I've been wanting to make a battery powered and Bluetooth low energy controlled recording sign, so I could ultimately send it to a friend who has a podcast. Since I've been recording lots of videos, it's at the top of mind. I've put these sub-systems into different products and projects in the past so this time the "new" aspect will be coming from the programming, the actual assembly of the box and attaching vinyl in a sane manor.

Here's my first log entry from June 15:

---



The first step of any project is to get the scope of how you’re imagining it will look down on paper. Rough dimensions are good but first figuring out what materials and processes will be required helps set the stage for later down the road.

For this project I figured I would need a way to cut some frosted acrylic to size. Approximately 11.8 x 3.93 inches (30 x 10cm). The idea being that I could mount any type of additional design to the customer facing side. Whether that be a negative cutout or positive cutout.

The “cutout” would be created by a vinyl cutter which could them be transferred to the frosted acrylic.

On the Inside

On the inside, some consideration has to be made for the LEDs. LEDs have an optimum viewing angle but also get dimmer from when you directly look at them to when you look at them from the side. The sweet spot looks like 30° to 30° that most of the LEDs maintain their intensity above 90%.

After doing some high school geometry, of which I have not done in a long time, I was able to determine that the optimal distance from the back of the acrylic is about 1.8 inches with a 4inch high view port. (note that in the scan of my notes, I actually did the cosine instead of tangent calculation)

The calculations lead to requiring at least 12 LEDs for a 30cm x 10cm configuration. That makes for a total of 36 LEDs to control. That may be 3 LED controllers right there.

A quick current feasibility study leads to the following:

1 color == 12 LEDs
1 LED == 15mA current
12 * 15mA = 180mA

So (a super quick and dirty calculation) a 4AH battery may last 22 hours of use.
6AH = 33hr
12AH = 66hr
24AH = 132hr

On the outside

The box itself would be made of MDF, or similar, with a 45° cut on all ends so that the guts of the MDF is hidden from view. The box could them be painted.

I found the design of others signs particularly interesting. One company used a textured paint. It reminds me of a LineX truck be liner consistency. It’s nice because it hides the joints where the 45° angles meet with the wood.

 

More Research (June 16, 2018)

I start the day with doing a little more research around the subject of vinyl and transfer paper. This is a new subject matter to me all together as I haven’t worked much with vinyl design.


I found a great article on the subject which breaks down the different kinds of transfer paper. It explains that there are pluses and minuses to both the clear plastic kind and the paper kind. After reading I think I’d be using a clear plastic or a low-tack paper for placing the vinyl in my design.


I’m also looking into prices for a Vinyl cutter. I didn’t realize they were so versatile. I may want to sit on it though. I don’t want to waste money on something I may or may not use a whole lot..


Reviewing my calculations from yesterday…

After briefly reviewing my calculations I made last night I noticed that I should have usedtan instead ofcos. This calculation yields thatx (distance from the acrylic to the LED surface) should be about 1.732 inches or 4.399 cm.


The above numbers match a bit more with what my gut would tell me with an effective viewing angle of 60°. I may look into other options to see if I can get a better effective viewing angle to make the box less tall.


Speaking of how tall the box is, it’s also highly dependent on the battery configuration. If I use a battery pack with a thickness of 1inch, that pushes the box out… 1 inch. I could also get fancy and put several packs in parallel but that seems like a crap ton more work for not much gain..


Back to the LEDs for a hot minute

Most of the LEDs I’m seeing, except for some non-uniform types, the wave forms drop off from 90% power at about 30°.




In the case above, it’s actually a bit closer to 25° on the left side of this chart.




I think the best way to handle this is to… test! Time to whip up a board I think would work for this configuration, with all the necessary components, and get it fabricated + assembled.

Depending on the cost, I may just get it done somewhere else..

 

Just whip-up a quick and dirty mockup and look at it.

There are just way too many variables to bother calculating anything here.
it's always going to require more LED's than you expect.

 

Good point. That's definitely the direction I'm going to go in. Better to get it right the first time than to spin a few PCBs..

 

Order some acrylic samples from a few different vendors. Acrylite seems to be great about sending out samples. Low cost plus free shipping.



I'm looking at something close if not the WDR58 option they have. Frosted enough so you don't see the guts inside, but not to much where you can't see the LEDs.

In the meantime I've been building up my library with parts I don't have. SnapEDA has been making that super easy though I've re-built some of the footprints they have. Building libraries is tedious but it makes BOM export exponentially easier once I go to fab.



Starting to do the layout for the functions I know will stay the same no matter what the position of the other components will be. Ideally everything will be placed on one side for ease of assembly and lower cost!



Once some of the parts I need for prototyping come in I'll whip up a box and test the LEDs for optimum distance.

 

You do not have to operate your LEDs at 15mA. They will probably produce some visible light at less than 1mA. Given how directional they are it will very likely be visible when looking directly into them. I recommend getting many 18650s to get the large capacity you desire here. You will want a charge/protect/boost (or buck/boost) board to get the required voltage for a microcontroller and to charge them properly, as well as to prevent any damages. You can use a microcontroller like the ESP32 that integrates bluetooth/wifi and supports low power operation. Also, try using a motion sensor to have it only display things when you are around. You do not want to waste power when no one is around.

 

Thanks for the input.

You do not have to operate your LEDs at 15mA. They will probably produce some visible light at less than 1mA.

I will likely set the current control to 15mA on the LED controller I'm using and then dial it back per each LED. The system voltage will be 3.3 so likely I'll need to bump it up a bit, or use a different rail, due to the loss through the LED driver.

You can use a microcontroller like the ESP32 that integrates bluetooth/wifi and supports low power operation.

