7/30/2015, OPP, the second generation ordered

This week there was a good amount of time to work on the next generation of OPP hardware.  The hardware design/layout is all completed and was ordered on Wednesday through ITeadStudio.  The cards should show up in three to four weeks.  All of the parts from Mouser have been ordered including the processor cards.  The processor is separate so all of the embedded development can happen even before the cards from ITeadStudio are received. Here are the major changes that are in this new version of hardware:

• The main processor has be changed from a MC9S08SE8 to a PSoC 4200 processor.  The original processor’s package has been End of Life’d.  That means either changing packages (which would mean moving to a surface mount package), or switch to a different processor with a development card so all the cards can keep using through hole parts.  I chose to switch processors.  The choice was then between the Teensy LC, STM32 discovery boards, and the PSoC 4200 Prototyping kit.  They are all Cortex-M0 processors, but only the STM32 and PSoC are 5V processors.  (Useful when enabling/disabling FETs without needing to voltage shift).  The STM32 boards have a larger footprint, so I ended up choosing the PSoC 4200.  Here is a quick link to the prototyping kit from Mouser.  The prototype kits are $4 which is below my target price.  Since it has 32 useable I/O pins, each card can support up to four “wing” boards.
• The original processor required a debugger to program the chip.  The new processor has a bootloader built in at the factory, so no debugger is needed.
• Connections between cards was previously accomplished using a 4 pin 100 mil header.  The new boards will use an eight pin ribbon cable to make wiring between cards easier.  The 8-pin ribbon cable adds a couple extra signals including 12V and a synchronization pulse.
• The solenoid cards used to use an individual spade terminal to kick the solenoid, and a 2 pin 100 mil header for each switch input.  The new solenoid cards use a six pin 4.2 mm spaced Molex connector to kick the solenoids and return the solenoid current, and a single 4 pin 100 mil header for all the solenoid inputs.  Having only two connectors will make it easier to remember the location of the connectors.  This was particularly a problem with the input card since it contained 16 individual connectors that all looked the same.
• The input only “cards” now have a single 8 pin 100 mil spaced connector for all the inputs.  There is actually no board needed for the input only functionality, the 8 pin header is simply soldered into the PSoC 4200 board.
• The new board supports an incandescent wing board.  Previously, the incandescent drivers were driven using a SPI bus which was bit banged from the main computer.  The new incandescent wing board can support eight incandescent bulbs, and communicates using the normal UART bus.  A synchronization signal has been added to insure that all incandescent boards can synchronize their blinking if desired.  Instead of using four  2 pin 100 mil spaced headers to drive the bulbs, a single 8 pin 100 mil spaced header is used.  Again, this makes it easier to remove and re-install cards and keep track of wiring.
• A card with support for a SPI bus or a Neopixel driver.  After talking to a bunch of people at Pintastic, many people were interested in driving WS2812 LEDs.  I have shied away from this for a long time because I don’t believe I have the artistic ability to utilize colored LEDs.  It is very simple to use the SPI interface which is part of the 4200 to command the multi-colored LEDs.  The 4200 running at 5V means no level shifting is necessary to talk to WS2812 based strip LEDs.
• Previously, each board (input or solenoid) had a different load of firmware on it.  The new boards have enough flash memory that a single version of firmware will support all of the “wing” boards.

The PSoC 4200 has two 8 bit wide ports on the left side of the card, and two 8 bit wide ports on the right side of the card.  (It also has a USB to serial port converter on top of the card, which kind of looks like a head because it is made to be easily snapped off).  Two “wing” cards are soldered to the left side ports, and two “wing” cards are soldered to the right side ports of the processor, making the whole setup look sort of like a dragonfly.  Up to four wing cards can be populated if desired.  In some cases, it might make more sense to add another PSoC 4200 board to keep wiring lengths to a minimum.

The new mashup PCB which is in the repository contains the following “wing” cards:  three solenoid drivers, one input board, two incandescent boards, and two interface boards per PCB.  The PCB is laid out to fit on a 10 cm x 10 cm board.  Each solenoid board can drive 4 solenoids.  The input board can drive a single SPI interface (probably used to drive multicolor LEDs).  Each incandescent board can drive eight incandescent bulbs (or LED bulbs).  The interface boards contain ribbon cable connectors for communication and a communication connector that matches the first style of OPP cards for backwards compatibility.  (Two pins are reserved on the ribbon cables to insure that the cables can’t be plugged in incorrectly.  These could easily be converted to support differential transmit/receive signals if I find it is necessary.)

