Weekly Update – July 15, 2017

In this update, I’ll be covering two primary things: Convention Appearances and Future Devices. At the moment, I’m simply waiting on updates for the V3 orders, so there’s not much to go on with that.

Convention Appearances

There are three scheduled and confirmed conventions that we’ll be appearing at:

  • BronyCon, on August 11-13, in Baltimore Maryland. This will be the final appearance of Matchfire Electronics. Future appearances will be as Phoenixborn Technologies.
  • Nightmare Nights, on October 27-29, in Dallas Texas.
  • Retro Game Con, on November 18-19, in Syracuse NY.

At BronyCon, we will only have the DigiBadges. There just isn’t enough time or money to get anything else done. I’m hoping that we can get a case design by Nightmare Nights, but that’s going to be a lot of pressure on Alabaster, so I guarantee nothing. Hopefully we’ll have them ready by Retro Game Con, but again, no guarantees. It’s not that long after NMN. Beyond cases goes into the next category:

Future Devices

There are currently two devices slated for certain investigation, and a handful of others that are categorized under “Hopeful Thinking.”

For the first two, I have Product One and Product Two.

Product One is a “DigiBadge Light” – Those of you that were around for the Kickstarter for the V1 should remember this. The difference in the V1 Light and V1 Hacker, or V1 Standard, was primarily that the Light used the ATMega168 instead of the 328. Additionally, the 328 on the Hacker was socketed rather than soldered directly to the board, and the screen was similarly detachable.

For the new DigiBadge Light, there’s a slightly different goal in mind. The V3 DigiBadge is significantly more expensive than the V2, with a final price of $50 vs the V2’s $15. This is due to a number of things, but primarily it is due to components. I personally assembled the V2s, which meant that I saved money in two ways. First, I could use components from non-standard sources, without datasheets and sent to me in ziploc baggies. Second, I didn’t have to pay an assembly cost.

However, there was a significant issue: Of the 150 boards and components that I assembled, only 85 were in functioning order at BronyCon. This was a significant problem. I managed to fix a decent number after the convention, which I then sold at Nightmare Nights, but that should never have happened in the first place. The root of the issue is split between time and equipment. As it stands, time would have been extremely tight again, and getting the proper equipment is incredibly expensive and also takes up space I don’t really have.

Assembly costs are completely necessary and will end up saving me a ton of time and a decent amount of money, too. Fewer bad units means more potential sales. However, assembly requires a different set of components. While I could potentially use the dirt cheap components like I used for the V2, I would have to purchase them myself, wait for them to arrive to me, and then send them off to the assembly warehouse. That would waste a lot of time and the money spent shipping them would probably make it not worth doing. For components on the V3, I had the assembly warehouse source them. This meant some things which cost $0.10 each on the V2 cost almost $1.00 each on the V3. And there are significantly more components on the V3.

The Light version will be trimming down things on the V3 to make a more affordable version. My current plans have the Light having the same ESP-12 module as the V3, but some other features will be removed. There will be no on-board USB support, and I’m considering making the V2 run off of 3 AAAs and then through a LDO regulator to a lower 2.7v. The Light also won’t have the ATTiny power controller, instead using its on-board ADC to monitor power. Because the ESP8266 has built-in Flash memory, I am also considering removing the SD card slot. There are a lot of things to take into consideration, but hopefully I can get the device to a cost that’s a little bit lower.

Product Two is a direct result of the Light. Without the CP2102, there will need to be a way to program the Light. For that, I intend on making a CP2102 breakout board. Why, when there are so many readily available on places like Amazon?

For one, almost all of them do not break out the RTS pin, which makes programming the ESP8266 a bit more difficult than it needs to be. And those that do often have them in an awkward location. A CP2102 board designed with breaking out the RTS pin in mind would be a lot easier to use, and could be consistently available for purchase with the DigiBadge Light.

Another thing is that none of the CP2102 breakout boards I have seen have a voltage regulator. The CP2102 does have a built-in 3.3v regulator, but it is very small and can’t power much of anything. Putting a LDO on the board should be fairly cheap and easy to do, and gives a significant boost to available power.

The final reason is simple logistics. I want to be able to sell the means to program the DigiBadge Light right alongside the badge itself. Yes, I could purchase and re-sell the programmers, but I would have to mark up such devices in order to make it worth it. At that point, people could get it from where I got it, for cheaper. I don’t like to rip off my customers. Then there’s also the matter of supply. I would have to hope that the supplier decides to keep them in stock, and in decent quantities. Otherwise, I’d be in an interesting situation if they decide to discontinue it.

Creating my own CP2102 breakout allows me to have the features I want on it and also to ensure they are available.

What about “Hopeful Thinking?”

Well, there’s a handful of devices I would like to visit or re-visit. The LED Matrix pendant, for one, would be a great thing to re-try my hand at. I have most of the supplies to build them, but at this point in time I know I could design a better version. I would love to investigate a different battery and make the PCB only slightly larger than the LED Matrix.

