I'm pretty excited about this.. it's a project that I've actually had sitting for over a year now. I've wanted to explore other ways of possibly replacing NVRAM without removing the original RAM chip from the game. Well, here it is... upgrade your game to NVRAM (no batteries!) without unsoldering the existing RAM!!! Just minimal modifications are needed to do this and the modifications can even be done while the board is still in your game if you're comfortable with a soldering iron!

sys11_nvram.jpg

As many of us are aware, Williams System 11 games usually had their RAM chips soldered directly to the board from the factory. To replace the original RAM with NVRAM usually requires the RAM to be cut off, pins desoldered, holes cleared of solder, and a socket soldered back onto the board. That works & is a clean solution, but it's out of reach for some people & there's always the risk of pulling solder pads/traces or otherwise harming your MPU board. I sell 6264 RAM replacements on eBay and normally recommend if your MPU is fully working, leave the RAM soldered in and use a remote battery pack. I wanted to offer another option though.. and I think this will work great for some people.

Installation / Minor Modifications to MPU
This NVRAM adapter board plugs into one of your game EPROM sockets and the game EPROM then plugs into a socket on the adapter board. You then snip a leg on Q40, pull the leg up and solder a wire to it (this is the chip enable signal for the RAM). Next, you solder a wire to U31 pin 8 (this is the read/write signal for the RAM). Then just connect those wires to the adapter board & you're done! No more batteries!

This mod should take less than 5 minutes for people experienced with a soldering iron since there is really no need to pull the MPU board from the machine and there are minimal modifications being made. The original soldered-on RAM does not need to be removed. It will stay disabled. If you sell the machine down the road, you can easily reverse these modifications and re-enable the on-board RAM.. then use the NVRAM adapter in another machine.

Steps can be seen below. Yes there is some soldering involved and you should know what you're doing to attempt this type of mod, but for many people this is far easier and safer than cutting RAM off the board and should be a piece of cake for anyone that is comfortable using a soldering iron.

***PRE-ORDERS & COST***
I'd like to get an idea of how many boards I should have fabricated, so let me know if you're interested by commenting on this thread or sending me a PM! If there's a large enough response, I'll make the PCB order & should then be able to start shipping them in about 2-3 weeks (takes a bit of time for boards to arrive & then will take a few days to assemble).

http://www.pinitech.com/lab/sys11_nvram.php

Cost will be $30.00 shipped. I'll include a replacement transistor for Q40, wires needed for modifications, fully assembled adapter board with NVRAM and instructions. Discounts will be given on 3 or more adapters.

I'm hoping this goes over big because I would really like to see this happen. Please contact me or comment on this thread if you're interested!

Steps..

step1.jpg step2.jpg step3.jpg step4.jpg step5.jpg

Picture of adapter "installed" in a Williams F-14 pinball. Please ignore the red/black wire with molex (that was the old battery connection).

Also, please note that in pictures mini-grabber clips/wiring is used. You could *theoretically* use mini-grabbers, but I am not recommending it. Instead, shorter wires can be used, possibly routed closer to the board -- trust me, I'll figure it out This was simply for testing purposes & snapping some pictures. I'll try to get a slightly nicer looking mock-up later this week that doesn't have the mini-grabber clips.

inside_machine.jpg

I'm using mini-grabber clips for testing & this is a prototype board. You *could* use mini-grabber clips & not have to solder anything (just cut the leg on Q40 and that would be it).. however, I'm not going to recommend that over soldering since it's not a very secure connection.

Hey if it works Honestly, I had mini-grabber clips hooked up in a Road Kings for MONTHS and it worked just fine. I'm just not going to *recommend* it as the way to do it.. but if you really hate soldering it would probably function fine for a good while. Just don't bump the machine a lot

Something else I can mention, but AM NOT RECOMMENDING ANYONE ELSE TO TRY.. is when I hooked this up initially without cutting the Q40 /CE line and with batteries still in the game..it had actually dumped current RAM contents to a new NVRAM. High scores, etc were written to NVRAM. I haven't tested this enough to know if it always happens or what situations (ie. brand new NVRAM versus NVRAM that has already been initialized). I also know it goes against any and all electronic principles since both RAM chips would be enabled at the same time. Just thought that was pretty interesting. DO NOT TRY THIS as it is untested and data corruption may/WILL occur.