Here is a quick picture of the mashup PCB:

There was a Pinside thread which discussed the differences between PROC and Fast pinball controllers.  Both PROC and Fast controllers are great solutions and available right now.  During that thread, both of the people that have the original OPP cards chimed in about using the OPP hardware.  (Thanks guys, it is always nice to hear that you have had positive experiences so far.)  The original poster mentioned that the OPP hardware was not going to be a possibility because there is no “Buy it Now” button.  I have stated to many people that I am not going to manufacture cards because I work on pinball as a hobby.  That being said, how would a person get the OPP hardware that they wanted?  Up to this point, all the people have contacted me via email, and if I think that they have a real project to work on, and they sound like they have a chance to finish the project, I agree to send them out hardware.  (In all those cases so far, I have both populated and tested the boards)  That is not a sustainable model in the long term.  My current thoughts are that people contact me, and I send them out the bare boards, and then they order the parts to populate the boards locally.  (Through either Mouser or Digikey).

When ordering boards (an order costs $30.50 which includes shipping to the US), ten copies are received from ITeadStudio.  (Each ordered board contains the eight individual cards).  I believe that it will take 2-3 PCBs for a pinball machine (depending on the desired features).  So for a single order, you can build approximately three pinball machines.  It doesn’t make sense for each user to order boards themselves since they will only use 1/3 of the order.  That is why if they contact me, I can take care of sending them the boards that I have available and they don’t need to order from a board house.  That should remove many people’s fears of needing to learn Kicad, talk to a board house, etc.

Let’s talk about the people that should not use this board set.  The board set does not support DMD interfaces, or old style displays.  It is assumed that a computer (such as a Raspberry Pi, Beagle bone black, etc will front end these cards, and will provide display capabilities).  The board set is tailored towards wiring a machine from scratch.  I would like to encourage people to put these boards below the playfield to make the wiring as minimal and simple as possible.  The board was designed as a distributed system, and bringing all the wires to the back box removes most of those benefits.  The one concrete reason why this board set may not work as a replacement boardset for a standard pinball machine is that the solenoid switches may be part of the switch matrix.  To make the solenoids fire as quickly as possible, I have assumed that solenoid switches are not part of the switch matrix.  I am going to attempt to update a machine this Fall without rewiring it, and just attaching the cards in the backbox.  Since I haven’t done that yet, I can’t be certain it is going to work without requiring wiring changes.  Both the PROC and Fast controllers are meant to be drop in replacements for many old boards.  Choose whichever one of those best matches your needs.

So how much is this going to cost to build a set of these boards to support a pinball machine.  I will use the SS3 features to calculate complete costs for the boards.  SS3 uses the following rev 1 boards:  two solenoid boards, two input cards, and six incandescent boards.  That means that it will require four PSoc 4200 cards to run the system.  The total cost for the whole system, including connector costs and shipping costs is $95.83.  Actually a more apples to apples comparison would be to drop out the connectors which brings the price down to $66.75.  (I’m guessing when they quote PROC/Fast prices, they don’t include the connectors that you need to actually use the boards.)  Here is a link to the Gen2Board Board Cost that calculates all the costs.  If I switched the insert lighting to WS2812 style serial LEDs, the cost would go down even further.

I am currently hoping that all the embedded code is updated, running, and tested by the end of this year.  It should not take that much work, I just don’t want to over promise and not deliver.

7/24/2015, Thoughts on Pintastic

I took SharpeShooter III to Pintastic as promised.  I’ll start by giving my thoughts on Pintastic, and then my thoughts on how SharpeShooter III performed.

Pintastic was probably the best pinball event that I’ve attended.  My feelings on the event are probably colored by the fact that I had all day Friday to talk to friends and just play a ton of pinball without the day ending in a six hour drive home.

