A Look At Chinese Value Engineering

November 21, 2017 by Anool Mahidharia 63 Comments

Seventy cents doesn’t buy you a lot these days. Maybe some sweets or candies at most. How about a string of LEDs that you can use to decorate your home during the festive season? [Amaldev] was curious to know what was, or wasn’t, inside these blinky LED strings which made them so cheap. He’s done a Christmas LED Light Teardown and shows how blinky LED string lights can be built with the bare minimum of components.

The string he purchased had 28 LEDs – seven each in four colors, a controller box with one push button and a power cord. Without even knowing what is inside the controller box, the cost of the product seems astonishing based on this BoM. The single push button cycles through eight different light patterns for each press. It even has a faux CE mark for the supply plug. Cracking open the case, he finds that the controller board is sparsely populated with just seven through hole components and a COB (chip on board) module. A simple, 8-bit, 8-pin microcontroller is possibly what controls the device.

[Amaldev] sketches out a schematic to figure out how it works. There are two arms with 14 LEDs of alternating colors, each of which is controlled by an SCR. Two GPIO output pins from the COB control the gates of each of these SCR’s. The button is connected to a GPIO input, and a second input is connected to the AC supply via a current limiting resistor. Most likely, this is used to determine the zero crossing of the waveform so that the COB can generate the appropriate trigger signals for the gate outputs.

It is unlikely that these products are manufactured using automated processes. The PCB production could be automated, but soldering all the wires, fitting it all in the enclosure and preparing the LED string itself would require manual labor. At US$ 0.7 retail on the street, it is difficult to imagine the cost breakdown even when the quantities are in large numbers. Maybe a combination of cheap components, recycled or rejected parts (mains cord/enclosure), lack of safety and protection measures (no fuses, no strain reliefs) and reducing the component BoM to an absolute, bare minimum, coupled with very high volumes lets them pull it off? What are your thoughts – chime in with comments.

New Part Day: A Fake Sun

November 17, 2017 by Anool Mahidharia 49 Comments

LED technology has improved by leaps and bounds in recent years, with what was once considered unachievable being common place now. Two of the main parameters of interest, total input power and conversion efficiency have been steadily increasing over the years. An efficacy of 120 lumens/watt is fairly common nowadays, and it may not be improbable to expect double this figure in the near future. Input power ratings have also steadily increased, with single LED units capable of 100 W or more becoming common.

But the Chinese manufacturer Yuji seems to have hit the ball out of the park by introducing their BC-Series, 500 W, high CRI, high Power, COB LED. Single, 500 W COB LED’s are not new and have been available since a couple of years, but their emitting surface areas are quite large. For example, a typical eBay search throws up parts such as this one – 500 W, high Power LED, 60,000 lm, 6000-6500K. It has a large, square emitting area of 47.6 x 47.6 mm. By comparison, the Yuji BC-Series are 27 mm square, with an active emitting area only 19 mm in diameter. This small emitting area makes it easier to design efficient reflector and/or lens units for the LED.

Luminous Flux is between 18,000 to 21,000 for a color temperature of 3200 K, and between 20,000 to 24,000 for the 5600 K type. Further, this high power rating is accompanied with a pretty high color rendering index (CRI) above 95. This allows the LED to faithfully reveal the natural colors of objects due to its wide spectrum. Electrically, it is rated for 12 Amps with input voltage between 35 V to 39 V. This translates to between 420 W ~ 468 W of input electrical power. Some quick math tells us that the efficacy ~~efficiency~~ works out to just a little over 50 lm/W, which isn’t all that great. But with light sources, you can have high-efficacy ~~high-efficiency~~ or high CRI, but not both – that’s just how the physics of it works.

At US $ 500 a pop, these eye blinders do not come cheap and may not find much use for individual hackers. But for some applications, such as studio and theatre lighting or photography, they may be just what the Doctor prescribed. In the video after the break, you can see [Matt] from DIY Perks give a rundown of the LED’s features and take it for a test ride.

Continue reading “New Part Day: A Fake Sun” →

Blinking An LED – Extreme Edition

October 13, 2016 by Nava Whiteford 58 Comments

This hacker’s video on blinking LEDs never got the recognition it deserves. At the time of writing clocking in at just 61 views, but it is indeed a work of art. Just trust me, scroll to the bottom of the article and watch it, you wont be disappointed.

Not convinced? OK, let me tell you about it and the world it has opened up in the Japanese maker scene. We’ve all blinked an LED. Maybe it was just to test a microcontroller, like the simplest Arduino example.

Or we’ve been a tad more old school and used the classic 555 to do it. Or maybe like me, you went through a phase of hacking together Phase Shift and other oscillators because well… it’s fun!

But [Junichi AKITA] has more extreme tastes, deciding that a custom IC layout is the way to go. [Junichi] designed a ring oscillator composed of flip-flops, then hand laid out each MOSFET placing each layer exactly where it should be fabricated.

The resulting design was then fabricated by an academic shuttle service in Japan (a bit like the well known MOSIS service). The result is a tiny circuit in the top right corner of the IC. Which of course [Junichi] then had to wirebond (check the video for a cool 1980s style Westbond machine which are still hugely popular in Japan).

IC Layout… by hand!

Optical image of the fabricated circuit.

[Junichi] bonded the die directly to a PCB (COB). I assume, purely for irony, a 555, and ATtiny based oscillator were also laid out on the board.

I guess you might have a couple of lingering questions. First you’ll likely bemoan your lack of your own fabrication facility (I’m still eyeing those used 1 micron fab lines that crop up on eBay from time to time). And secondly you might be asking yourself… why?

Both these questions are somewhat answers by the MakeLSI project. This growing project in Japan seems to have opened up semiconductor fabrication to all kinds of projects.

While my Japanese isn’t good enough to fully understand what’s happening it’s clear there are many awesome projects going on. Including joys such as IC layouts designed in vector graphics packages (Inkscape) and die images packed with interesting layouts, anime characters and QR codes.

For more awesome images and information (unfortunately all in Japanese) you can check them out on Facebook or on their homepage.

Continue reading “Blinking An LED – Extreme Edition” →

A Completely Open Microcontroller

October 10, 2016 by Nava Whiteford 58 Comments

We don’t know about you, but the idea of an Arduino-class microprocessor board which uses completely open silicon is a pretty attractive prospect to us. That’s exactly [onchipUIS]’s stated goal. They’re part of a research group at the Universidad Industrial de Santander and have designed and taped out a RISCV implementation with Cortex M0-like characteristics.

The RISCV project has developed an open ISA (instruction set architecture) for modern 32-bit CPUs. More than 40 research groups and companies have now jumped on the project and are putting implementations together.

[onchipUIS] is one such project. And their twitter timeline shows the rapid progress they’ve been making recently.

mriscv_bonding — Die directly bonded to an OSHPark PCB

After tapeout, they started experimenting with their new wirebonding machine. Wirebonding, particularly manual bonding, on a novel platform is a process fraught with problems. Not only have [onchipUIS] successfully bonded their chip, but they’ve done so using a chip on board process where the die is directly bonded to a PCB. They used OSHPark boards and described the process on Twitter.

The board they’ve built breaks out all the chip’s peripherals, and is a convenient test setup to help them validate the platform. Check it, and some high resolution die images, out below. They’re also sending us a die to image using our electron microscope down at hackerfarm, and we look forward to the results!

Continue reading “A Completely Open Microcontroller” →