Also I just want to say that I did look into this from a PLD (programmable logic device) perspective -- CPLD or FPGA. It would require *way too many* logic chips to be replaced. This really is the easiest way I've come up with to do an NVRAM modification without removing the soldered-on RAM or replacing a huge amount of logic happening on the board.

For Williams System 11B/C the R/W line is *slightly* easier to tie into. I'll mention an alternate way in the instructions. Still requires soldering, but not to U31 pin 8.

Understandable. I was *really* hoping for a truly plug-and-play solution as well and that's why I've sat on this for over a year. I recently had a chance to look into PLDs more and at what components would need to be replaced to recreate the R/W + mem protect & /CE lines for the RAM if I created an adapter board that fit into the CPU socket. Way too many chips to then duplicate the function of, including buffer chips on the CPU lines that would also be needed on some type of plug-and-play solution.. it's just not feasible when those signals are already on the board and could be tapped into with minimal modifications. The $0.10 transistor has to suffer! But I'll include a replacement transistor for free.

Great question. I just took a quick look at Space Shuttle schematics and I think this would also work in System 9 MPUs. The adapter would go in U20, the R/W would come off U43 and there's a 2n3904 transistor setup for the chip enable that is the same as on the System 11.

Hey Mark, that's pretty cool. What did that kit cost? I'm seeing prices as high as $160 for some of those multigame kits. I didn't see a price for your version of it though via a quick search. If board size + additional components + additional time for assembly are all factored in and you can compete with $30 NVRAM adapters and have a completely plug-and-play solution, I'd say go for it. I just thought I'd offer an option that is not out there yet for System 11 machines.

I actually expect to be able to drop the price to $25 + shipping or possibly less if enough people are interested that I can get a larger board order (100-200 boards). If not, PCB board prices in smaller volume will be greater and that needs to be factored in.

Yep, it's pulling up A11. I actually thought about killing power to the original RAM, but did not know if there was a chance of some type of feedback or interference messing with it.

I had figured the problem with grabbing /CE off the emitter is that Q40 transistor is stopping the RAM from getting enabled until the MPU resets properly. Well according to what I see in the schematic anyway. So I decided.. easy enough, cut the collector lead on the $0.10 transistor... the /CE on the original RAM then stays pulled HIGH via R64 (this disables the original RAM). Then solder a wire to the collector on the transistor (/CE signal) & run that to the adapter board. For the read/write signal, you should be able to use a solder pad on the W5 jumper on the board on the System 11B/11C boards. The pure System 11 & System 11A's would need the wire soldered to U31 pin 8.

Really it's nothing *overly* complicated, and I would love for an even better plug-and-play design, but the CPLD stuff seems overkill. I know that there will be some people that don't want to do *any* soldering/mods to their board. But that's what mini-grabbers are for -- only need to cut one transistor leg!! (joking of course)

Yeah I've noticed prices on Ramtron rising since the announcement that Cypress was buying them. It was a few bucks cheaper per unit last year. Luckily I bought a bunch a while back, so I'm hedged for a good bit. It's also not the quickest way to make a few bucks, but I've enjoyed creating this type of thing & getting some other options out there for people. Plus I get to use it in many of my own machines and avoid having to switch out batteries each year I really really hate batteries after having to clean up a ton of alkaline damaged boards!