Here’s how the weekend went…  Took off work a little bit early and drove down to Sturbridge.  I escaped Boston before the traffic got too bad which made for a more pleasant drive.  When I got there, Gabe and Derek had  a good number of dollies available to move machines from cars.  I moved the machine in and put it back together fully expecting to have issues.  Twenty minutes later, it was all together, kicked on the power, and the machine was up and running.  (I had every tool I could think of to fix any issues, but it turned to not be necessary.  The biggest issue I had was with one of the light sockets shorting out, which made two or three of the GI bulbs not work.  I tried to fix it with the soldering iron, and found out that the insulation for the socket was failing and didn’t have another socket with me.)

Now the bad.  On Tuesday and Wednesday, I added a couple new features.  I didn’t really have time to test the changes very thoroughly, and hoped that I got them right.  It turns out that I introduced a couple of bugs, that I didn’t find until I got home after the show.  I should have known better than to update the code without testing it that close to the show, but they were new features that I really wanted to have available.  (i.e. having the Aerosmith Sharpshooter song play when the game was in attract mode, and adding a timer to the kickout hole so that only a single callout would happen.  The kickout hole change ended up introducing a bug where the kickout hole would not fire if certain modes were active.  I checked on the machine every couple of hours, and there were a number of times that I saw the ball stuck in the kickout hole.  It took me less than 5 minutes to fix it when I got it home, but unfortunately I didn’t have time to look into it until I got home.)

Thursday, after I finished setting up, a random person walked into the room and he got to be the first person who is not my family to play the game.  It worked well, and I watched a couple other players play a couple of games.  Since it was part of the Restoration Throwdown Contest, it was in a room with only one other machine.  It was a strange setup, where I’m guessing that most people didn’t really understand the point of the machines in the room.  I changed the machine to start up automatically when plugged in, tested it, and it worked first time.  I left the show and drove back home on Thursday night.

Friday was the big day.  I woke up early, and drove back to Sturbridge, getting there at about 7 am.  Plugged the machine in, and it came right up.  (Oh, yeah, I never ended up getting around to installing an on/off switch, so yeah, it is simply plug in to turn it on, unplug to turn it off).  Turns out that there isn’t much to do until about 8:30 am, so went to the McDonalds and grabbed some coffee.  Saw LTG (got to thank him for answering so many of my technical questions on pinside), and saw Jersey Jack and talked to him for a bit.

At 8:30 am, the pinball began, and really continued mostly non-stop for the rest of the day.  There were probably about 100 or more machines to play, and so with the number of people there, I never had to wait to play a machine.  Every couple of hours, I would go back and check on SS3, and make sure that it was running properly.

There were a couple of new games that I wanted to try in the dealer area.  First one was Full Throttle.  I got to play three  or four games on it, and it was a blast.  It is really, really fast.  It is really, really fun.  I hope one of the local arcades, or someone who buys new machines in the area grabs one.  Next machine was the Lexy Lightspeed.  I really wanted to hate the machine because I tend to dislike pinball mixed with video games.  It turns out that I think that Gerry did a good job of striking a balance between those two.  It was less video gamey, and more pinbally than I was expecting.  I enjoyed the game a lot, and wished I would have been able to play a couple more games to get a good feel for some of the modes.  I have to say I was very impressed.  The flipper buttons lagged the flippers by a couple of milliseconds which was strange, but I think you would get used to it.  Last game was the Hobbit.  I ended up playing three games on it, and it was so lacking in features, I couldn’t really get a good feel for the game.  The machine had one of the rollover switches disabled so that you couldn’t get into some of the modes.  Seemed almost like I was playing a whitewood.  We played in groups of four, and I never saw Smaug.  Maybe that was one of the things you couldn’t get to because the switch was disabled.

A friend from CT showed up with his daughter and we got to play a ton of games on Friday morning.  As it got towards noon, I decided to grab lunch at the restaurant by the lake.  Glad I did it at that point, because the lunch took over two hours to get our food.  Sloooowwwww.  It was time for me to give the talk on creating SharpeShooter III.

