"The Black Knight Rises" - Week 3 Progress

It felt like I was beginning to turn the corner in week 3.  With the fabrication and restoration steps complete, and reassembly underway, it seems I might just make my self imposed deadline.



Day 15 - Saturday:
With the playfield clear coat happily curing without me, I turned my attention back to the cabinet.  I measured and cut out the coin door and bottom panels, then glued it all up.

Clamps, a woodworkers best friend.  Plus heavy stuff to weigh down the bottom panel while glue sets.

Once it was assembled, I was able to then measure for the top shelf and the playfield backstop (not sure what else to call it), cut them out and glued them to the cabinet.

Read on for the day by day recap...




Troy stopped by and dropped off the rebuilt drop targets.  They now operated perfectly and looked great doing it.

Old and grungy on the right, new and shiny on the left.  Troy worked some magic on the drop target rebuild.

The improvements are even more noticeable from the back side.  The new springs really made a difference.

Day 16 - Sunday:
Pushing forward with the woodworking, I sanded down the cabinet and got it ready for finishing.

I then focused on the backbox.  Not only was the backbox being constructed from solid walnut, I had decided I would use a nice dovetail joinery to really catch the eye.  Since this was my first real dovetail project, it took most of the day to get it right.  I didn't have enough spare wood to replace even a single piece if I made a mistake.  Luckily by nightfall I had it glued up and perfectly square.


Day 17 - Memorial Day:
This would definitely be a working holiday for me.  I got an early start and cut a cable and ventilation hole between the cabinet and the backbox.s

Next I sanded the backbox.  The sanding took a while since I had left my dovetails "proud" and walnut is very hard, but when I was done all the joints felt nice and smooth.

I then applied a couple coats of Danish Oil to the cabinet and backbox, inside and out.  This would need to cure for 3 days before applying a polyurethane topcoat.  The walnut really responded well from the oil, going from a dull light gray to a much deeper warm brown.

On the playfields I applied a fresh coat of gray paint to the undersides of both.  The original gray paint was marred and dirty, and I wanted a clean backdrop to show off my wiring and electronics.  I was very careful while painting, positioning the playfields vertically to prevent any paint drops from hitting the inserts, and I rolled very slowly to prevent splatter.  I used Behr Premium Plus Ultra (paint & primer) Exterior Flat in "Dark Pewter", PPU18-04.  I lightly sanded before painting, and one coat of paint was enough.

A fresh coat of paint on the underside of the playfields really cleaned things up nicely.

My next tasks was to rub out the playfield clear coat.  I had a judgement call to make on how aggressive a rub out I wanted to perform.  I could simply use rubbing compound, or I could go aggressive with 1500 grit sandpaper, followed by 2000 grit, followed by rubbing compound.

I know, I know... "Show the clear coat already!"  Happy?  I am.

I think the more aggressive approach might have helped a bit with the insert decals not being flat, but after test rolling a pinball on the playfields for a while, I decided the aggressive rub out wasn't worth the time or the risk of cutting through the clear coat.

I then tried some rubbing compound in a couple areas, and honestly I couldn't tell a difference.  After about 10 minutes, I decided the clear coat was best left untouched if I wasn't going to do any sanding.

Skipping the rub out, I then began the meticulous process of playfield reassembly.  Now... where did I leave those parts...

Ahh, there's the parts!  Well, some of them...

A dual playfield configuration like Black Knight is much more challenging to assemble, as certain components and parts need to be installed before the two playfields are married together.  I didn't have any kind of assembly order checklist, we had disassembled too fast for me to make any special notes on what has to be installed in what order.  All I had noticed during disassembly was that one of the upper flippers had a screw obscured by the lower playfield, so it would have to be installed before the marriage.

As I would come to find out of the next day, I should have paid more attention during disassembly.  Luckily the upper playfield is joined to the lower playfield by just 5 screws.  I got quite adept at separating and rejoining them.

On the top side I got the wood siderails installed.  The lighting in this photo makes the insert decals look worse than they really are.  I didn't bother to paint the ugly yellowed white areas, as they will be hidden, even though VID1900 says do it.

