Lunar Den Lights

So the front of our house has a bunch of convenient outlets, but they are all run off the same 15A GFCI breaker. Knowing that likely wasn’t going to be enough power for our show, we ran quite a bit of our 2020 show using extension cords from a set of 20A outlets we have in the back of the house. This obviously wasn’t very convenient or probably safe. Each extension cord also had a relay box on it, which worked well, but provided multiple potential failure points.

This year I wanted to have dedicated outlets just for the show in front of the house. I also wanted to eliminate all (or most) of the little relay boxes we used to power the show on and off. Since the pixels and controllers all use power even when they are “off”, I actually want to cut the power to them when our light show isn’t running.

Starting in 2021- control boards and pixels are all powered separately. The Raspberry Pi and BBB SBCs (controller computers) will stay on all the time so I can update and test them during the day. The controller boards themselves, differential receivers, matrixes, and all pixel strings are only powered (using this system) while the show is active. This is for safety and energy savings.

So, I installed two dedicated 20A outlets in front of the house, along with a power management and monitoring system for them.

UPDATE (January, 2022):
The system worked flawlessly for our show in 2021. The SSRs were quiet and didn’t even get noticeably warm. While I anticipated a much higher load based on PSU ratings, and spec-ed for it accordingly, neither circuit exceeded 4A during the show’s operation! It is safe to say that this build exceeded my expectations.

The ONLY issue we had was the GFCI outlets I installed seem to be overly-sensitive and we had a lot of nuisance trips. I’m going to try installing a different GFCI brand for testing over the Summer. We had them trip randomly in perfectly dry conditions, and with specific hardware like brand-new PSUs and differential receivers (that do not trip other GFCI outlets or breakers in the house). I have the same model GFCI in another location inside the house and confirmed that specific components would trip it, while not tripping others, so I believe this is a manufacturer issue. I may also switch to either GFCI breakers or another internal option as I often had to reset them in less-than-ideal conditions outside.

UPDATE (November, 2021):
My intent was to keep the controllers powered up all the time and only power the pixels on using this setup, but it doesn’t work. I believe if the controllers are powered up, even if by a separate supply, they attempt to send power out to the pixels via the capes. None of them would boot consistently without having the pixels energized. If they did boot- they often wouldn’t connect to the cape when it was powered up. At this point I honestly don’t understand why separate power is even an option for these controller capes. I’m currently only using Kulp controllers (A K16, K40, and four PocketScrollers), so this may be different for Falcon, Hanson, or Hinx controllers.

So for now- I’m just powering up the whole show when I want to install updates or load new sequences during the day, and then shut it down when I am done. Next year I might pursue trying to switch the DC just to the pixels using MOSFETs, so I can keep the capes powered.

UPDATED BUILD (9/11/2021):
I wasn’t happy with the way my first build turned out. The cover on the first DIN rail box I bought distorted the displays on the power monitors, and I wasn’t comfortable with the whole SSR setup since I know they can get hot, and if they fail- they usually fail spectacularly. I didn’t like them mostly exposed in a plastic box. I also wanted to add supplemental breakers on the load side of the SSRs, and there weren’t any slots left in the old DIN rail box.

So…

First, I found an “8-Way” DIN rail box with an actual clear cover that fits the power monitors and the new breakers. Here is the box and breakers that I bought:

I also found a larger metal junction/utility box to put the SSRs in. I got it from Menards, but they should be available at about any home center.

I mounted the SSRs in the metal box. I also hooked up an outlet to power my Raspeberry Pi, or anything else, since there wasn’t an outlet in that corner of the room anyway. It’s a GFCI since it is near a laundry tub in an unfinished section of the basement. This outlet is not switched by the SSRs. I am using thermostat wire to separately (now) drive the two SSRs from the Raspberry Pi.

I then routed the power up to the new DIN rail box, which contains the power monitors and supplemental breakers. The neutrals and grounds run through to the outlets outside. The hot leads run through the power monitor’s inductive pickups, and then through the breakers before heading out to the outlets. The power monitors also have hot and neutral leads for their own power and voltage/Hz monitoring.

This makes for a much cleaner and safer install. The SSRs are safely enclosed in a metal container in case they fail, and also to provide much better heat conduction. Once I have them loaded, if they do seem to get too hot- I may add vents and a cooling fan. I’m not expecting them to get too hot since I over-spec-ed them at 40A each. If they get really hot- I may swap them out for contactors instead.

The outlets and power monitors are only energized if the SSRs and supplemental breakers are on, so I have a quick indication of the system status. If the power monitors are on- the outdoor outlets (and main show power) are on. If one or both are off, and a power trigger has been sent- something is wrong. An overload should trip the supplemental breakers first, saving the SSRs, otherwise everything is still on 20A breakers in the main box. Of course the outside outlets are GFCI-ed as-well.