I also want to visit further on the idea of a magnetic field viewer. It would function along the same principle as the Elektrosluch, except instead of directing two inputs into an audio output, it would direct an array into a visual output. It would require a lot of fiddling on my part, but it’s theoretically possible.

Another thing I want to work on is a remote control vehicle core. Due to the cost of getting anything with custom wireless certified, I would build it around something that’s already certified, and it might not be practical, but it’s something I want to look into at the very least.

That’s all for now. As always, you can catch me on Discord, on Twitter, or on Facebook.

Weekly Update: June 3, 2017

Hey guys!

I’m late again. Was working on the latest V3 prototype last night and suddenly it was 3 in the morning. Oops. At least I grabbed a shiny picture:

So, what makes this different from the previous versions? Well, pretty much everything. It’s largely the same circuitry, but everything’s been moved around and reorganized to fit better and have fewer traces having to go entirely across the PCB just to go back to where they started. There are a few changes, though.

First and foremost, I’ve decided to try and make a separate “PowerBoard” for the device. This consists of a boost-buck regulator instead of two separate regulators, plus an ATTiny84 to control it. The ATTiny84 has 11 GPIO pins, of which two are being used for I2C communication with the ESP8266, one is being used for monitoring battery voltage, and the last one is being used to control the voltage regulator. This leaves seven GPIO pins free to use for the device the PowerBoard is attached to. The V3 ties one of them to a button, to turn the power back on when that button is pressed. The ATTiny84 runs directly off of the battery power – or USB power if it’s available – so is, essentially, always on and monitoring the inputs. Its power consumption is fairly minimal, and will be minimized as much as possible. It’ll likely be put into a sleep state when the regulator is off, turning it back on when it’s awoken.

The PowerBoard being an I2C device does a few good things for the V3 board. First, as it monitors the battery voltage, the I2C ADC can be removed. That ADC was, roughly about the price of the ATTiny84. The single voltage regulator instead of two is also about even in terms of price. Additionally, since the PowerBoard is an I2C device and we’re not controlling the voltage regulator directly from the ESP8266, we’ve freed up a GPIO pin on the ESP. That has now been tied in to a transistor which controls the backlight, removing the need for the I2C Digital Potentiometer.

In addition to that, there’s the six unused GPIO pins from the ATTiny84 that are broken out, in addition to an ICSP header for programming said ATTiny84. In all, the 18-pin breakout header has pins for all three voltages: +5v USB power, +3.3V Regulated power, and the Battery raw voltage. The regulator can handle anything from 1.8 to 5.5v, making a LiPo battery perfectly suitable (Although I’d recommend adjusting the ATTiny84’s programming to not discharge it too much. And there’s no charging circuit). There’s also two ground pins, the ESP’s three SPI pins, the two I2C lines, the ESP’s ADC line (Which is linked to the buttons, but with the right resistors you can add more. Or you can read them from something else), and all seven of the PowerBoard’s GPIO pins. One of them is linked to the power button, but this can allow for an external wakeup from another device.

The PowerBoard also allows for a great deal of versatility in V3 design. If I wanted to make a LiPo powered V3, it would only require changing the battery and a different PowerBoard with a charging circuit included. If I decided to redesign with a more robust and powerful regulator, it would only require changing the PowerBoard.

I’ve ordered some prototype PCBs for the PowerBoard, which will let me test the voltage regulator circuit independent of the V3, and I’ve inquired into what would be necessary for manufacturing it. It may turn out to be too expensive to do in this fashion, but at the very least I’ll have the circuit tested and not have to worry about that on the V3.

Until next week, you can find me on Discord, on Twitter, and/or Facebook.

Weekly Update: May 27, 2017

So, this one’s a bit late. I apologize. It’s been a long weekend at my primary job and it’s worn me out.

On the plus side, that’s now over, AND I have three days off this week. Once my DigiKey order arrives, I’ll be able to hop into some more testing. And speaking of testing – A major landmark in this past week! I managed to upload code to the ESP-12 on the new board. It looks a bit funky, as I had to horridly cobble on some headers into places where they were never designed to be in order to bypass the CP2102 chip, but it worked.

Look at those horridly-cobbled-on headers!I also soldered on the SD card slot, and that worked mostly flawlessly too. The test program picked up the SD card and files just fine, and while a separate one didn’t I’m not too bothered. The SD card is also much easier to remove, unlike the first prototype where it was incredibly difficult. As you can see, it’ll have to be moved to the side some to allow for proper access to the mount hole, but that’s a minor issue.

However, I ran across two issues when attempting to put the rest of the devices on the badge: One, I forgot to order the 1% resistors for the buttons (Oops…), and two: It now isn’t uploading any more programming. I’m not entirely sure at which point the uploading failed, so I’ll have to go back through step by step and test things as I go to see where things went wrong.