I also added the "Magna-Save WHEN LIT" stickers.  The green isn't a perfect match, but the camera certainly makes it look really bad.  In person, the color match is much better but still off.  I had to apply these stickers after clear coating as they are laminated - I was unable to find non-laminated versions.

The "Magna-Save WHEN LIT" decals aren't a perfect color match, but still much better than before the restoration.

Progress looked much better under the playfield.  I started off with the big components first, namely the playfield mechanical assemblies, plus the 3-ball lock.  I also mounted some ChameLED's on the upper playfield.  Though you can't tell from the back side, I used red ChameLED's for red inserts, green for green and blue for blue.

Under the playfield I got about half of the solenoids, the 3-ball lock, and some ChameLEDs installed.

Day 18 - Tuesday:
Tuesday was dedicated to playfield assembly.  Of course, it was not without challenges.

My first big challenge came when I went to install the ball ejector assemblies.  Remember, I had purchased three brand new units, and I had made sure to get the same "left hand side" position of the mounting brackets, so I thought I was golden.  During installation, I noticed that the mounting holes on the playfield didn't position the cam in the correct place to hit the ball.  Setting old next to new, I finally realized that the assembly was a mirror image of the originals.  Plus, the cam arm was longer, and would hit the ball enough in the fully retracted position.

New (left) is a mirrored arrangement of the original ball ejectors, and didn't line up correctly on the playfield.  Also notice that the cam arm is about 1/4" too long, allowing it to block the ball in the fully retracted position.

What I ended up doing is using two new assemblies to rebuild the old assemblies.  I transferred springs and clips from new to old.  I also took the old assemblies and straightened the bent metal bits, and finished with a nice Dremel polishing.  Almost good as new.

But I was still short ball ejector.  Luckily I found that one of the three locations on the playfield aligned correctly with the new assemblies I had bought, so all I had to do was cut off about 1/4" from the tip of the cam arm, and it worked perfectly.

Okay, I guess Tuesday wasn't completely dedicated to playfield assembly.  I took a few minutes to install the backbox hinges on the backbox. These are Williams style hinges, and they use 3 bolts per hinge to attach to the backbox.  As you can see in the photo below, they are too long for my shrunken cabinet.

Oh yeah, that's right, you haven't see the cabinet yet!  Check it out!

Walnut ply cabinet and dovetailed solid walnut backbox, finished in Danish Oil.  It calls to me, "put me in the living room".

This is an excellent indicator of just how much size I cut out of the cabinet design.  The backbox is about 2 inches thinner, and the cabinet lost about 3 inches in depth.  The height of the cabinet is still standard height, as least on the back side.  The midsection lost about 6 inches, and the front lost almost 2 inches.  In addition to the weight savings from reducing wood content, I used 1/2" maple ply for the cabinet bottom, instead of 3/4".  One day I play to weigh The Black Knight Rises and Modern Firepower to see if this really saved much weight.

Back on the playfield, I cleaned and polished (to a brushed stainless finish) the metal ramps and rails as I reassembled.  I also started installing the posts, using new plastic posts.

I love how the old tarnished metal bits look brand new with a few minutes of polishing.

Under the playfield, I got almost all solenoid assemblies and switches installed.  I cleaned and adjusted each switch as part of the installation.

Solenoid mechanical assemblies and switch installation is mostly complete.

Day 19 - Wednesday:
The last playfield assembly to install is the Alvin G. drop target.  Troy's comment that the white targets looked like Tom Cruise's teeth still haunted me.  So I decided to try printing my own in red ABS on my 3D printer.

As long as I was designing my own drop target, I decided to use a Williams style target head on the Alvin G. style stem.

The red ABS from iC3D produced incredible surface finish.

The design work went really well.  My first test print revealed that I only needed a minor tweak to the design to raise the target head a few millimeters.  I had never used ABS before, primarily because of the smell and toxic fumes, but I knew ABS would be needed to stand up to a pinball.  PLA is just too fragile.  Nylon may have worked too, not sure.  Printing in ABS was a challenge until I finally got it to stick to my bed by reducing the first layer height, and cleaning the bed with 91% isopropyl alchohol.  Once I had the adhesion issues resolved, the prints came out fantastic.

