For a university project I started working on RGB LED Shutter glasses, which are based on the idea of a kickstarter by Macetech. However, these glasses miss some features I was looking for and therefore seemed a good idea to improve upon. Also, they are pretty expensive and most components were given by my university so I was very glad my idea was chosen for the project. That said, our project goal was pretty simple: reverse engineer these glasses and make them react to sound.

However, during the project I stumbled on some pretty cool other use cases for these glasses and therefore decided to develop it further in my free time. However, our prototype is very rough. The LED's are on a pcb which is taped on some 3d glasses and a four threaded wire is connecting the LED pcb to the microcontroller. Not quite nice looking...

Since I want to use these glasses for myself (and wearing them at festivals), I want to make them look more professional. So I'm planning to create a 3D design which is bulkier than normal glasses but doesn't look stupidly big. This means all my electronics (except power) must fit in a small case in the glasses itself. For that, the legs of the glasses seem the best idea, especially if only one leg is used. This volume is approx. 4x6x1 cm.

The electrical components for in the glasses' legs are:

  • Wemos D1 mini pro
  • 3 buttons and a pull down resistor for each button
  • Small microphone (max4465)
  • Connector for power (located at the end of the legs of the glasses, so not in the said dimensions)
  • Power delivery circuit (constant 5V, max 5A)

The Problem

Therefore I'm looking for a way to get quite a lot of power to these glasses. The WS2812B I'm using has a max current of 60mA per LED, of which I have 72 which accounts for 4.32A. Normally these leds function on 5V, but they also seem to do fine on a bit less (li-ion voltage of 3.7-4.2V). So in the worst case they can suck 21.6W. Later on, I'm also planning on increasing to more leds while still maintaining the same surface. So for that I will be using the sk6812 mini, which I'm able to squeeze 100 leds on a new verion of the pcb. This will account for 50mA per led, which is at peak 25W.

Since I like to reuse the whole power design, I'm looking for a way to get constant 5V, max 5A. The way which seemed the most ideal to me was using a powerbank which has quick charge. To trigger this I could use a qc trigger board (as suggested here) and then step the voltage down to the desired 5V 5A. However these two boards are pretty large and certainly won't fit in my glasses.

I also found it's pretty simple to trigger qc, but since I don't want to waste power pulling a constant 12V doesn't seem like a good idea.

Last thing was that I saw a usb c qc trigger board on aliexpress, but these boards are very expensive.

The question

So I'm interested, how I can efficiently pull power from a QC power bank and step it down to a constant 5V and a max 5A, while maintaining a small size?


I'm looking for a small circuit which can trigger QC and convert it to a constant 5V and a peak of 5A. How can I achieve this? Also, other options are welcome as well.

Bonus info

Earlier on, I said that I wanted more features from my LED glasses and will tell here more about that.

My glasses are currently able to display approx. 30 cool presets, but I have quite a lot of others in the making. Some of these presets have quite a lot of customizability, like setting a color (or on some a palette), showing different forms or reacting to sound or not. I also can show text and adress a certain letter with a color. And last but not least, it sort of works with ARTNET (recieves data, atm doesn't tell it's an ARTNET device). Also, I like to make an app which can control a certain preset and configure all the different options of it.

  • \$\begingroup\$ By QC I assume you mean Qualcomm QuickCharge? Or do you mean any of the technologies that accomplishes the same thing? Also, 60mA per LED is a fair bit for LEDs not being used for illumination, are you sure you need that for every LED? \$\endgroup\$
    – Hearth
    Mar 12 '19 at 16:06
  • \$\begingroup\$ @Hearth Just QuickCharge indeed. The power is quite high indeed, but I'm not sure how much power is necessary when at a daytime festival. Also, I know someone who is interested in using my glasses (if fully functioning) for a light show at a party (since it can work on ARTNET). \$\endgroup\$ Mar 12 '19 at 16:49
  • \$\begingroup\$ Modern indicator LEDs can be seen perfectly fine in a bright room at less than 10mA, sometimes as low as 1mA each. LEDs meant for illumination need a higher current, though. Also keep in mind that on a 5V supply, you could probably just barely put two red, yellow, or maybe even green LEDs in series, running the same current through them both, because those types have a lower forward voltage than blue, violet, or white ones. \$\endgroup\$
    – Hearth
    Mar 12 '19 at 17:00
  • \$\begingroup\$ @Hearth As specified, I'm not using a normal LED but I'm using a Neopixel. To be exact the WS2812B and later on the SK6812 mini. So, both the voltages for the package and the LED's in the package are in parallel not in series. \$\endgroup\$ Mar 12 '19 at 17:37
  • \$\begingroup\$ Ah, I didn't realize that. I don't know what neopixels are, I'm afraid. \$\endgroup\$
    – Hearth
    Mar 12 '19 at 17:42

The way I solved this in the original macetech RGB Shades:

Full power is too much, cap it in software

It turns out that running 12-15 watts of LEDs on your face is too much. I have customers asking me frequently for reduced brightness levels, even though the stock RGB Shades firmware ships with brightness capped to 1/3rd of the maximum brightness (with 5 adjustable levels reducing the cap to less than 1/10th the possible brightness).

If you're in a darkened room, even 1/10th brightness over 68x3 = 204 LEDs can be quite blinding. Running them at maximum brightness also can cause the panel to get hot to the touch and physically uncomfortable.

By shipping the shades with brightness capped in software, they run fine from a normal pocket-sized USB power bank. Customers still want it dimmer, so there is no need for a high-current supply. Designing for maximum LED output is a distraction in this application, you will find yourself looking for ways to reduce the current, not increase it.

