I have a smaller/cheaper one from Oitez: EScope. Cost a bit over $100 10 years ago. Has been fairly valuable for SMT stuff. Can take pictures and even movies.

Microchip Details (resized).jpg

eScope (resized).PNG

Circuit Board Pricing: $1,000 for a new SPIKE II CPU isn't totally out of line. Factory markup for PCB assemblies runs about 6X I think. There might be $150 or $200 worth of parts there.

If you need a new WOZ CPU they are priced at $200 from JJP, if they had them (not available now). It is an MSI H81M-P33 Celeron motherboard and you could get one from Newegg (again, if they had them) for ~$120. Why? Because these are mainstream PC parts, not vertical application-specific parts.

I just bought a Multipmorphic SW-16 I/O board for $45. Pretty simple board - probably right at that 6x markup.

WOZ_CPU (resized).PNG

It is a fair comparison in the sense it is what Stern's competition is offering (JJP) for WOZ.

As another data point is American Pinball. Houdini uses Multimorphic boards. I suspect the CPU they use is the P3-ROC which can be replaced for $175 from Multimorphic. Not sure if that can be loaded with Houdini code by a user. They don't elaborate on this on their site.

600% is normal factory multiplier. A new ECU for my 2001 Honda Civic costs $800-1000. If that's what you need then that is what you buy.

Adding a PC or SBC to make Multimorphic P3 work? That changes the math a lot. But long term the more general the solution the more likely it is to be around after a while.

I'd bet that one of two things will happen: 1) Someone will entirely reverse engineer the SPIKE II platform and sell it themselves or 2) Stern will release lots about it some day in the future so people can work on them.

SERVICE PARTS (Spares) COSTS

If you're building something - say a clothes washer - then here are some build quantity thoughts:

Suppose your target building 1,000 units at the current revision for market. You actually have to build a large percentage more on the line, say 30% more based on reliability information.

These "spares" will have to be packaged, cataloged, managed and stored for the "service life" of the product. This adds directly to the cost of the 1,000 units you plan to sell.

Each revision will need its own set of shared/unique parts to be stored. What prompts revisions? 1) Parts availability. Manufacturers regularly issue "End of Life" warnings for parts they are done making. Users that count on these parts are usually offered a "one time buy" opportunity to get what they think they need to support this washer revision. 2) Better designs. Moore's Law has made giant leaps in price/performance for electronic stuff. People expect more and newer parts can do that faster and cheaper. 3) Manufacturing efficiencies. Improvements are constantly made to the "current" product that improve its price/reliability/manufacturability.

These spares have a large percent-contribution to the factory % markup, part of that ~600%.

The problem with pinball is that the customers expect the thing to work forever. If/when this becomes impossible then the customers will have to adapt. That, or replacements will have to be available through other avenues so they can keep it running. I'd say the ARM family of CPUs actually make this much more possible.

So: put down that cell phone and start studying electronics! ( - :

Another thing to keep in mind:

Remember that JJP Celeron motherboard I mentioned? You can't get one from JJP or Newegg. So what do you do if that fails? Start scrounging for parts to fix your 2013 WoZ? I'd say so.

Stern's stability hasn't forced this burden on their customers, as of this writing. You can pay "a lot" and keep moving or scrap it.

Oh my. That shifts the table a lot.

Hum, there have been lots of chipsets. Do you anticipate trouble in the future finding this one in current-production motherboards?

Keep the info coming, I'm learning about machines I don't have regular access to.

The 2 big elements of "chipsets" are the North Bridge & South Bridge. I did PC servicing for a number of years and I was constantly fighting the "wrong drivers, BSOD" problem when replacing a motherboard for an existing installation. The only way around this without a total reinstall was what we called a "kernel repair" operation, it could use the basic BIOS to get to the point where it selected the proper drivers for the new chipset.

Actually it looks like the best idea is to use "simple" boards like FAST or Polymorphic with the primary controller being a compute platform using USB. This level of hardware abstraction allows lots & lots of flexibility.

