So, I've seen a few DIY threads popping up, so I figured I'd add mine to the list.

I've been working on a DIY pin for a few years on and off (mostly off, you know, wife kids chickens dog life etc).

Originally my theme was Big Trouble in Little China but the wife pointed out, the first time I do something I tend to mess it up a bit, so shelve the idea I really want, and go with the backup which is Skyrim! A TESV:Skyrim pinball machine.. sounds pretty good to me actually, and I have 200+ hours playing skyrim, so I kinda like that theme idea.

Sooo with theme in hand we proceed to the nitty gritty. I'm doing my own electronics because I think PROC is massively expensive for what it is and what you get. Right now the plan is a 3 board design. Master controller (UBW32 pic32 board) with custom shield doing the game rules, switch input and lighting. A Solenoid control board for driving up to 32 solenoids. All the A/V work will be done on a raspberry pi. Yes, it will be LCD. One of the first things I decided was that it would have some form of colourdmd and the only way to do it was an LCD.

I've bread boarded the solenoids for testing, so now I've got a schematic design going in Eagle. Once I have that back from being fabbed I can start work on the playfield whitewood. The solenoid driver is run by a propeller chip, which can run 8 threads simultaneously, which is fantastic for timing all the solenoids on/off with PWM at once. 1 core will be driving the HC595's that turn the solenoids on/off, the other 7 will each run the timing of 5 solenoids (last one only running 2, tallying up to 32). The nice thing is I have this all running inside the chips cache so its running at its full 80mhz speed. It spends its entire time basically doing millisecond timing of all 32 solenoids which is nice to detach it from the master board.

The raspberry pi takes commands from the master board that basically just tell it play sound effect X, play images Z etc. So far its working well. It feeds out to an LCD of 1366x768 size, which is a nice physical size for the backbox. OpenGL accelerated SDL2 with no X running is very nice, boots fast too, probably in about 10 seconds-ish?

It can decode mp3 and play wav files, outputting to a car amp with a sub and two satellite speakers in the backbox, so nice to have real stereo sound and bass going on! Its working really well so far. I have all the fonts, sounds, music and speech ripped from Skyrim. Its about 8gbs worth of data, obviously I wont be using them all. There is some really great call outs in there.

Powering this is two power supplies, a regulated high voltage psu just for high power coils like flippers, and a good computer psu, the computer psu will drive plenty of power for the LED's and micro controllers off of its 3.3 + 5v (30Amp), and its 12v rail has 70Amp which is plenty of power to drive the car stereo amp for the sound + bass.

I have a rough playfield sketch on my tablet, with way too much wedged into the design, but the design will have crossing orbit loops shots, drops, scoops, magnets, etc. (until stuff is removed) Upper playfield that will be kind of like a cross between the upper pf on WWFRR + SOF. A loop with drops, that let fall past them to the lower pf. I want more on there than I can get so its in a state of being culled down somewhat.. eg: Im not sure if I can get the drops on the upper, because that uses quite a bit of real estate below for the mech which would interfere on the main playfield. (I have a nice 4 bank drops from a Alvin G. Big House thats fairly slim, mechanism wise, think I can move the solenoid below the push bar to make it even slimmer.).

This is a widebody because... I had a widebody lockdown bar on hand so thats what I drew my design for...

Anyway, the plan is to start a whitewood as soon as I get a working solenoid driver board going.

I have a lot of mechs and things already, even if I dont use any of them, Ive collected various parts for testing and playing with to see what works and how, and how they might factor into the design I have. Not everything will have a place..

- several drop target mechs for testing (all gottlieb, 1 alvin g)
- the 3 LAH magnets for center playfield (I need to figure out how deep they are recessed into the PF)
- bally 8ball deluxe inline drop mech
- old ball trough from an williams widebody indy.
- stern auto launcher
- Some data east mechs (ripper, kickback, new old stock scoops, vuk, slingshots)
- williams flipper mechs with 15411 coils

I think the only mechs I really need to get are some pops. It seems that ebay pop bumper prices negate buying new, since you have to replace the yokes, plastics etc anyway, there is no saving over buying new and pops are expensive, oh well.

random note, the coin door already has token mechs and takes PAPA tokens just fine, configurable for free play or 1 to 4 tokens per credit. Do I do random things out of order? yeah... guilty.

