How to Build a Disinfection Chamber

When facing a virus without a cure, you should leave no stone unturned in your efforts to fight it. In early February, I learned how ultraviolet (UV-C) light was being used as a tool for disinfection in Wuhan, China. This sounded like a tactic worth exploring, especially as my company (GoodRobot.com) does a lot of electronics and custom prototyping. It’s taken a few twists and turns, but I’ve finished my working prototype: a UV-C disinfection chamber that can be used for disinfection.

Here’s a short video of the finished unit in operation:

Mask disinfection in action

If you’re an organization located in/near Toronto (hospital, care facility, first responder, etc) that has limited PPE and could use one of these disinfection chambers, please get in touch (alan [at] goodrobot.com), I’d be happy to get one to you, create more, and test with radiometer or a calibrated UVC dosage meter if you have one available. Your feedback would be extremely helpful so I can continue to add improvements, and build more of these units.

For those of you farther away, let me share how I built this prototype (by using a modified toaster oven), and perhaps you can make one too.

So first, lets start with the list of materials used to make it:

Second, this project involves serious risks that are not to be taken lightly: This isn’t just boiler plate, so please don’t dismiss these risks — if you don’t take care, you could permanently damage your eyes and/or electrocute yourself. UVC light in particular can blind you or cause skin cancer, and must always be contained in a suitable enclosure or restricted area. So if you’re new at DIY, lack protective/measurement equipment, and don’t have access to a qualified electrician and/or electrical inspector, then this project likely isn’t for you. I wouldn’t normally publish the details of a project like this (that involves genuine risk at this level), but doing so also lets competent makers have the opportunity to save lives. So trying anything based on what you see/read here is entirely at your own risk, it may injure or kill you, or others. Enough said.

Before we dive into the details of the unit, here’s a quick review of the system and how it works…. The reason I chose to modify a toaster oven is because it gave me a head start on many items it’d be laborious or costly to make from scratch. The toaster oven: offers a nice non-combustible surface to mount bulbs, a built-in timer for controlling the unit (bottom knob), a built in door, a pre-wired power plug, plus a hollow interior space (on the right side behind the knobs) to contain the ballast that drives all four bulbs. The power comes from the toaster’s existing wall plug, and one wire is run through the timer knob (it cuts the main power when the timer runs out) and then out to the ballast. The ballast has a pair of wires running out to each bulb/socket (passing through the Cable Glands so the wires can’t be cut or damaged by the sheet metal). So that’s the overall system in a nutshell, actually pretty simple as far as electronics goes.

So from a design perspective, how do we know how much light is needed, and how much the system produces? Well, a good start is this document (one of many excellent decontamination documents available here), which recommends ≥1 J/cm² of UV-C exposure for masks. If we look up the data sheet on our 18 watt Philips bulb (below) it tells us that the bulb produces 51 µW/cm² at a distance of one meter. So, if I’m calculating this correctly… I know that a sphere will capture all light emitted by a bulb (regardless of distance), so I can measure the relative intensity at different distances by looking at the sphere’s surface area and using the ratio of spheres of various sizes. So using that rule of thumb and calculating for a distance between mask and bulb of .1 meter (plus accounting for two bulbs, a pair on both the top and bottom), I arrive at .0102 W/cm² at the .1 meter distance which gives the required 1 J/cm² in about 98 seconds [Edit: first UVC radiometer tests suggest 112 seconds]. That’s reasonably consistent with a UV test strip I used in the chamber which showed a “germicidal dose” with about 30 seconds of exposure.

Datasheet from Philips (18 Watt bulb specs are 2nd row from bottom)

We can also make a similar calculation to estimate the safety of the operator. One source estimates a safe maximum of 30 J/m² of exposure in an 8 hour day. The results vary greatly both by distance as well as the filtering properties (or lack of them) of the glass of your toaster oven. So this is an area where you’ll want to take an actual calibrated measurement before you trust the safety of your eyes to it. I highly recommend adding a shade #5 filter as an extra precaution if you want visibility into the oven, or simply use aluminum tape (common for ducts) and block off the window entirely. As an extra safety precaution, during my build and testing I wore a set of welding googles… never hurts to be extra careful. But putting a UV test strip right against the window I found that a “germicidal dose” took about 20 minutes. But that still remains strong enough that I’ll still be adding a filter (probably Shade #5) to the window as an additional measure. You’ll also want to take your oven into a dark room put a light in the interior and determine if there’s any leaks. I found a couple of pinhole areas where light was leading out, something I remedied with the aluminum duct tape on the interior, as well as an extra strip to cover a less-than-perfect seal of the door.