The Raspberry Pi and logic-level converter board are in the small case on top of the outlet box. I have it hooked up to the show network via Ethernet now, as I just don’t trust WIFI for this stuff. I also have each SSR controlled separately now as-well, so I’m using two GPIOs (more info about this below). Here is a diagram of the data-side from the Pi to Logic Level Converter to the SSRs…

Safety note:
SSRs are never truely OFF. There is a tiny amount (usually around 3mA) of “leakage current” through the circuit when the SSR is not triggered. While this is not enough to be overly dangerous- you should still never consider an SSR-controlled circuit to be “Off” unless the source power is cut completely.
To use a real-world example- I have a small network switch for our show plugged into one of the outlets. The leakage current is enough to actually power it- it stays on all the time, even when the SSRs are “Off”.

Following is the original (Flawed) build, with additional info:

The primary components are:

Unfortunately as of this writing, the 12-slot DIN box I used is no-longer available on Amazon. Hopefully they will have them back in stock soon. Here is the listing and a link to it, but right now only smaller sizes are available:

I’m also using an old Raspberry Pi 2B with a WiFi dongle running FPP 5.x to control it. It seems like there are only Raspberry Pi 4s for sale most places anymore, but if you check eBay or shop around- you can probably find a 3B+, which is ideal since it doesn’t need any extra cooling or anything special, plus it has WiFi built-in. I’m using a 2B because I had some of them in a junk drawer, and I’d rather re-use it than recycle it. I could have run this setup using a D1 Mini and WLED or one of my relay apps, but I’m trying to make use of my old Raspberry Pis, and I think FPP will be a more reliable choice to manage it.

So, here are the basics…

As you can see in the main picture above, the control box is attached to the wall next to our main breaker box. I installed two new 20A breakers specifically for it, and ran their power out to the control box, along with neutral and ground wires. All are 12Ga.

The Hot wires first run through the Single Phase Energy monitors, which monitor current using an internal inductive loop. I wanted a way to see at a glance how much power our show is using, and more-importantly- see how close we are to actually overloading the circuits. These things are awsome!

From there- the hot leads run to the 40A SSRs (Solid State Relays). I decided to use 40A SSRs since they were only a couple of $$ more than 20A ones, and I figure the larger capacity will generate less heat than pushing a 20A one to its limits. I tend to treat things like these, and buck converters, the same as a PSU- you should never plan to run more than 80% of the rated capacity for any significant length of time. Magnetic contactors (relays) could have been used too- but they are noisy and their coils use more power and can actually generate more heat.

The SSRs are triggered by GPIO pins through the logic level converter. These are necessary because the 3.3V logic output of a RPi GPIO is just barely enough to drive the SSRs, and is said to not be reliable. The logic level converter bumps this up to a more-typical 5v.

The energy monitors are DIN-rail mounted. The SSR heat sinks have clips in the back that conveniently fit on a DIN rail, but they are too large and aren’t the right form-factor for a DIN box like the one I bought. As you can see in the pictures, I ended up cutting out the openings so they would fit through, which ends up exposing 110v terminals outside of the main box cavity, but under the protective sort-of-clear door. I’m not super-happy about that, but it is what it is. Fortunately they are covered, and I added some electrical tape over them (yellow) for additional protection. I’m the only one that will open it anyway, and I don’t plan on leaving this setup if we eventually sell the house.

Here are some pics of the inner workings, discussed above:

So, the RPi runs FPP (Falcon Pie Player) 5.x in Remote mode. I have it configured to toggle the GPIO on and off using FPP’s built-in GPIO controls and a single-channel Pixel Overlay model. The Overlay Model is independent and can be used via scripts to toggle the GPIO, without having to build it into any sequences. I can incorporate this (and other power management like the FM Transmitter and Projector Power) into my show playlists, toggling the power on just before the show begins, and off when it is done. For now, I decided to just use one GPIO to control both SSRs. In the future, I can easily split them or add more if I want to have finer control of which outlets are on.

On the other end, I installed 20A GFCI outlets in a weatherproof double-gang box with an in-use cover. These outlets will run all of the differential receivers and pixels. The primary controllers, including the Master controller, will remain on all the time. They use minimal power and I don’t want to have to build-in additional delays to wait for them to boot up. I actually now keep the FPP Master controller running all the time, even during the off-season, as I plan to use it for other lighting projects, and use it to test new FPP versions that come out over the course of a year.

Safety note:
SSRs are never truely OFF. There is a tiny amount (usually around 3mA) of “leakage current” through the circuit when the SSR is not triggered. While this is not enough to be overly dangerous- you should still never consider an SSR-controlled circuit to be “Off” unless the source power is cut completely.
To use a real-world example- I have a small network switch for our show plugged into one of the outlets. The leakage current is enough to actually power it- it stays on all the time, even when the SSRs are “Off”.

I did decide to make one change to this. I’ll post pictures when I’m done. I decided it would be safer to add supplemental 20A breakers between the outlets and the SSRs. This is so if there is any kind of short-circuit- the SSRs won’t blow up before the primary breakers trip. I’m adding another DIN box and DIN-mount breakers above the current box. Just waiting for the parts…

I’m also considering installing a DPDT bypass toggle switch so the power out can be manually overridden in case there is some kind of an issue with the SSRs or RPi controller.