Right now I'm using a BMD-300 (nRF52832 module) as I'm both intimately familiar with Nordic's hardware and firmware. Wifi was definitely a thought but I think I'll forgo it on this round. Maybe, in the future, I'll make a Wifi version!

Also, try using a motion sensor to have it only display things when you are around. You do not want to waste power when no one is around.

I'm liking the motion sensor idea. I designed some PIR night lights a while back and they're still running strong since October. I may be able to mush that into the hardware somehow. That would definitely help with the power situation.

 

You can code an esp board with the arduino IDE. It is very intuitive and easy coding. You should really consider it. It has many integrated features, including capacitive touch sensing, WiFi/Bluetooth, and many GPIOs. It is economical, and available from many vendors. Here is a tutorial on it.

 

You can code an esp board with the arduino IDE. It is very intuitive and easy coding. You should really consider it. It has many integrated features, including capacitive touch sensing, WiFi/Bluetooth, and many GPIOs. It is economical, and available from many vendors. Here is a tutorial on it.

Nice, I'll have to grab one and play with it at some point!

 

All the parts that I need for a quick and dirty prototype is in. I just cut some fresh vinyl off the vinyl cutter and transferred it to the acrylic samples I have. (See pictures below)



I still need to dead bug some LEDs but hope to have something working within the next day or so. I grabbed some 1/4" MDF from Home Depot but I fear it's a bit too thin to make the box. I'll play with it and see.

 

Box is built up. I just need to finish placing and wire wrapping the LEDS to the driver. Dwadbugging was tedious but much easier than expected for a 2x2mm LED.



Not bad for a first attempt if I say so myself

 

I did a quick and dirty test with an Arduino and an Adafruit Breakout of the TI TLC59711. Looks like 4 LEDs cover a 4x3in area well. Where my eyes did not see the difference how close the LEDs were, my phone definitely could. I moved back the LEDs until the hotspots in the middle were nearly indistinguishable. I think the dev board is driving the LEDs at just about 15mA which is what I was hoping to do.

 

Looking good. Also, another way to cut down on power consumption is to PWM the LEDs. You can easily do this with a mosfet and microcontroller (the LEDs may need more current combined than your microcontroller can supply). You can PWM them at 15mA or so, but you can have a low duty cycle. This offers greatest efficiency and best light output for lower levels of brightness. You can also acheive different effects easily.

 

Im messing around with a more archaic version of an LED driver, compared to TI's, by Toshiba. (TC62D722CFNG) The sucky part is I need to feed it a clock signal for the PWM to work, the nice thing is that I can set a duty cycle individually for each LED (should that be interesting). I think the processor I'm playing around with has a built in PWM peripheral. I can just run it at 50% duty cycle, hopefully, in low power mode when I'm not connected or broadcasting via BLE. All the little nuances to be figured out I'm sure.



Had some spare time to focus on the layout and component selection. Went through a few different power supplies which inevitably led to some yak shaving. I think what I have now is a good first pass. Final checks on the schematic and footprints/routes and it'll be off!

 

Final PCB came out good after the aforementioned yak shaving.



There's TONS of space on this board so no need to put anything on the back. Even the battery will be stuck on the top.

Ordered the PCB through Macrofab. $220ish out the door. It would have been about $157 for parts, the boards and stencil (not to mention my time!) There are too many things to complete on this project. So paying the extra $63 is worth it IMO.



Lots of these prototype vendors seem cool, especially considering they can handle raw .brd files from different CAD packages. Cost wise most of them never make sense but Macrofab fit the bill.

 

The $220 included acquiring the parts and populating the board for you?

 

The $220 included acquiring the parts and populating the board for you?

It's everything. (boards, procurement, assembly, shipping)

The only thing that I don't get out of it is if I used OSH Park I'd get 3 PCBs instead of two. But honestly I typically run out of patience after the second board so the third doesn't get populated anyway!

 

Ahh I can see your images much more clearly now that I'm on a PC and not a phone. Considering how long it takes to order all the parts and put them on a board yourself, that's not too terribly bad, especially if your personal schedule is otherwise busy.

 

Sort of update:

I was busy over the weekend installing emergency lights into our John Deer Gator at the fire department. That took way more time than expected. Looks good now that it's done though!



Not many updates though I was able to get the OTA firmware to compile for the NRF52810 + S112 soft device. The latest SDK does not have much built in support for the 810 as it was laser focused on the NRF52840 (which is a rad chip by the way; yes, I said rad)

Hopefully will be able to recycle some code from different places so I don't have to re-create from scratch. For instance, I popped an accelerometer in there as I'm curious to see if the tap interrupt is sensitive enough to handle me taping the enclosure. I know I could go down the route of capacitive touch but I'll try this first and see where it leads..

 

Spent more time building out the firmware infrastructure. This NRF52810 has a much smaller flash compared to his big brothers so I've been carefully adding features making sure not to add too much bloat. Adding fprintf for instance added over 4096 bytes to the app code for maybe a single line operation.

The firmware is in a good place but some work still needs to be done to prove out the LED driver. Determining when to sleep, when to advertise, etc.

I also started spinning up a react-native environment which is a bit new to me. There's allot going on to compile a react-native program so when you do just one thing wrong it can send you on a wild goose chase. Bluetooth seems to be working there too but I have to start building out the functionality. Luckily all of that is in javascript (for better or for worse) which I've spent some time with.

Also, re: hardware, I did create the enclosure in Sketchup. Sketchup has been my go to for a while for mocking up enclosures. I even designed a CNC machine in there in a previous life.



Eventually I'd want to make the enclosure out of some nice veneer bamboo on a CNC machine rather than making the cuts myself. It may take al little longer but it will be muuuuch more accurate. First I gotta get the CNC machine working again though.