I have some ideas for all the modes in the game. there will be some simple modes and some complex modes and a 'wizard' mode or two. some hidden things.

rough ideas from my notebook (probably using things not even in the layout, some very rough notes, probably far too skyrim knowledge centric too, which can be big problem)
- use magnets when the dragon priest activates (some coloured lightning bolt inserts will flash when active), dragon priest will be drawn in the lower pf.
- cave scoop, can start modes.
- draugr drop targets
- Mammoth Giant 'ripper' kickback in pf, those damn giants playing golf with you...
- orbits move which hold you are in.
- depends on which mode you complete in which hold grants what reward (eg: riften gives thieves guild, whiterun gives companions.)
- travelling to all holds gives wayfarer bonus
- some of the stationary targets will be guards, some bandits. LED will randomise red for imperials, blue for stormcloaks.. if that mode is on, otherwise could be bandits etc.
- right scoop for certain objectives
- left scoop for alternate objectives
- completing two objectives in same hold grants thane bonus, which gives a mini multiball to destroy dragon.

so for example
- in the rift, left scoop starts thieves guild mode (orbit shot in each direction with no other switches to stay hidden, then into the pops for picking lock. N pop bumper hits opens lock and finishes mode). right scoop starts a 7000 steps 'dragon shout' mode (spinners, not 7000 spins.. that would be crazy!)

Obviously this is very rough and random, and probably too complex.

Toys on the playfield.. I'm looking at some of the McFarlane's Conan and Dragon toys to see how things might fit.

I addeed some messy shots, a WIP solenoid board.. electrocuting myself with the flippers, and the spaghetti mess that is manually wiring the PIC32 to the PI to a support eeprom.. and two little screenshots of the WIP title screen from the PI coming rom the PIC32

omg wall of text!

sdrv.png
xflippers.jpg
xmess.jpg
s1.png
s2.png

Thanks guys. I am hoping to get the solenoid board out for fabbing on friday so I can really start in on the whitewood design.

right now I have a meanwell 48vdc/12A psu tuned to 50vdc. I plan on replacing it with the Antec one suggested on the proc site that does 70v + 24v

A little progress, solenoid driver test boards from the fab!

boards.jpg

got a huge batch of parts from mouser, time to start testing some PCB's!

Would I have used PROC if it was not "massively expensive", sure.

so lets see..

My goal is for my boards to handle 128 switches, 32 solenoids, bunch of gi + inserts etc.

I can buy 16 shift registers, 74hc165's for less than $4. But if I buy PROC boards its going to cost... $80 to read 16 switches, you need a total of 8 boards at $480.. Thats 8 x SW-16 boards for $480!!

Yeah 480$ to read 128 switches, do I think thats massively expensive? HELL YES.

Want to control solenoids? all it takes is a MOSFET to sink, you can buy 32 IRL520's for $24. Say it cost $20 to make a board, thats less than $50, but for PROC boards, you need two power driver 16 boards for $100 each at $200.

So far were what $60ish vs $680..

Want some GI? the PROC led board is not bad, $90 bucks, you'd need two, one for gi, one for inserts, so $180. I can buy 10 ws2803's, each one does 18 LED' so for $20 you can buy 10 of these, 5 for GI, 5 for inserts etc, make a board for $20, sure, so again under $50 there..

Thats $110 vs $860.. oh, lets not forget to add in the P3-ROC to drive them, thats $275 cost there.

I've got a $40 UBW32 pic32 board driving mine.

That brings us to $150 vs $1,135...

Maybe it will cost me $200 or $300 once I'm done.. Thats still much better than $1,135!

So whats not overpriced here?

Since this is a hobby, my time is not on the clock. I dont need to recoup costs and pay wages to employees. That makes an obviously huge difference which PROC are trying to make back in a VERY niche market.

If I could buy boards that supplied all I needed for $300 would I buy them? hell yes I would.

The Pinheck / Tommy Board by Ben Heck + Longhorn Engineer they want to sell for around 400$, but I think they are doing 64 switches and 32 solenoids. Those boards look really nice but not out yet, and they target a DMD, where I am using an LCD. I believe the final board rev is inside of Americas Most Haunted right now at the MGC.

If that was out now, I'd probably buy one and manage my design around a DMD, but when I started this, that was not anywhere near completion and I decided I wanted an LCD display.