It's still a possibility, just has sat on the back-burner with everything else I've had going on in 2013. I'll be looking at this some more early this year in terms of tweaking the design/instructions slightly so the solder points are a bit easier. I'd really like to release something that doesn't require desoldering of existing RAM, but also want to be comfortable with the modifications I'm asking people to make and have it be as simple as possible.. otherwise it's just going to be a "install it for my own use" type of thing. I've had the prototype that is pictured in this thread running in my F-14 for the last 9 months & it has worked out great.

Half of my machines are converted to NVRAM (Data East and any that had sockets at the RAM already and a few I did when boards were out for other repairs). The other half, mostly System 11's and WPC, aren't converted. Something like this would take care of all the System 11's. I could easily remove boards from my machines and install sockets in all of them, but unless I have the boards out for other repairs then it's not on the priority list. Plus I know that's the situation other people are in and would like to come up with a different solution.. so I'm leaving my own machines as-is for motivational purposes

Thanks for expressing interest!

If this keeps receiving some interest I may very well just have to go forth and make a few dozen of these adapters up! I have been back and forth on completely writing off the idea, but seeing some occasional interest on this thread out of nowhere has me back to considering offering it for sale.

It shouldn't take too much to effort, the existing design has been working fine all this time in my F-14 and I already have the majority of the instruction manual done from when I worked on it about 10 months ago. This would be a "modify-at-your-own-risk" affair though and I'll need to make that very apparent on any product pages and throughout the instruction manual. I don't see it being a big issue if people know what they're getting themselves into (ie. soldering and slight board modification). It's not plug-and-play that some people need, but it's a light amount of soldering for those that know what they're doing and it's quicker and far less invasive than cutting or desoldering RAM off the board. Heck, people modify their own video game systems with mod chips, etc that have dozens of wires -- this is two wires. Ideally I'd love to have it be completely plug-and-play in a CPU socket with some CPLD but I'm not quite there yet with designing something like that -- maybe later this year or next year I'll explore that option more, but it's also going to make the adapter more expensive whereas this would both be available now and can be offered for sale around the same price as a standard NVRAM adapter.

I think I have an alternate installation method that involves adding a few more components to the NVRAM adapter board, but won't require cutting the leg of the 2n3904 transistor on the MPU. If it works, that method would be 3 wires instead of 2 and also would need a resistor to be soldered in parallel to a resistor already on the MPU. All of that would then be fully reversible.

I should be able to allow the NVRAM board to work with either method, just hook up via whichever method suits you. I'm going to test this out today and will report back.

I was able to use the alternate method successfully, however it means changing the pcb layout to allow for both methods of installation. So the good news is there's the potential for a few different methods depending what is done with the board design.

I don't have any estimate on availability. This isn't something I want to rush out the door without having adequate documentation and then as a result have my email inbox full of installation questions from customers.. since, I just don't have the time to devote to that right now. My schedule is getting busier now that warm weather is on the horizon and some of the pinball projects I've been working on are going to have to take a back seat for a bit again. Going forward I will likely be reworking the board design, updating the documentation / installation steps and then look for a few people to test everything out and provide feedback. Then if everything pans out there I would open it up for orders.

It's also difficult gauging interest on something like this here on Pinside. If this was available NOW based on the people that have commented on this thread and via PM I could maybe sell 12-15 units immediately, then I'd have a few sales here and there each month after that. Selling a small handful of something like this isn't really enough to justify making it a top priority or having time wrapped up in supporting it and I've come to terms with that when creating products. When I had initially announced this I was hoping for larger interest and to be reasonable sure I could sell 50-100 units fairly quick. If that was the case, it would have been fast-tracked and available a few months after the announcement. But, there's a number of things about a product like this that limit an already small audience, so I can understand why its only received a relatively small amount of interest.

Thanks for the feedback, I like getting perspectives like this & the non-electronic people are definitely a huge consideration. Even though people mod video game systems and other electronics that involve a lot more wires and hacks, wires are often hidden from view and the modded item likely wouldn't be traded/sold to different owners as often as something like a pinball machine.