The back side isn't quite as clean where I had to print supports.

  Installing the new targets was easy, just had to loosen a couple screws and unhook a spring.

Comparison of the Williams style drop target head to the Alvin G. style.

A nice surprise is that my freshly printed drop targets actually work better than the original Alvin G. targets.  This seems to be because the thick target heads of the Alvin G (which create a flat bridge for the ball to roll over) had a propensity to get caught on the playfield during the up/down transition.  The thinner Williams style doesn't have this problem.

The thicker head of the Alvin G. drop targets gets caught on the playfield wood when they drop or rise.

With my 3D printer happily churning out a couple more drop targets, my next task was to get the rest lighting installed on the playfield.

During installation of the ChameLEDs, I ran out of red LEDs.  I quickly counted red inserts again, and realized I had 30, not 24, red inserts.  Since I had only assembled 24.  I would have to order more LED's and resistors and bake up a 6 more red ChameLEDs.

Alvin G. drop target assembly installed (top right), and lots of ChameLEDs!  Well, lots, but not enough...

One of the reasons I'm building The Black Knight Rises is to uses it as a testing platform.  Installing the ChameLEDs is the first test of this lighting solution that I developed a couple years ago.  I was very disappointed that one of my main design attributes, that you can leave up to 8 ChameLEDs attached as a single larger PCB and wire them all from one end, didn't really work on the Black Knight layout.  The inserts were so close together, I had to snap the ChameLEDs apart to individual units, and even then they were a touch large, making it hard to fit them all on the playfield.

I had designed the ChameLEDs based upon the Firepower insert spacing, and they would have worked 10x better on Firepower.  The challenges I had with making the ChameLEDs fit on Black Knight have me questioning my approach.  I could easily shrink down the size, and further reduce manufacturing costs, if I abandon my goal of having modular, resizable LED strips that scale anywhere from 8 down to 1 LED's.

On the top side, I got the rest of the posts and metal bits installed.  That allowed me to install the rubbers and the plastics.  For now, I just used the old plastics, as I still have to remove the metal posts from the old plastics and transfer them to the new plastics.

Almost fully assembled, starting to look like a real pinball playfield.  Of course, it is!

Day 20 - Thursday:

I continued on Thursday by installing some bayonet style pinball bulb holders, for the locations that bulbs need to protrude to the top side of the playfield.

With assembly complete, and since I was still missing some ChameLEDs, I decided to do some Magna-Save electromagnet testing. I hooked up my new 38V power supply to one of the magnets, powered it on, and manually actuated the small solenoid that completes the circuit.  The magnet powered right up, and grabbed the ball right out of my hand - all the more impressive since the playfield was upside down at the time.

Using my ammeter, I measured the current running through the magnet coil, and was pleasantly surprised.  It would have a max current spike of about 0.56A, but once running the continuous load wasn't event registering on my meter.  My new Chameleon Power Driver was designed to handle 5A loads, so I thought I could power the magnet directly from my power driver, and eliminate the small solenoid Williams used.  I also noticed some extremely mean Transient Voltage Spikes (TVS) that would produce blue fireworks at the leaf spring contacts every time I released the switch.  On a few occasions, the TVS would continue for several seconds and turn the leaf contacts white hot, and I noticed the plastic leaf switch actuator on the solenoid was starting to melt.  When this happened, I had to cut power to make it stop, as it seemed self sustaining.  It's possible that the flyback diodes needed to be replaced.

I decided the solenoid/switch assembly for powering the magnet was suspect, another reason to power the magnet directly from my power driver.  I hooked the magnet to my Chameleon Power Driver, and sure enough it powered right up.  The first thing I noticed is that I was now seeing max amps of around 2A.  Seems like the leaf switch assembly was restricting current flow, probably due to worn leaf switch contacts.  The magnet still grabbed the ball from my hand, but it seemed about the same strength as before, certainly not 4 times stronger based upon the increased electron flow.