So here are the quick links to the videos of the talk.  The first video is the main portion of the talk, and the second video are the questions and answers.  The sound is not that good, because, well, I didn’t speak directly into the microphone.  Here are the links:

http://www.ustream.tv/recorded/67230530

http://www.ustream.tv/recorded/67232530

Link to the slides:

Click to access SS3Presentation.pdf

Luckily not many people showed up for the talk, and of the people that did show up, I knew most of them.  I don’t know why random people would show up and watch the talk.  I’m guessing that they were tired and needed a place to sit and rest.  The strangest thing was that Roger Sharpe showed up to hear the talk.  I hope I didn’t bore him too much.

Another nice thing about the first year of a show is that it is pretty small.  Just walking around the halls, you would run into people.  There were three or four times that I ran into Roger with my family, and I’d point at him and go, that’s Roger Sharpe, and my kids would go, “ooooohhh” (with awe in their voices).  They only know of Roger Sharpe as the person that the machine is based around.  I wish I would have had a chance to watch Roger play the machine.  I’m guessing he would have been able to hit the shots, and I could have talked him through starting a couple of the modes.

So, finished up the talk, and a bunch of local friends had some questions.  We meandered over to the SharpeShooter III machine, and I took off the glass, lifted the playfield, and showed the OPP hardware.  I also pulled the backglass so people could see what was back there.  At one point, I was pulled up the pinball framework, and showed you could run the code, simulate switches, and play a “game” on my laptop.  It was a great discussion, because some of the things that we talked about gave me some really good ideas on what is next for the OPP.  More on that in another blog entry.  After an hour of talking in the seminar, and another half an hour answering questions, my voice was spent.  Mark and I adjourned to the lake for some relaxing.  It was a great chance to catch up with him.

The pinball hall was supposed to close at some point, but even after announcing it would close in five minutes, nobody moved, and we continued to play for another couple of hours.  We then moved  over to the Friday night JJP after hours party.  I missed the discussion, but just when I thought I couldn’t play any more pinball, Mark and I found out that they had opened up the hall again.  We ended up playing another two hours of pinball, and that is when I got the chance to play the Hobbit.

Saturday was much less pinball related.  The kids showed up, but both had been sick for the previous week, and didn’t feel like doing much.  They played about three or four games of pinball, and then decided they would rather relax by the lake, or go swimming in the pool, or play a round of mini golf.  The pinball hall was really loud with all the machines, and that turned them off a little bit.  I got a chance to catch the Roger Sharpe talk and ask a question.  I grabbed a couple of games on some machines that I had missed such as Sopranos, and Simpsons Pinball Party.  Teardown was Saturday night and they had a band playing which was really fun to hear as you are working away.  I wish I would have been able to spend more time listening to the band.

So there were a couple of issues with SharpeShooter III which I just didn’t have time to fix before the show.

1. First major issues was the incandescent driver boards.  I calculated the input and output impedence for the shift registers, but ended up messing up the math.  This meant that the clock pulse was reflecting on the long lines, and ended up glitching the lights.  I started with 56 ohm resistors on both the clock and data lines, but they should really be increased to 500 or 600 ohms.  That would make the game look a lot better.
2. I would say that the code was pre-alpha.  There was definitely a bug at the show that would get the machine into a state where it couldn’t be played.  My bad for trying to update the code on the first morning of Pintastic and not having enough time to really test the changes.  I should never have done that, and should have had a code lock for a week before the show instead of trying to add new features.
3. The game is too hard.  To start a mode, all inlanes and drop targets need to be collected.  Drop targets are pretty easy to collect, but the inlanes are difficult.  Since the drop targets don’t get reset until the mode is started, or the ball is lost, it can be very difficult to get the last inlane.  The code should be updated to reset the drop targets in easy mode  as soon as they are collected.
4. Sorry to all the people that played the game and saw some of these errors.  Next year it will be that much better, and hopefully will be complete.  Should only take a short amount of time to fix these issues, but with the time crunch, I was not able to get it done.

I’m tired of typing and need to do some other things.  I promised my wife to work on many things so that I could get a pass and work on the pinball machine for all of June.  That pass has now been returned, so now I have to work on a bunch of items on the to do list.  (Of course, I’m still working on other pinball things when I have time, but probably nothing major will happen for the next few months.  Updates to the blog will probably only happen once a month for the next couple of months.)