I still think PROC is "massively expensive" for what you get.

well I was looking at P3PROC and it explicity says to use SW-16 boards, and no mention of a built in 4x16 matrix.

sure, you can use a solenoid driver to control lamps but thats overkill. I'm using LED's and they drive much better from proper LED driver chips.

1 - whats overkill about wanting things like attract mode sweep patterns, and individual light control over the GI?
2 - Absolutely will I be using RGB LED's under several inserts. RGB leds are not overkill.

Did you get any further on your nightmare befoer xmas table? that was a pretty cool looking design (do you plan on updating your website?)

Well I dont want ALL RGB, but some inserts I do, for example in Civil War mode, your either Imperial (RED) or Stormcloak (BLUE) and targets will change, gotta defeat all your enemies. etc. Full RGB on everything is overkill yes (imo WOZ is... overdone, I dont like woz). In another mode, same inserts might be yellow, green, orange or purple... It wont be a WOZ vomit of rainbow colour tho.

16 switches for $80.. yeah.. I've already said what I thought about it.

So, made a little video of the test solenoid driver, using LED's to simulate the solenoid going to GND so I can test and write code without melting solenoids I'm at a size limitation of my PCB software so I need to pay out some cashola and its a fight in my head of what I'm going to do which kinda sucks.

The vid shows the driver hooked to a test MCU and testing the all important watchdog chip, which should something cause the system to go screwy, will shutdown all the solenoids so nothing burns out. (basically it takes in a heartbeat that must flip/flop every < 60ms or it triggers.)

I am looking at a new driver chip, this one would drive two FETS at a time instead of one as currently seen on the board. Its also cheaper, so I reduce the chip count by half and cost by nearly half as well (plus its faster, but were talking 12ns vs 30ns)..

yep, I have Eagle hobbyist license so 6.4"x4" (160x100mm).. I've been testing DipTrace and its very nice but not Linux native, it runs under wine which sucks, but I can run it under XP on virtualbox which I dont really want to do, however I can create 100 new parts in it in the same time it takes to make 1 new part in Eagle, man if I could get Eagle with DipTrace's footprint/parts creation..

Last time I looked at kicad it was a steaming pile of ........... I should really look at it again, but I dont fancy having to create EVERY single part I use since it has virtually zero existing standard library support.

so yeah, its all or nothing for Eagle $$$, I get a native linux app, and a workflow I'm used to and large existant part library (thanks sparkfun + adafruit, etc)... or I pay for DipTrace and get say 1000 pin limiation, but making parts takes like 5 seconds but I have to run it under virtualbox..

oh man its decisions decisions.

I dont mind the eagle ui, just have to understand its modal. Its not the best, I REALLY_ miss right mouse click to pan.
I tried kicad again last night, its still an abortive mess. going to try and convert one of my schematics on the weekend and give some real time to it and see how it goes.

I spent several hours this weekend on kicad, I built it form source for the latest version from git.. ooh man is this freaking terrible. Everything is so manual I might as well edit everything in a text editor and manually type IC1 X=1.50, Y=2 it would be faster. The whole workflow is REALLY bad. find parts, mess around with the schematic.. create a netlist.. try and find a footprint for every part in the netlist, there is no connection from a part to a footprint, you can attach a 100 pin QFP to a 4 pin dip schematic part, then mess around with that in the pcb tool.

Some file, fp-lib-table didnt exist.. whats that? dont know. lot of googling led to a script someone wrote on eevblog, run that against the /usr/local/share/kicad/modules directory and move it to where your schematic is. apparently you need it but kicad will not create it! oh.. .kicad_pcb file didnt exist for my schematic (since, well, I'm making a new schematic), so it will not edit the pcb layout... so lets manually touch that file.

no thanks. I'm done. This is not a workflow I want to get into.

I really want Eagle. My existing schematics are in it. I use the nice DRC rules from Sparkfun, I have all my parts, my workflow, key shortcuts. The cost is a real punch in the guts tho. I can upload a board directly to oshpark without messing with Gerbers, which is nice!

I now have a much better appreciation for what I'm paying for. I love open source, heck I write open source software too but for getting work done, I don't want to have to fight the tool and the workflow is really bad. maybe in a couple of years, who knows, but I think right now its not for me.

I did learn a new trick for Eagle that I wish I knew about ages ago!

