The Modern FIrepower Pinball Project - Shocking Discoveries

Though overall progress has been slower these past few weeks, wiring is coming along nicely.  48 lamps have been wired up, and I have about 30 more lamps to wire.

Lighting wiring:  Red wire is shared 5v power, green wires are ground control lines that run back to the LED-Wiz.

In the pictures, the red wires are providing shared 5v power to the bulbs, and I used all green wires for the individual grounds that run back to the LED-Wiz.  I have no idea which lamp connects to which port, and I don't need to know since my software will map the lights.

Yes, pinball bulbs are typically powered by 6.3 volts, but I already had a 5 volt 2 amp power source handy, and I found that the brightness was just fine using only 5 volts.  I will cover the power supplies in a later post.

For now, read on to discover the shocking oversight I made in my electrical design...

Lower playfield wiring for lights go to LED-Wiz Dev 2.

I've also connected the Pinball Solenoid Power Driver to LED-Wiz board number one, though I've yet to run wires to the individual solenoids from the Power Driver.  The black wires are the individual switched ground connections that control the 16 solenoid outputs.  The green wire is the shared ground.  The red wire provides the 50V source power to the LED-Wiz bank select, so both banks 1 and 2 are being set to the 50V power supply voltage, which is the same voltage that the LED-Wiz will see coming in on the 16 solenoid connections.

LED-Wiz Dev 1 connected to my Pinball Solenoid Power Driver board.

Looking at the empty wiring spools piling up, I think there's about 150 feet of wire on the playfield now. Plenty more to go.

Troy recently paid a visit to review the progress, and to make things more interesting I decided to take a break from wiring lights and test out the Pinball Solenoid Power Driver circuit board.  We spent a few minutes going over the wiring, making sure everything was connected correctly.  For the first test we hooked up a single flipper.

Another view of the wiring connecting LED-Wiz Dev 1 to the Pinball Solenoid Power Driver Board.

With safety checks complete, we powered up.  I was relieved when nothing blew as the 50v life force coursed through the copper veins.  Since I didn't have flipper buttons booked up to play with, I handed Troy the keyboard and told him which key would trigger the flipper.  Troy happily tapped away, but my confusion set in when nothing happened.  It seemed like the software wasn't actually sending the command to the LED-Wiz, but at the time I couldn't fathom why.

Beginning trouble shooting, I tested the voltages.  I had Troy holding one probe on a ground connection while I sampled various points on the LED-Wiz and Pinball Solenoid Power Driver boards.  All readings were perfect.  

Still confused, I decided to manually ground the circuit for solenoid 1, doing the LED-Wiz's job for it.  I was so absorbed in the task at hand, I neglected to warn Troy what was about to happen, and Troy was focused on keeping the ground probe in position. 

BAM!

Like lightning, the flipper jumped to life.  I jumped, even though I knew it was coming. Poor Troy was twice as startled, and for a second I wondered if he had received a shock; he had, just not the electrical kind.

After apologizing I convinced Troy to reconnect the voltmeter probe, and for fun we triggered the flipper over and over again.

BAM! ... BAM! ... BAM! ... BAM! ... BAM!

The noise was victorious, proving that the power driver design was solid, and we continued firing, faster now.

BAM! BAM! BAM! WHOOOOOOOOOSH!

I couldn't believe what I was seeing, as the LED-Wiz board exploded in flames.  Hot blue flames and white hot sparks erupted from two chips on the circuit board like fireworks, finally burning itself out after a very long 5 seconds.  The PC winked out at the same time.

In the briefest of moments I had killed an LED-Wiz and the computer.  

Troy was quite alarmed, and was still standing directly above the smoldering remains of the once great LED-Wiz. As fast as I could I shut down power, but for the electronics it was too little too late.

I was more confused than I had been in years, with an extra dose of angry on top.

Troy worked skillfully to calm me down, and eventually I turned my attention to the PC.  I pressed the power button and the boot screen brought a sense of relief.  Later testing would reveal that only one USB 3.0 port had been damaged, knocking out USB 2.0 functionality, but surprisingly 3.0 devices still worked on that port. The PC was still good to go.  Simply amazing, since the PC had just taken a direct hit from a voltage spike.

The two lower bank chips celebrated the 4th of July a bit early with a fireworks show.  The center control chip also burned.

I found myself explaining to Troy what I thought had happened.  A reverse breakdown voltage spike, caused when power is cut to the solenoid, had propagated to the LED-Wiz, but at first I couldn't understand how.  I knew this voltage spike was a normal occurrence for solenoids, and I had diodes installed all over the power driver board to send the voltage safely back onto the 50v line.

Then, in a moment of clarity, I realized my oversight.  The voltage spike didn't simply disappear onto that 50v line, rather it added to the overall system voltage.  The same voltage that was being fed to the LED-Wiz, a device which has a 50v limit.