The clips in the pictures weren't going to be the way I would recommend installing, technically you could use them but as you mentioned what happens when you sell the machine? Instead there would be several solder points on the board to tap into signals that cannot be grabbed at the game ROM socket.

Method #1 was cutting the collector leg on Q40, bending it up and soldering on a wire to pick up the chip enable, then soldering a wire to W5 (jumper) on System 11B/C MPUs to pick up the /WE signal. So that was a 2-wire installation, but involved cutting a transistor leg.

Method #2 would be a 3-wire + low-ohm jumper installation. A low-ohm jumper on R65 would be required to drop the voltage at the base of Q40 so that transistor doesn't activate and on-board memory is disabled. The /WE signal is again picked up off of W5 on System 11B/C MPUs. The two remaining signals are /RESET (at R68) and /RAMCS (at U35 pin 6 or the Q40 emitter leg). These signals would go to the adapter board that contains on-board chip select circuit (re-creation of the Q40 chip select circuit that's on the MPU).

In any case, although these methods are feasible for a board like this.. I still have my own concerns about making something like this available as a product and supporting it. It's asking for people to make minimal solder connections to their MPU to install it, but there are varying skill levels out there and someone that hasn't really done much soldering attempting on an expensive MPU -- well I don't need the headache if something gets hosed despite putting 100 disclaimers all over the place Plug-and-play seems to be the way to go with this stuff and also makes it available to a much wider audience. So, I may just continue sitting on this design, calling it a proven concept and instead work on a plug-and-play CPLD design down the road. If someone else is already working on a plug-and-play NVRAM design that won't require removal of existing RAM, please let me know so I'm not duplicating efforts if and when I start putting some time into that idea.

This design never went public so I'm the only one with it running in a machine. Never sent out any prototype units even. Anyway in my F-14 it's been working well, even with a non-permanent installation (mini-grabber clips).

I could probably map out 10 ways of picking up the signals or otherwise not having to remove your existing RAM.. some more invasive than others. If I add a few components to my board, you wouldn't need to cut the transistor or solder onto its leg... which I'll admit is a little sketchy... but you'd have 3 wires and one other component to solder to the MPU. I'm looking at some alternate methods of acquiring some of these signals like piggybacking them at the on-board RAM and I have a few ideas in mind. If I do release this as a product, I want to be outlining the easiest methods of installation in the instructions or have some clever way of making the install much easier for people.

A CPLD design at the CPU socket as mentioned would be the most plug-and-play method, no wires needed. Problem is, I haven't worked with CPLD's yet so there'd be a big time and financial investment there which I'm unable to make right now. This along with cost of investment in stocking up on *whatever* CPLD was used to avoid immediate obsolescence kind of has me putting that type of design on the back-burner and instead going back to the "pick up the signals already present on the board" design. I don't see recreating signals and adding $20-30 more in cost to the board to achieve the same thing a simpler method may be able to.

Anyway probably more than you ever needed to know when you were just wondering if someone had installed this

If you're interesting in testing a prototype if/when I have one available, send me a PM or let me know here. That goes for anyone really. I may do a small test run of whatever simplified method I come up with later this year and send them out just to see what people think and to judge if it's simple enough or documented well enough to sell as a product.

---
http://www.pinitech.com - "Pinball Inspired Technology"
Kits, upgrades and test equipment for pinball machines

Still on-hold right now. I have a couple other tweaks to this idea that I'd like to try, but will not have a chance to get back into it until after Christmas. Possibly getting back to this a bit sometime in January.

I'm still keeping in mind there are some people that wouldn't really mind soldering a few wires on their boards.. and that as far as modding goes on things like video game consoles, etc.. it's a pretty light amount of modding compared to a few dozen wires. I've just been back and forth with ideas trying to make it as plug-and-play as possible before releasing anything.. if I do release this one.

---
http://www.pinitech.com - "Pinball Inspired Technology"
Kits, upgrades and test equipment for pinball machines