Garrett Mace- macetech's LED control hardware / software, the element14 interview

Garrett Mace is an electrical engineer and electronics enthusiast located in Pullman, Washington. His startup company, macetech LLC, began as a part-time hobby in 2008 and recently with his co-founder Jason Moungey made the leap to a full-time business.. Most of macetech's products are focused on LED lighting technology for artists and hobbyists. In addition to designing new products for macetech's online store, Garrett provides custom electronic design consulting services through macetech. His designs are used in art installations, major televised events, performing artist tours, and art/music festivals.

He is pictured here wearing his latest product, LED Shades, being funded on Kickstarter.

Sophi: The last time we spoke, macetech was located in California. Now you’re in Washington state?

Garrett: Yes, Pullman, WA, it’s about as far east in Washington as you can get without being in Idaho- about 80 miles south of Spokane. Definitely not on the west coast...it's a college town, a large college Washington State University  is here, and about 5 miles over is another large college, Idaho State.

SK: Is it a good area to have moved a business to?

GM: It’s a small town, a quiet town, with about 30 thousand people in it when the students are here. There are a lot of technology businesses here- Digilent is here--they make education boards for colleges, FPGA stuff. The people here are well educated- if you walk around with a gadget, people will comment “oh is that an Arduino, can I see the firmware”, it surprises me how in tune with the maker community this area is.

SK: Did you have another job when you first started macetech?

GM: I worked for a company called Comfile. They are a company that is based around (Atmel) AVR microcontrollers and they make small to mid-level PLCs (Programmable Logic Controllers) and essentially provide tools to automate small machines.

SK: So a less expensive automation product?

GM: The flagship project was based on an Atmel ATmega128 with a ladder logic interpreter and a basic interpreter that ran side by side. The core module is called a CUBLOC and Comfile was started much like I started my own company. The owner got out of college, had some ideas, made a few little parts, started working part time and eventually went full time...now they have about 40 employees.[01]

SK: How long were you at Comfile?

GM: About 7 years.

SK: How did macetech get started?

GM: While I was working at Comfile, I was trying to keep up with the hobby scene in general.

I was talking with some friends about how RGB LEDs are really cool but unless you set up a complicated matrix strobe system you have to use to use a bunch of IO on a microcontroller. There should just be a simple way to control RGB LEDs.

While talking about this, I found a new chip from Allegro, the A6281, and it's a small chip, a 3mm by 3mm QFN that is a shift register with 3 current controlled PWMs attached to it.

You could use this to make an RGB pixel. I made a circuit board in Eagle, and discovered that the chip would fit between the pins of a square RGB LED.

I eventually ended up with a circuit board that I thought I should make. So I sent off for prototypes from Batch-PCB (edited: Sparkfun started BatchPCB and ownership was transferred to OSH Park last year) soldered them by hand, tested 5 of them and was pretty happy with it.

Maker Faire Bay Area was coming up and I decided that I wanted to make 150 of these and that was my first interaction trying to find an assembler.

I was referred to a company in China called OurPCB. They agreed to work with me on a small assembly run. It was affordable to run some parts and I had some of those parts, called ShiftBrites, at Maker Faire, in a shared booth with other people.

I gave some samples to some people including Hackaday. After I posted an article on my hobby blog about the product, people were saying in the comments that they wanted them.. So I set up an order system and all of the ShiftBrites I had made were bought out.

So then I thought I could possibly make a business out of this. I had always wanted to start my own business but had never seen a way to get there....so the ShiftBrite just organically caused that to happen.

So that was level 1, having sold 150 ShiftBrites, I could roll the profits back into the business.

And at this point my roommate, Jason Moungey, was getting wind of this and offered to help with the business, ship stuff and put money up to buy more. So we decided to go in halvsies on a larger order of these and actually start a business. The next thing was we ordered 1000 ShiftBrites and set up a better store.