So back to the chamber (e.g. oven) itself (which should remain unplugged from now on if it wasn’t already). One of the hardest parts is disassembling the thing, it has lots of little screws (be sure to have a container to save them in) as well as sharp sheet metal, and interlocking parts. Since I didn’t have anything to go by, I generally removed every screw I could find (including the typical tricky ones underneath the rubber feet of the unit, which in turn are hidden by a removable rubber plug in the center of the foot). Then I pulled most of the unit apart. In future, I’d like to figure out how to properly take apart the door and hinged swivel. But I’m sad to say that in this case I simply pried it off (imagine removing both side panels as you would a pair of “earmuffs” with the door in the middle still attached to them). Here’s a few photos of some of the screws required for removal (don’t forget there’s a few just inside the door as well).

Here’s a look at how you remove the screws hidden in (and beneath) the rubber feet of the unit.
Don’t forget several of the screws on the inside of the door openings (both sides). To the right is the main timer knob I’m not using to control things.
Here’s a few photos (apologize for low quality) of the removed door. The middle and left photos show how the door (the part with my hand on it) connects to both side panels (what I described as a set of ear muffs). So in addition to removing all the screws holding on those side panels, you also need to pull out two metal connecting tabs (which slide into the slots my finger is pointing to (in the rightmost photo ). In summary, removing this part was a real pain, and I should have worn gloves for it too because of the sharp edges.

So once you’ve removed those “ear muffs” which consist of the door and two side panels (please share if you figure out a way to remove the door hinge instead as that likely would’ve been easier). You’ve now exposed most of what you’ll need to begin the next step of installing your bulbs and electronics. There’s really three main remaining disassembly steps at this stage.

Step 1: go ahead and remove the top panel cover. You can put this off till later if you like, but might as well finish all the sheet metal parts first. In addition to the visible screws holding this panel down, you’ll find it’s also attached by a narrow strip of curved metal that runs along the top opening of the door (in the rightmost photo above you can just barely see it in the upper left part of the photo). So just unscrew that metal cover strip first (screws are underneath it in the door opening) and then remove it before removing the top panel. The reason you remove the top panel is so that you can drill the holes that will secure your bulb socket and holder with bolts.

Step 2: you can go ahead and remove the fan and heating elements (all contained in the interior cavity behind the controls on the right side of the toaster oven). The fan is held on with a handful of screws, pretty easy to remove. The four heating elements are the bars that run through toaster interior. I suggest removing them because they give you a bit more room, but the holes they leave can also double as the pilot holes that you can drill a bit bigger and use to put your cable gland and wires for the bulbs through. In the leftmost photo below you can see the entire fan unit that comes with the toaster oven (you don’t need it), so just remove those screws holding it to the wall of the interior and it’ll come right off. The middle photo below is zoomed in on the lower part of the fan unit, but on the left side of that photo you can see one of the heating elements (a rod) poking into the interior cavity and on the right side of the photo you can see where the power cord comes into the cavity from the bottom of the toaster oven. In the photo on the right, you see the panel that’s on the other (right) side of the toaster oven. Again we see those heating rods protruding (and a metal rod welded to them). To remove these rods, you must unscrew a small bolt near each of the rods in the interior area of the toaster oven oven (In the rightmost photo below you can see how each rod is held with a rectangular piece of metal that’s held by that bolt). Once unbolted, and the wires severed from the rods, you can go ahead and pull out those heating elements.

The leftmost and middle photos show the interior cavity that’s within the right hand side of the toaster oven (just behind the control dials on the front of the toaster). These photos are prior to modification, so you can see the attached fan in the middle as well as the intact wiring on the left and where the power cord comes into the cavity on the bottom right. The rightmost photo shows the left side of the toaster with the panel removed — no electronics on this side other than the rods attached to the heating elements that pass through.

Step 3: Now with the fan and heating elements removed, we’ve got a lot more room to work with in the interior cavity that will eventually house the ballast . In the leftmost photo below, you can see the detached fan unit and all the space it opens up (that’s one of the UV-C bulbs I’m holding to give you an idea of scale). In the middle photo below, you can see the wiring going out to all the control dials on the left side. In my retrofit, I chose to use only the bottom timer dial and nothing else in order to keep wiring simple. I didn’t even use the tiny “on” light indicator, deciding that the fewer AC wires I had running around the better (and the UV-C lights are their own indicator if you don’t mask the glass completely). Most of these wires for the unused upper two knobs were unpluggable, and I cut everything else that wasn’t. The lower timer knob interrupts the main power going to it (red wire coming directly from the power cord and then going to the load next). Instead of the normal load of the heating elements, the terminal from the other lead of the timer is where I attached the leads from the power ballast, in this case the connection was soldered, not plugged, so I had to be very careful removing and resoldering the new wire and it’s strain-relieving clamp on the insulator. I recommend you double check that you get the hot/neutral connections of your ballast correct too (hint don’t just go by wire colors look at the spade sizes of your power connectors to determine) and ensure it’s plugged in correctly into the wall too as I can’t speculate on how this ballast is affected if things are reversed. The photo on the lower right shows what I got once all control wires were unhooked and I had a basic connection between the timer knob and the ballast. In that same photo, you can also see the single black cable gland with red wires inside of it. Later, I decided to use two separate cable glands to minimize any wiring running between the top and bottom of the toaster oven interior. In the same photo, you can see the holes where the heating rods used to be (four in total but I only used two), I drilled two of these holes out larger in order to put the cable glands.