I started simulating game play by triggering the magnet for 9 seconds, then turning it off, letting it rest a bit, then powering it again.  It was all running great... until it wasn't.

First, you might be wondering about the 9 seconds.  Don't know where I got that number, as according to the information I've now read, the original Black Knight only powered the magnet for about 5 seconds at a time.  I was definitely over-driving this magnet.

Second, in my glee I started decreasing the resting time between magnet activations. I would power it for 9 seconds, wait 5 seconds or so, then power for 9 seconds again.  I really couldn't believe how perfectly it was working... until is wasn't.

With a pop and fizzle, some magic smoke came out one of the transistors on the power driver.  I turned everything off to examine what happened.  A few minutes later, during my examination, I checked the magnet and realized it was burning hot.  I freaked out and immediately checked the playfield surface, and was relieved to see it wasn't burned or damaged.  That was close.

So the first lesson learned is that electromagnets can produce a lot of heat, and should only be run for short stents.  The limited on-time for the Magna-Save feature is not just to make the game harder, it is to prevent the magnet from setting the playfield on fire!

The second lesson I learned is that the dreaded TVS had destroyed my power driver board.  If you recall from The Modern Firepower Project, I discovered that TVS's of up to 180V traveled upstream from the solenoids, through my MOSFETs, and fried my control board.  I added TVS diodes to the power driver and that resolved the issue by giving the TVS a pathway to ground.

The TVS diodes I hacked onto the Chameleon Power Driver v3.0 to get Modern Firepower running.

When I set about designing my new power driver, the Chameleon Power Driver v4.0, I was determined to engineer the problem out of existence.  The good news, is that my new design prevented the TVS from leaving the power driver and affecting the controller board.  The bad news is that the power driver sacrificed itself to protect the controller board.

Now that I've had a few hours to think about it, it seems obvious that while my new design provided isolation for the controller board, it did nothing about managing Transient Voltage Spikes.  I should have simply included a TVS diode in the design.

This was all especially frustrating as I lost a day of productivity troubleshooting the same TVS problem I solved 4 years ago.



Day 21 - Friday:
With 1 week to go, I needed to get some last minute parts ordered.  First, I had to get some new TVS diodes.  For Modern Firepower, I had used diodes rated at 1.5KW (1500 watts), and found that one wasn't enough.  I wired up four in parallel, increasing their capacity to 6KW, and they have worked perfectly for years.

Online I found that common sizes were 1.5KW, 5KW, and 15KW.  I almost went for the 5KW version, but seeing that it was a little lower than the 6KW I was successfully using, I sprung for the 15KW version.  Upgrading to the more expensive 15KW diode also seemed appropriate considering the incredible TVS light show I saw the magnet produce on the leaf spring contacts, this thing is a beast.

I also finished designing two acrylic panels for the backbox, one to cover the monitor, and the other to cover the speaker and DMD, and ordered them from Ponoko.  Even though I had already designed the panels, I need to tweak them slightly based upon the final dimensions of the backbox, plus the size restrictions that Ponoko has on their acrylic panels. 

Frustratingly, Ponoko kept rejecting the design files I produced from two different CAD packages.  I had to switch to Adobe Illustrator to get a workable solution.  All of this took longer than it should have, so productivity was pretty low.  I was able to get a few more things assembled on the playfield, and started making plans for where to install my electronics on the playfield.


Week 3 Complete - 1 Week Remaining:
Now more than ever I'm doubting my ability to complete The Black Knight Rises in time for the Southern Fried Gameroom Expo.  On my project plan, at this point I was hoping to be fully wired up and testing code.  I haven't even started the wiring yet.

The cabinet is on track.  All that is left is to give it a polyurethane protective coat, fabricate a DMD/Speaker mount, and install the hardware.

The playfield looks great, and is mostly assembled, but still needs to be wired up.  Plus, I need to assemble a working Chameleon IO Controller board, 6 more red ChameLEDs, and a new Chameleon Power Driver to replace the one I just toasted.  My goal is to get most of this accomplished over the weekend, leaving Monday through Thursday for coding.

But if anything slips at all, and I'm sure it will, I won't make the deadline.