(SK: ShiftBrite, from the macetech website: “The ShiftBrite is a high-brightness LED module containing red, green, and blue elements. It uses a simple clocked serial interface to receive a 10-bit brightness value for each color, resulting in over a billion possible colors. Each input is buffered and output on the other side of the module. This allows each ShiftBrite to repeat the signal to the next…”)

SK: Who did the manufacturing?

GM: OurPCB in China, we've since moved on due to quality issues.

SK: Is Jason still with you?

GM: Yes, Jason is located in San Francisco and we split duties. I do tech support, engineering and shipping because I am located where the stock is. Jason is handling business development, accounting, work with government agencies and distributors and doing some marketing, stuff like that. He encouraged me to actually go for it and although we were both working on the business part time for a number of years, we came up with other products like brighter versions of the ShiftBrite and the Chronodot.

The Chronodot came about because I made a breakout board for the DS3231. And I am really happy with that chip, it is so accurate, rock solid.

So this is not a super exciting story like I was homeless and I taught myself to be an engineer....

SK (laughs)

GM: I was very fortunate because while I was working fulltime, I was able to work on macetech in my spare time, then transition into my working at Comfile part time so that I could spend more time on macetech. This was mostly because the owner of Comfile had done the same thing when he started out. And he was just paying it forward. Then I did go fulltime with macetech two years ago.

SK: Are you still involved in Comfile at all?

GM: No. I'm fulltime macetech.

SK: Do you have anyone else working with you besides Jason?

GM: Not at the moment. We do have friends who help out occasionally- which I think is true of a lot of small businesses. We have had a quite a few friends who helped us assemble LED Shades. And who help us out with the booth.

SK: You’ve got a successful Kickstarter campaign going right now for the LED Shades. Tell us about them.

GM: We first built LED Shades in 2012, and they were built because of a conversation on IRC...someone showed these shutter-shade glasses, a plastic frame with LEDs in them, they don't do anything but at night they are better than nothing.

I wondered if you could do an LED matrix on shutter shades, and if you could see through the matrix. It started out just designing in Eagle, but as the circuit evolved, it because clear that yes, you could fit an LED matrix on the shades.

The first revision was a lower resolution, 20 x 6 LEDs, it used strobed matrix and shift registers and was on a 2-layer circuit board with standard rules. As in true Maker Faire tradition, I assembled it the night before Maker Faire Bay Area 2012 and showed it off in our macetech booth with the other standard products.

People went crazy over it, we had so many pictures of people trying them on, people kept asking us are you going to make these?

After MF, we of course wanted to follow up on that, which began a torturous journey, the design had some problems, the board flexed when you put it on your head and it turns out that flexing circuit boards and QFN chips don’t work very well together. You end up with internal faults inside of the chip itself which you can't fix.

SK: How do you even troubleshoot that?

GM: It became obvious when nothing would fix the glasses besides replacing the chip itself! We'd flex the glasses and they wouldn't work, so it was clear that it was a mechanical failure. We eventually moved to a different solution, the AS1130, a very small chip, which can control 132 LEDs per chip. So we have 2 of those on the current LED shades. We increased the resolution, we went to a 4-layer board, we had to use 4 mil PCB production rules to get that working. So we were able to launch the 2nd version of these at Maker Faire Bay Area 2013 and we sold out our stock of 60 shades by the middle of the second day.

SK: What was the price?

GM: We had them priced at $160 in 2013. We had hoped to get the price down, but when you're building things in small quantities, with a lot of hand work, a lot of 3D printing, the cost of everything goes way up. So that's the price point we had to put them at, and now we are selling them at $140.

I still felt disillusioned with the 3D printed parts that we were using for the hinges. And we were not yet in the volume where we could justify injection molded parts, so I came up with this idea to use the PCB board itself for the mechanical parts.

This was designed using Autodesk Inventor, which allows you to simulate and see how the hinges rotate and then I designed the parts for the cutouts in Eagle.

The hinges are really stable, I call them "puzzle hinges" because they fit together like slot and tab. No injection molding, no length and expensive 3D printing, no post processing. Just get the PCB panel, pop them out and put them together.

SK: How are they powered?

