). This would have the effect of rendering an area of FRAM inoperable. As before, this may be unlikely unless the high scores and player initials are stored in different areas within the FRAM chip. You might want to use your logic probe to make sure that all the address lines (A0-A10) are toggling - just in case you have a cheeky open circuit somewhere.
As you can see, I’m a bit stumped by this one but I do have some suggestions (which have already been suggested in other replies).
If you’d like to do this, a suitable replacement can be found at RS components (part number 538-158) – they are about £3 each. Please note that this chip has an extra control pin (pin 26 – Chip Enable 2). This pin would need to be pulled up to 5V by connecting it to pin 28. Let me know if you’re not sure what I mean, and I’ll attach some photos.
Yell if you have more questions. Good luck.
Thanks for pointing me at the schematics. I’ve had a good look at them now and I’ve also re-read all the posts in this thread. I’m still a little unsure about why your scores are resetting to 0 but I have a couple of ideas.
Looking at the schematics, it is obvious that this System 11 design is the basis of the WPC system that followed.
I can also see why they moved to an ASIC in WPC. The System 11 design made extensive use of PIA devices (Peripheral Interface Adapters) for IO expansion and connectivity – which is much easier to achieve using a single ASIC. I’m currently doing a similar thing by embedding my WPC CPU, Fliptronics 2 and DMD driver board into a single CPLD device – but that’s another, much more complex story.
I can also see that System 11 uses a similar system for powering the SRAM/NVRAM. The RAM is continually powered either from the main 5V line (+5VDC on the schematic) or a battery derived voltage (CMOS B+ on the schematic). I note that there doesn’t seem to be a real time clock (RTC) circuit so I’m assuming your pinball machine doesn’t know or store the time/date.
Anyway, back to your problem.
The first thing I noticed is that the NVRAM you have fitted to position U25 is an old Ramtron/Cypress FM1608-120-P6. This device is known as FRAM – Ferroelectric NVRAM. This just means it uses a slightly different internal technology to function as NVRAM. I had a look at the device datasheet. This is a relatively slow device – taking around 180ns for each read or write cycle. A (modern) SRAM would only take around 30ns.
Is this the cause of your problem? Unlikely but it is possible the slow access speeds of your chip could potentially cause problems if they are significantly slower than the SRAM that the Williams designers had in mind.
I also noticed (from the photos that you posted) that you don’t have onboard batteries (you may have them somewhere else and connect them to the board using wires – but I can’t tell from the photos).
Is this the cause of your problem? Highly unlikely as you are using a form of NVRAM. The data should be retained when you turn off the machine even though the batteries are not fitted (which is the whole point of fitting NVRAM in the first place). Please note my previous comments – you can leave batteries fitted even if you have NVRAM – it won’t do any harm even though they aren’t needed.
This is going to be a bit tricky to diagnose I’m afraid – particularly as I don’t know how the RAM is mapped to the CPU. By that I mean that I don’t know where (within the RAM) that the high scores and player initials are stored. I’ll stick a couple of potential causes on here – none of which I’m particularly confident about – but they are the only things I can think of.
It is possible that you have a faulty area of RAM – but this faulty area would have to be the area where the high scores are stored. The reason I’m not happy with this theory is that it would mean that the high scores and player initials would have to be stored in different areas of RAM. The RAM is divided into 8 separate 1K blocks – so it is possible (however unlikely) that the RAM could have a faulty block.
It is worth bearing in mind that FRAM (as do all components) has a limited lifetime – but FRAM/NVRAM lifetime is affected by the number of times you write to a RAM location. This is known as Write Endurance. Admittedly, this number is big (around 100 million cycles) but modern software that writes to NVRAM take this into account and is written to spread the load across the device. For example, a memory location in the disk used in a PC (solid state disk – SSD) would have a significantly shorter lifetime if it continually wrote to (and read from) the same locations. This is why these devices have clever ‘wear levelling’ algorithms in them that spread the load and make sure that (as far as possible) all memory locations are used an equal amount of time. The Williams engineers probably use the SRAM to store all sorts of information many times – and they almost certainly didn’t consider NVRAM and wear levelling when they wrote their software. So, there is the possibility that the NVRAM has locations that have simply worn out. Again, this is unlikely but possible – and I can’t say for sure as I don’t know how the system partitions/allocates/uses the available address space.
Another possibility is a faulty address line (either on the CPU board or within the FRAM). This would have the effect of rendering an area of FRAM inoperable. As before, this may be unlikely unless the high scores and player initials are stored in different areas within the FRAM chip. You might want to use your logic probe to make sure that all the address lines (A0-A10) are toggling - just in case you have a cheeky open circuit somewhere.
As you can see, I’m a bit stumped by this one but I do have some suggestions (which have already been suggested in other replies).
- We know that the RAM is working at the basic level (or the machine wouldn’t work at all)
- We don’t know if there are faulty areas within the RAM (either due to faulty RAM locations or a faulty address line).
- We know that player names are stored – but not the high scores.
If you’d like to do this, a suitable replacement can be found at RS components (part number 538-158) – they are about £3 each. Please note that this chip has an extra control pin (pin 26 – Chip Enable 2). This pin would need to be pulled up to 5V by connecting it to pin 28. Let me know if you’re not sure what I mean, and I’ll attach some photos.
Yell if you have more questions. Good luck.
Last edited:
Fubar
Site Supporter
Sounds interesting - care to post some more details about this project? I'd love to hear about it. (In a new thread though!)
That will be because its the nearest score to the score you currently have. Normal behaviour for later system 11 games onwards afaik, if your score was higher than default 4 but lower than default 3 then it would say highest score - whatever high score 3 is.
I'll start another thread at some point. I'm currently looking for a donor TAF machine so I can graft the new electronics into the machine and completely ditch all the Williams electronics. I'll hopefully have something before the end of the year (unless someone knows of a donor/non working TAF on the market somewhere - meaning I can get it working sooner
)Good to know it's supposed to be like that. It's just misleading by saying "Highest Score" rather than "next high score " or "score to beat".
Bk2k has the same text but lists No1 so I wasn't sure.
Back to zeros
Since early May it's been fine, I assumed it was because I changed a transistor (C30).
Games been on and off lots. But today I was moving a smart plug around so was testing that. Diner went on and off a few times in the space of 10mins.
Then zeros.
As I thought it was fixed I didn't put the new NVRAM and new ROMs in but now they're in
The old ROMs look really old and as I took one out something fell.. but they still have all their legs and none of the ram sockets or nearby components look broken so I have no clue what it was.
Since early May it's been fine, I assumed it was because I changed a transistor (C30).
Games been on and off lots. But today I was moving a smart plug around so was testing that. Diner went on and off a few times in the space of 10mins.
Then zeros.
As I thought it was fixed I didn't put the new NVRAM and new ROMs in but now they're in
The old ROMs look really old and as I took one out something fell.. but they still have all their legs and none of the ram sockets or nearby components look broken so I have no clue what it was.
Ashbo
Site Supporter
That was the bug in the system. You'll probably be fine now ...... I'll get my coat....