With Eagle closed, if you edit your ~/.eaglerc, add the line 'Interface.UseCtrlForPanning = "1"' or with Eagle open, in the command line of it at the top of screen, type 'SET Interface.UseCtrlForPanning 1' then just hold control down and move the mouse... YES damn easy panning, why is this not the default or even documented? aaaugh.

anyway...

So I have spent the last two weeks testing Diptrace and Proteus8 in virtualbox (and in wine, for diptrace just to see), making the same example schematics of another board I made, (hard in Proteus8 as the demo cant save!).. Eagle import into Diptrace is ok-good, not fantastic (labels and other things dont come across into the PCB design and such), Proteus8 has no Eagle import. Proteus8 seems more about the simulation that creating schematic/pcb it felt like. DipTrace is very nice and easy to work with but feels... less powerful?

Eagles price is just too hard to swallow, thats not even adding the autorouter on (which is nice but not required).

Which leaves me the option of DipTrace, which is basically 300 pins (75$), 500 pins (145$), 1000 pins (345$), 2000 pins (595$), unlimited (895$). I'd probably look at 1000 pins. I've a few hours converting the master board schematic and started adding in the solenoid driver and hit the 300 pin limit

I really dont want to mess with virtualbox and want a Linux native app.. I can live without autorouter... its still 1000$ vs 345$

Eagle 7 is going to drop sometime soon... and diptrace is getting some new library management features, gui updates, and other things...

In the end, I stumped for DipTrace 1000 pins, just today inside a few hours, I have already migrated my master control board and full sized solenoid driver board completely over.

Not a lot happening this week, just kinda completing the design of the master control board, at this stage its just going to be the switch matrix (16 x 74HC165s), fed into the PIC32. Rather than daisy chain them and do 128 reads, the pic32 has 80+ pins, so I'm going to use 18 pins, 16 pins to directly talk to each chip and 2 pins do to the clock/load. This way, I can get all 128 switches in 8 reads and its nice and fast. Put the read routine into an interrupt it should be able to read it very very fast.

Right now I just need to figure out what kind of connector I'll use. The good old locking molex KK 1.56 spaced connectors are HUUUGE on PCB space. I'm thinking I might use simple 2 row .1" pitch pin spacing shrouded header with the good old IDC type flat ribbon cable. I do want to stick with whats commonly used, thats KK 1.56 stuff. ugh.

this is an older image of what it looked like last week using the .1" pitch x 16 pins. there are no standoff mount holes and some other things missing from the pic here as well.

pcb3.png

Slowly pushing some stuff to github. Not a lot there, only the solenoid driver stuff right now, DipTrace schematics and the source for the driver. There is a routed pcb but really needs re-routing.

Trying to ducment stuff too (thats a mishmash as I'm pulling stuff from my master Skyrim document, where a lot of design detail doesnt need to come over)

(DipTrace has a free version that will let you view the schematics as far as I know, but thats it because the pin count exceeds what the free version has.)

https://github.com/stu/system_shock

All source code is GNU GPL-v2, all hardware schematics/designs are CERN Open Hardware License v1.2

I want to keep this fully opensource so everyone can benefit/contribute/use etc.

spent the weekend routing the master board.. ugh. happy with it now. I'll sit on it for a week and look it over a few times then get it fabbed. I decided to go with the KK-254 molex locking connectors, basically the 0.1" version, the little brother to the very common in pins KK-156 0.156".

aaanyway.

Once I'm happy-ish I'll add the master schematics to the git repo and write up some more documentation and push it to the public.

Ideally later on I'd like to replace the DIP40 propeller with the QFP version, save a HUGE amount of pcb realestate... and probably replace the switch 165' shift registers with SOIC ones and again save huge space.. but I really HATE trying to hand solder the SMT stuff. With everything being DIP, right now anyone could build a board and repair it, going to some SOIC parts and QFP-44, meh, makes it that much harder.

anyway, gonna sit on it and review it a couple of times this week and see if anything jumps out.. tired of looking at it right now

pcb4.png

thanks for all the feedback and well wishes. Lets answer some questions;

ecurtz; I'm not an electronics guy either, I'm a software guy. The resistor networks are pretty tiny. They shouldnt get in the way, those molex connects dont wiggle about so should be good... and if not, thats why you create another revision after testing

blondie7575; Those 0.1" connectors are ONLY for the switches. The solenoid/flashers use the large KK-156 connections that all current pins use, so power on those should be no problem. You can see the larger size on the solenoid board.

pcb5.png