I had channeled the voltage spike directly to the one device that couldn't handle it.

Now that I had a theory of the problem, I needed to verify my analysis.  I knew I didn't have any tools capable of measuring the voltage spike, so I turned to my friend Dale, a jack of all trades who used to moonlight as an EE, and he gave me lots of solid advice and also steered me towards Tenma multimeters as an affordable solution.  

I picked up Tenma's 72-8400 handheld oscilloscope multimeter and soon captured the elusive voltage spike in action.  Though very short in duration, I observed spikes in excess of 180 volts!  No surprise then, the LED-Wiz had died of a massive voltage overdose.

The dreaded reverse breakdown voltage spike is huge at almost 200 volts, though brief at 0.3 milliseconds long.

The next challenge was finding a way to suppress the surge before it could cause damage.  My searching led me to Transient Voltage Suppression (TVS) diodes, also known as Silicon Avalanche Diodes (SAD).  TVS diodes are extremely fast devices with picoseconds response time that, once the breakdown voltage is achieved, allows the voltage an unimpeded path to ground.

The TVS diode (bottom) is surprisingly large.  A normal pinball solenoid diode (top) is provided for size comparison.

The most important parameter of a TVS diode is the standoff voltage, below which the diode blocks the flow of electrons - this value should be higher than the system voltage.  I picked up the very popular 1.5KE series of diodes in various voltage ratings, the highest of which was a 62v jobber with a 53v standoff voltage, high enough to allow electricity to flow normally through the system at idle... At least in theory.

Installing the TVS diode was easy, one lead to the 50v line and the other to ground. Turning on the power, I noticed that the diode instantly got hot.

The Transient Voltage Suppresor (TVS) diode is connected on the left side of the Pinball Solenoid Power Driver board, across the 50v power and ground lines.  Note, the power supply wiring is not connected in this photo.

My watt meter revealed that on start-up, 23 extra watts were being consumed by the diode, which gradually decreased to 6 extra watts after a few seconds, and the oscilloscope showed very light clipping of the 50v power signal.  Apparently the 53v standoff rating was overly optimistic, as the diode was clamping the normal system voltage.  I proceeded to test the solenoid firing and was pleasantly surprised to find that the voltage spike was now kept in check, never peaking above 60v, and even then only for a few microseconds.

The voltage spike is clamped by the TVS diode just above the normal 50v.  Some ringing is visible on later cycles.

Troy returned to participate in the testing, and we hooked up a brand new LED-Wiz.  Interestingly, we noticed that the replacement LED-Wiz was a newer hardware revision that is not mentioned on the company's website.  I don't know if the specifications have changed any due to the new bank chips, though for now I will assume there are no functional changes.

The new LED-Wiz hardware revision 4.3 now uses surface mounted chips and resistors.  It is unknown if the specs changed.

Loading up my pinball engine software, again we found that the LED-Wiz didn't seem to be responding.  Troubleshooting the software, I finally realized I had hard coded the flipper and the other cabinet buttons to be gamepad buttons, which I didn't have hooked up, and my attempt to use a keyboard was a non-starter.  I don't quite remember why I coded it that way, but my original concept was that the cabinet buttons would be on a 2nd U-HID device, and they would always be configured as gamepad buttons.  I guess it never occurred to me that bench testing might require the use of a keyboard before all the buttons are hooked up.

Hooking up some real buttons, we tried again.

BAM! BAM! BAM! BAM! BAM! FLASH!

At first I thought Troy had taken a picture; he was bent over underneath the playfield in the area where the flash had originated, but since the flipper was no longer responding I knew something had popped.  Turns out that the TVS diode had failed, not surprising since it was overheating even on the stock system voltage.  Fortunately, the failure mode of TVS diodes is to short to ground, and the flash was the power supply fuse doing its job.

The LED-Wiz survived all our testing, but I obviously needed a higher rated diode.  The next available size up was 68v, which has a 57v standoff voltage.  The 5v jump was larger than I wanted, but other options weren't available.

Fast forward another week, and the testing commenced again.  This time seven solenoids on the lower half of the playfield were hooked up, and the new 68v diode was installed.  For close to an hour Troy and I had fun playing pinball and checking out the features.  Eventually, the 68v TVS diode gave up the ghost, and the power supply fuse popped in response.

The replacement LED-Wiz has so far performed like a champ, and the TVS diodes have prevented the deadly voltage spikes from wreaking havoc with it.

I will continue to test with the TVS diodes, hopefully finding a solution that lasts years instead of hours.  For my next test I will run four identically rated 68v TVS diodes in parallel; apparently this increases their power rating, and I'm hoping by distributing the workload they will last much longer.  If that doesn't work I will try the next higher rated diode, a 75v TVS.  I'm getting uncomfortable with the higher ratings of these diodes, as I feel I am overworking the LED-Wiz board, but with time I am sure I will find the appropriate solution.