Funny but I've never worked on projects where silicon availability long term was a primary consideration. Pinball really demands this.

I feel your mods could either help or hurt, flip of a coin.

If there are part leads that are hanging out then giving them mechanical support (silicon?) would help.

And manuals describing the function trees that comprise the service menus. Pretty basic stuff. Figure it out yourself.

Each via (top/bottom interconnect) requires drilling. My 1" x 2" PinRemote PCB has roughly 25 of them, all drilled. As complexity, parts count and space conservation increases so does via count.

Just for the record.

There is a lot of conjecture here that is offered without proof, which is part of human nature. "Lord Mounts", or vibration isolators are designed to do just that, but resonances rule the day and they can easily make things worse if the system isn't understood.

The problem is two pronged: You must understand what vibration patterns are produced (pretty hard for a pinball machine) and then design PCB mounting to counter those. This is beyond the scope of pinball manufacturers.

At HP in the late 70's I did environmental testing on a desktop calculator product. We strapped it to a special table where we could control the frequency, intensity, mode (direction) and duration of the shaking we were going to subject the unit to. Most tests were pretty benign, the unit just ran quietly. There was one test though, where the operator scanned through the frequency spectrum at will - looking for settings that "produced the most noise". The most noise meant something inside was at resonance and was getting maximum deflection. Sure enough, 30 seconds later or so the machine would die and the test was over. Find the parts that were broken off the board (usually large caps in the power supply) and devise some way to secure them. In some cases a PCB was attached at both ends but not in the middle so it just bowed like crazy: it needed support in the middle.

I enjoy reading about adapting drone ideas to pinball and seeing what parts are being used to do this but won't be modifying my machines this way as some sort of crazy experiment. I'm just going to trust the manufacturer - and think about replicating the boards myself some day.

I did say this type of testing is beyond ALL manufacturers.

Stern "invented" the idea of letting the customers test the product, and they don't even ask those same customers to send in log dumps to help them fix the products like some newcomers do.

Someone else coined this saying: PINBALL IS HARD

We had a tongue-in-cheek saying at HP: "Don't comment code. It's hard to write, it should be hard to read."

The biggest comment I remember from the 9845 HP project was "DML [date]" Ah, Dave Landers wrote this and later changed it. Funny now. Worthless.

Later we would see comments like "implemented as a singly-linked list" only to find out later that it was doubly-linked.

Or "disk I/O queue sorted by shortest-seek-time-first" only to determine that it is done by user job priority, then biggest writes first.

Now I only comment the tricky stuff and always in some snarky way.

Writing code to detect that is a bit trickier than it might seem. Do-able, but then you delay intro and people get pissed at you ( - :

Hum, wonder if my Instructables "Carvey" could do that...

Nevermind, it is hard enough troubleshooting the circuit without having to troubleshoot the PCB!

Great stuff you are sharing here Mr. Walker

I agree that the lamp/switch/relay stuff is pretty easy, but the graphics (like WOZ) and sound can get pretty CPU intensive.

Good to know, thanks.

My focus is on the most venerable foundation to build things on so they don't become obsolete. Celeron? (JJP), maybe. Whatever Stern uses? Don't know but probably not.

The ARM family seems to have pretty solid legs. If, as you say, it doesn't take much CPU then build your foundation on that family.

Slightly off subject, but this is the type of info I like to see about technology families. A "Good Until" date.

Nice processor, think that P3ROC suggests this one.

SBC (resized).PNG

I don't think the processor is the issue per se, it is the chipset

Since your abstraction layer is USB - of course. Port a USB driver to VAX/VMS and use that.

My point is for the other things that comprise a pinball machine the user community needs products with legs. Ones that can be around for as long as these things live which appears to be Methuselah's age ( - :

Did find this picture today which talks about a logical division of tasks. I'm talking about the Media Controller (right/red) part:

Pinball Controller Task Split (resized).PNG

Trick answer, but CGC does for MMr & RFMr