Excellently awful eBay trash (with schematic)

Oh Clive! You made me yell at my monitor! It is NOT some "glitch" coming through the MCU's power supply! It is not "odd." I knew exactly what the issue was as soon as I saw you bust open the unit, when no large capacitors were in evidence. I knew right away that the problem was that potted module hiding a large (100uF?) capacitor. You rightly identified the problem as an inrush current. Where you went wrong was failing to reduce it. In my storied EE career (ha!), I saw this inrush issue innumerable times. Some young EE would put a FET between the source and load, then hit the FET gate hard producing a nasty dV/dt voltage on the FET's drain. If there was some large-valued capacitor on the drain, it meant that a nasty inrush current would result: I(inrush) = C*dV/dt, easily reaching 10-20 amps, or more, if the power supply can supply it. If the supply cannot, as your various USB supplies cannot (due to their internal 5V switchers' current limits), then the output of the supply will collapse to zero. This is exactly what you have here. The faster the edge, the larger the inrush current. What you want is to slow down that dV/dt on the FET drain pin, and IT IS EXTREMELY EASY TO DO with the circuity you have shown. All that is needed is to append a small capacitor across the NFET's G-D terminals. In conjunction with the 1k resistor driving the gate, these two components create an integrator out of the NFET, much like an opamp's integrator that uses an RC network to define the ramp output it produces. This capacitor is usually called a "Miller capacitor." (Long ago, Mr. Miller must have discovered the integrator and got a capacitor named after himself.) An opamp-based integrator uses the opamp as the heart of the circuit. Due to negative feedback, the opamp will do everything it can to hold its negative input to ground, a "virtual ground," the same potential that is applied to the opamp + pin. To this node, you add the junction of the RC series circuit. The other end of the resistor goes to a DC voltage, which establishes a constant current through the resistor; the other end of the capacitor goes to the opamp's output. This forces the resistor's constant current to pass through the capacitor, resulting in the opamp producing a linear voltage ramp. That's how an integrator works. Now think about the NFET in this "ionizer." 15:06 When the NFET is turned on through the 1k resistor, the NFET is acting as an inverting amplifier whose output (the drain voltage) is 180 degrees out of phase with the input, the gate. The gate voltage goes low-to-high, the drain goes high-to-low. The NFET has some gain in its linear region. Not a LOT Of gain, but an appreciable gain, maybe around 50. You can model this as an inverting opamp once the gate reaches the turn-on voltage of the FET (Vth). If you then add a capacitor across the FET's G-D terminals, you now have an integrator. The difference is that the gate will NOT be held at ground voltage, but will be held (briefly) at Vth as the FET starts acting like an integrator when the FET hits its linear mode. The ramp current is (Vdd-Vth), where Vdd is the supply voltage of the MCU, and applied to the 1k resistor. This current passes through the added Miller capacitor. Since i = C*dV/dt, then dV/dt = i/c. Substituting for i, you get dV/dT = (Vdd-Vth)/(RC). Instant integrator! If you put an oscilloscope on the gate during the ramp time, you will see the FET's gate voltage "paused" at near Vth while the FET briefly operates in its linear mode before going into full saturation, after which the gate voltage will drift up to Vdd. Science! EDIT: To finish the exercise, let's say you want to limit the inrush current to 1A, that the capacitor inside the potted module is 100uF, and that the FET's threshold voltage (Vth) is 1V. You now have all the pieces in place to make the calculation: 1A = 100uF*(5V-1V)÷(1kΩ*C). Solving for C, you get C=0.4uF. That's a little too big, so I would change out the 1k/9.1k resistors to 4.7k/100k. Now you get C of around 0.1uF, which is a common junk box value you're likely to have on hand. This should result in a dV/dT on the FET's drain of about (5-1)÷(4.7k*0.1uF) = 8.5V/mS, and a surge current of about 0.85A. Without this change, I'm guessing the board as-is has a dV/dT about 500 times larger, so the inrush current wants to be about 400A! You see why your USB bricks are collapsing. Your external supply (1:53) probably has a large value output filter capacitor on it, which helps mitigate the problem. Although that USB monitor you have in series may also affect the surge current in unknown ways. It's not a perfect ramp, but you'd be surprised at how linear it is at the drain is and more importantly, how well-defined it is by the equation above. By fiddling with the values of R and C, you can slow down the rise time of the FET drain to significantly reduce the inrush current to the ionizer module (or any other capacitance on the drain), which must have a large input cap buried inside its potting. Try it: add a 100nF capacitor across the G-D of the FET and change the two gate resistors to 4.7k/100k. It should fix the inrush problem. Lesson learned: inrush currents can be huge and lead to power supplies buckling due to those surges, but it's very easy to fix. Sorry to ramble, but I think fledgling engineers should learn this stuff.

Q: You have a activated carbon filtet? A: yeah. Q: But your device is glued shut. How do the user changes the filter? A: Don't worry, the air doesn't go trough the filter, so it won't need changing. Q: but if the filter doesn't filter, why have a filter in the device? A: Obviosly so we can advertise, that the product has a filtet... we ain't saying it does anything, but it does have it!

Manufacturing: "Did you run the design through QC first?" Naaa, we'll just wait for Clive to buy one and figure out what is wrong.

Ooh, it’s like modern computer cases! Completely sealed with fans that can only pull air from inter dimensional planes of existence!

looked like at first a dollar store mono bluetooth speaker that fits into a minivan cup holder

Microcontroller: "Here I am, brain the size of a planet, and they ask me to switch a circuit on when a button is pressed. Call that job satisfaction? Cos I don't."

The use of a barrel jack rather than a USB power port input also suggests this had a previous life as a better designed, possibly older, device.

A MOSFET, a Micro , 4 resistors, 1 capacitor, 2 diodes and a PCB just to save on a mechanical switch? Great case for other projects!

It’s like it was designed by a neural network that hasn’t finished learning.

What a treat especiall!

You know you're getting good value for money when Clive says "this is rilly strenj" so many times.

"Then I got a big blackweb one, nothing happening... Then I got an even bigger power supply..." Was expecting to hear "still nothing", followed by "Then I called photonicinduction to borrow his 100A supply, and it went pop"

This falls into the category of "Why does this exist, and what could we have made with the wasted materials?".

Clive gave this thing way more attention than the original designer did

Can't believe you didn't opt for a can opener on the bottom of that case, it's the perfect shape for one!!!

Since the module is a capacitive load you need to make the MOSFET switch slowly. That could be achieved by a capacitor parallel to the 10K resistor that connects the gate to ground. MOSFETs are so fast that they glitch the power supply like hell when they got a capacitive load. Problems I faced for the better part of 25 years as an electronics engineer.

I genuinely have no idea what you are talking about once we get to the circuit board portion of the inflight entertainment but I find your videos so enjoyable that I watch every single one you post. I'm hoping I'll learn by immersion someday haha

Drills into case: "Hope there's not a big lithium battery inside...." Classic Big Clive. Grabs popcorn. 👍

I like it when the bomb disposal expert grabs his drill and says "We're going in"! As screwed up as this thing is, they seemed to have got one thing right - It stands the right way up!

I know I can’t be the only one that subscribed because his voice, inflection and cadence are so soothing to listen to, right? I found myself walking around listening, not even learning.