Photos of the right interior cavity of the toaster: with the removed fan (left), the original control wiring (middle), and the new ballast with wired in rough (right).

Once things were cleaned up, I bolted in the ballast (drilled holes to mount it). All wires exited to the toaster interior via two black cable glands (you can see them on the upper and lower right sides of the cavity), and each interior lamp had a separate pair of wires running to it (this required me to use a marette on the single yellow wire going into the ballast, allowing me to have a separate return wire for each bulb). Here’s a view of how it all looked when complete (note: the angle of the marette is poor, as are the wiring color choices, but you should note that no red wires are coming into the marette, all red wires run directly from the ballast to the bulb’s socket and all return wires come into the marette and connect to the single yellow wire of the ballast):

The completed wiring, sorry about the angle and wire color choices.

The information on how to wire the bulbs is given by Fulham, they offer a great lookup table on what wiring diagrams you should use based on the make and wattage of bulb(s) you’re using. Here’s the diagram I chose to use for my four 18 watt bulbs (YMMV):

Here’s the wiring diagram I used based on Fulham’s available charts for my four twin bulbs. I recommend doing your own due diligence on this one. You can also see the single yellow wire coming out from the ballast here.

Now lets shift to the interior of the toaster oven. I ended up drilling three holes for each bulb: two to hold/mount each bulb socket, and then another for the metal bulb holder. Spend some time on alignment to make sure you get this part right (I used the sockets themselves to align the holes I drilled). Depending on the size of the holes, you may get a bit of wriggle room for error and adjustment, but not much. Here’s some photos of what that mounting looked like:

Newly installed bulb sockets/bases (left), wiring in back of socket (middle), full bulbs installed including metal holders (right)

You bulb socket may vary, but in this case there are four pins, each with a terminal that extends from the socket. So as you can see in the middle photo above, the red wire coming from my ballast needed to be be run over to the second bulbs pin (nearest camera), while the return wire which is the farthest pin from the photo (should be yellow but it’s black, apologies) is hooked to one terminal but also brought over to the adjacent far pin by the short yellow yellow wire you see here. In the rightmost photo above, you can also see the full bulbs installed with the wiring running out to the black cable glands on the far side. I’m not happy with the wires being in the chamber like this, though this seemed the best option for a quick prototype. A sheet of plexi or other cover cover (even metal) might be useful to protect against curious fingers or anything else that could snag/damage/expose wire. In theory, the sockets themselves could be butted up against the wall and remove this issue entirely, but the light would be quite uneven with less uniform coverage that way. In the rightmost image above you can see the metal bulb holders too. Pushing the bulb into the socket slowly helps snap them into the metal holder, but mine were a very tight fit, so I bent them a little bit more open than stock because I didn’t want to risk breaking my bulbs with too much pressure.

Last, here’s a photo of the finished unit.

It works!

The unit is operated by by the turning the timer knob on the lower right (note: I discovered that it won’t “ding” when complete unless you turn it to a sufficiently high number of minutes then adjust down to the minute or two that you need). Unlike a microwave, it has no safety interlock and so it will keep running if you open the door — DON’T, because this directly exposes you to dangerous levels of UV light. And as I suggested earlier, you’ll want to do a calibrated measurement of how much light is passing through the glass on your unit and see if that’s within safety specs. Assume it’s dangerous till proven otherwise, and a quick fix is just to cover the entire window with aluminum foil tape, or install a Shade #5 or better filter. On my unit I noticed that at the wrong angle there’s also a small visible crack of light on the rightmost side of the door, so I’ve added some material to make a better seal.

I’m happy to report that in addition to masks, I’ve also placed phones and other inert objects in here as well for decontamination. Who knows, maybe I’ll try a few grocery items in here too (or any item with a reasonably flat surface that normally requires washing down).

Again, if you’d like one of these units to test reuse of masks or PPE, please reach out to me at alan [at] goodrobot.com if you’re located in Toronto. Or feel free to contact me with any other advice or feedback. I’m particularly keen to measure the results with a calibrated UV-C meter. Thanks for reading!

Technology can augment human capability.