GM: Since the RGB LED Shades are very bright, they can require quite a bit of power sometimes. The best solution we've found is to use a rechargeable lithium-ion USB battery pack.  5V Power is supplied either through the USB connector or through soldered wires.

SK: How many have you promised as rewards on Kickstarter?

GM: Right now we have committed to 165 pairs of LED Shades.

SK: How long will that take to manufacture?

GM: We've already sent out 80 of them for manufacture for the early access backers. After that we're going to streamline the design a bit and make a sleeker product which will go out in February to the rest of backers.

SK: You started out with LED control in 2007, and in 2007 there really weren’t many choices of LED control hardware out there for people to make projects with. Now the market is full of LED control systems and some of them from 2007 must by now be obsolete?

GM: I don't think that LED technology is immune to the forces that govern the rest of the engineering world. There will be advancements and a lot of approaches will be obsoleted either through a technology no longer being manufactured or through market economics. For example the new WS2811 chips have really turned the LED market on its head. We will be moving to a chip more like that simply because the A6281 is not relevant for standard brightness.

SK: What are the differences between these two chips?

GM: The A6281 can handle up to 150 mA per channel so that's a lot of brightness. It has 3 channels, which you can use an RGB LED on, or you can parallel a single color to get a lot of brightness.

The A6281 was interesting technology in 2007- there weren't very many other options. The ShiftBrite at that time had a lot of popularity. Now that's tapering off quite a bit because there are so many other competing options like the WS2811, which are cheaper and in some cases easier to use.

The WS2811 cannot control high current LEDs, it is limited to about 20 mA per channel. It is basically a shift register chip with a simplified protocol using one wire for communication. That was the advantage of the ShiftBrite, you can control many LEDs with full brightness control easily using just one microcontroller....and that is always going to be a popular concept, people will just decide which hardware to use as the technology changes.

One other factor is established systems or established experience, we still have people buying ShiftBrights and a lot of them are people who have already put them into lighting installations. They're buying them because they have established expertise and established software and hardware. It makes sense to use something you know, because sometimes it costs more to move onto something new. We'll offer both technologies as long as we can.

SK: Do you think that the applications for custom LED lighting are going to go away? There are a lot of bright flashy LED things that you can simply buy on eBay. Even the lighting controls with music shown on some of the holiday decorative houses- Wizards of Winter- is now available commercially. So do you see the LED lighting industry going mainstream?

GM: Sure, something like that could happen- you could even apply the same argument to many business sectors such as 3D printing and microcontrollers with the Arduino.

SK: In context of LED lighting, it's art or design because it's unusual, but when the output of the people becomes saturated, what happens to that industry.

GM: People will put a lot of effort into something if they know it's unique, but there are a lot of people who will do something just because they think it's cool. So when it becomes saturated I think the enthusiasts are still going to be there, but you will just see more of the mass market type things in comparison. So if Walmart came out with a fully controllable Christmas light show package, I don't think that would kill the DIY Christmas forums at all.

SK: So that thing, that fully controllable light show package exists. It's about $50.

GM: So I think that this will feed back into the enthusiast community and make what they do easier, it will give a jumping point to do bigger and better things. Because enthusiasts are always going to want to do something bigger and better and I don't think they're going to run out of headroom there. Whether it be LED pixels or making your house annoying to your neighbors around Halloween or Christmas. If anything, there will be more enthusiasts, but they might be drowned out a bit in the commonality of the mass market version.

SK: Do you have any advice for new startups?

GM: I would recommend that early in the startup phase, it's good to develop relationships, so there's definitely value in being near or in a large city. If you're out of that phase,  your business is in the teenage years, you have some legs and you're established some, that's might be when finances start to get tight- so moving somewhere less costly makes a lot of sense. And doing a lot of your business over the internet can be a really good move.

SK: Do you see macetech for the time being remaining in Pullman?

GM: Yes, Business needs to stay here- the last couple of moves really took a lot out of us. It's a climb back to where we were before the move. There is some struggle with the business.