Filter analog supplies to reduce power rail interference

With normal USB cables, we saw pretty significant power rail bounce depending on the LED pattern. This interfered with a later MSGEQ7-based microphone audio sensor. The solution was to use a small inductor and resistor to filter the power rail, and to have custom power-only USB cables made that both had thicker power conductors than normal USB cables, and were thinner and more flexible than normal USB cables since they only had two conductors.

Much of the audio-specific design process is captured here, if you need additional information (and OSHW design files): http://www.macetech.com/blog/node/144

  • \$\begingroup\$ Damn, never thought the one who inspired my project would find this question. So, I knew the LED's can be very bright but I wanted my design to have as few limitations as possible. So I guess I'll stick to the powerbanks I have lying around. I also know you later found out that Muse used your glasses, what is your estimate of the brightness they used? And thanks for pointing out that noise is quite a large problem for the microphone. I did noticed that the 5V rail of the WEMOS was pretty noisy, but I think I might need to take a better look at the 3.3V line as well… \$\endgroup\$ Mar 14 '19 at 16:33
  • \$\begingroup\$ Also because I’m already going a bit off topic, I hope you don’t mind if I ask some more questions. How did you solve the noise from the button (debouncing)? And why did you opted for the MSGEQ7 instead of a normal microphone amplifier like the MAX4465? And last, how are you able to solder the WS2812B pretty quickly? It took me an awful lot of time (first time SMD soldering, so I think it was 10hrs for a single board) and since I have some more pcb’s laying around I may want to solder them for my friends. \$\endgroup\$ Mar 14 '19 at 16:33
  • \$\begingroup\$ Muse tech department taped a small USB power pack to the left side of the shades, sacrificing some runtime for lack of a tether. I don't know if they increased the brightness in software, though it is possible to do so and would be fine for short periods. Based on the relative brightness I think they're still running about 1/2 or 1/3 maximum PWM. The 3.3V line may indeed be a bit quieter. I used the Shades Audio Sensor in an upgraded LED staff with a Wemos, and ran from the 3.3V rail without issues. However, that sensor board already has power supply filtering. \$\endgroup\$ Mar 15 '19 at 3:51
  • \$\begingroup\$ For button debouncing, my solution (old code due for an overhaul) is available here: github.com/macetech/RGBShades/blob/master/buttons.h No fancy tricks to save memory, just two buttons and timestamps. I wanted some frequency selection, so went with MSEQ7 and my own mic amp solution for compactness. The first times I soldered the RGB Shades, I used solder paste, a stencil, and reflow oven. Stencils are cheap now from the same places you get your PCBs, and you can use a toaster oven (carefully) to replace a reflow oven. \$\endgroup\$ Mar 15 '19 at 3:53

Skipping all exciting details, the question boils down to how to make a portable power with 5 V 5 A output (25W).

First variant, QC (v3 and v4) can output 4.6 A at any voltage level. And the 5V is the starting default. Therefore, to get 4.6A you don't need to have any "trigger" at all, provided that the QC source is designed and is capable to deliver this amount of current. Not all banks/adapters can do this, especially the ones that use standard Type-A port for output, and then use off-the-shelf cables (because a normal Type-A connector can't carry 5A). You need an adapter with "captive cable" of proper ampacity.

Second, if a "power bank" has advanced QC and enough total power, you can indeed "trigger" the high-voltage higher power mode (9V, 12V etc.). Once you "trigger" this level , the task is trivial - it is to convert the 12 V into 5 V 5A using a buck switching converter. Actually, the 5 A output level level is not completely trivial, and can't be done in "small size", typical power density is no less than 3-4 cm3 per 1W, so you are looking at about 90 cm3 volume space.

enter image description here

Alternatively you can design your own battery-powered supply (power bank) with desired output. It will take likely a battery of Li-Ion cells, balanced charger, and DC-DC converter to 5V whatever amperage. The first "standard" way seems to be the "more effective".

  • \$\begingroup\$ Thank you very much! If I indeed can get 4.6A that will account for 23W, which is more than enough for my first prototype. Also it's 92% of the max wattage for my planned new PCB with more LED's, so that seems pretty sufficient (so 8% less brightness). But now the question, which types (or brands) of battery packs have this feature? Or is there a standard for this type of Quickcharging? \$\endgroup\$ Mar 12 '19 at 17:49
  • \$\begingroup\$ @JeremyvanWieringen There is a standard, it's called QuickCharge. It's a Qualcomm-specific thing, so it won't (in theory) be as widespread as the open standard USB-PD, but it's reasonably common. It is, however, complicated enough that you should expect compatible battery packs to be much more expensive. \$\endgroup\$
    – Hearth
    Mar 12 '19 at 18:21
  • \$\begingroup\$ @Hearth Ah okay, so I thought maybe besides Qualcomm there were more standards which would allow it \$\endgroup\$ Mar 12 '19 at 18:28
  • \$\begingroup\$ @JeremyvanWieringen There are, but you stated above that you were asking specifically about Qualcomm QuickCharge. \$\endgroup\$
    – Hearth
    Mar 12 '19 at 18:32
  • \$\begingroup\$ @Hearth Oops, my bad... I meant just any technology that could achieve a quick charge. But I do think I may have found an option. It's a Anker Powercore, but I'm not sure if it can output 4.8A on two ports or 4.8A max in general. It does however state it can't do Qualcomm QC 2.0, so a bit weird... \$\endgroup\$ Mar 12 '19 at 18:45

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.