Additionally, I discovered an issue in the way the voltage regulator(s) were set up. While they seemed to be functioning (I only managed to test the USB regulator), they wouldn’t stay on even when the ESP was programmed to keep them on. So I’ve begun looking at ways around this issue. Currently, I’m poking through a design that uses an ATTiny running directly off of the battery or USB power (Yay diodes!). The ATTiny can be run as an I2C slave, meaning no extra pin usage on the ESP-12 – In fact, what would have been the power line would now be a free GPIO line. I’ve done some testing with the ATTiny85V that I have and it looks good so far. The ATTiny85V has just barely the number of wires I would need, though, and for curiosity’s sake I looked into the ATTiny84. The 84A has 11 I/O pins to the 85V’s 5, and was actually slightly cheaper on DigiKey. I sent a request to the PCB Manufacturer to get the price of the 84 to see if they can acquire it at a similar price. Additionally, if I use the 84, I could potentially drop the digital potentiometer and drive the backlights from the 84. The digital potentiometer isn’t extremely expensive, but being able to offset the additional price of the 84 some would be great.

I have a DigiKey order that I’m waiting on that has a few 84s to experiment with along with the 1% resistors. I’ll be designing a new PCB to get prototypes of once I have a few more things figured out and adjusted, as there are a handful of things that didn’t line up on their footprints quite so nicely.

If you want to yell at me for being extremely late with this update, you can find me on Discord, on Twitter, and/or Facebook.

Weekly Update: May 13, 2017

Updates, updates, updates!

Last week I mentioned the project I was working on for the lightsaber interior. I can happily say that they are all assembled, tested, and working and will be shipped out soon. Look, I even have a picture!

I managed to impress myself with this one, as usually there’s at least one device that needs to be fixed or tweaked to make it work. These guys all worked like a charm, first time. Granted, they’re super simple, consisting of an ATTiny85, a 10k ohm resistor, and a MOSFET, but I’ve learned to never underestimate my ability to make some stupid mistake and put something in backwards.

The test PCBs for the V3 won’t be here until Monday, and given the way things are going that means I probably won’t be able to touch them until Wednesday. I’ve been busy at my regular job (I don’t recommend working in a grocery store during the holidays… and there are a lot more of them than you think), so haven’t had a chance to really fiddle with the code much either.

So instead, I want to talk about the next product I want to develop, once I have the funds and the V3 is finished. It’ll take a significant amount of time to work out, but I’ll be able to work off of the V3 as a base to get it working a lot quicker. It’s also something that Michael (Who I made the ATTiny/MOSFET boards for) has been prodding me for a while to make.

What is it? Why, a tricorder! To be fair, it’s not actually a tricorder from Star Trek, but it is along similar lines. Simply put, I’m going to be taking as many sensors as I can get my hands on and jamming them into a single handheld device. An accelerometer, magnetometer (AKA, a compass), GPS unit, temperature and humidity, barometric pressure, a variety of gas detection sensors, distance measurements… While this list includes a number of things I want to include, it doesn’t include everything. And just because something is on this list doesn’t mean it’ll be on the final board. I do intend on having some designated expansion slots on the device as well, as there are some things that are simply too expensive to include in a base device. Take, for example, the LinkSprite JPEG Color Camera – It can output to an RCA display, and can also take JPEG pictures. Incidentally, the ESP8266 can display JPEG pictures, allowing it to display directly the images this camera takes. It’d be a cool addition, but not worthy of adding $50 to the pricetag. On the other hand, if someone wants to add it to their device, that’s  great!

I’ll be back to talking about the V3 next week. You can ask me questions about the V3, the ATTiny/MOSFET, the tricorder, or any number of things on Discord, on Twitter, and/or Facebook! See you next week!

Weekly Update: May 6, 2017

Greetings, everyone, and welcome back!

This week, we have a handful of things to talk about. I did the “last” testing required to get the V3 design to where I could send it off to get another set of prototypes printed. I also worked a little on the code, getting the badge code and Pride Flag code functioning. I still have to re-implement image and slideshow modes, along with rebuilding the menu system, but that can take some time to be done. I also am having issues with GPIO0 on this device, but even with those issues I can still upload code so I’m not terribly worried. Just annoyed. Hopefully those PCBs, along with the others, won’t take terribly long to get here so I can start testing them out.

On Sunday I mentioned a secondary project. This one requires a little bit of explanation, since while it’s a simple device, it’s incredibly specific in its purpose and design. My friend Michael is part of the Golden Gate Knights, a group of Star Wars Lightsaber enthusiasts who train in the proper methods of saber combat and stage choreography. Michael is working on a saber for a friend of his, and wants to introduce a slight flicker to the blade. These blades aren’t your typical saber blades, because as mentioned in their web page, blades with LEDs going their whole length tend to have LEDs die under repeated impact. Instead, they use high-powered LEDs at the hilt – Enough to require a heatsink and a significant power supply.

So, while the requirement was fairly straightforward – Apply PWM to the feed of one of the LEDs – the power required made it outside of almost all “Standard” approaches. To that end, he contacted me, seeing if I had any input. My design was fairly simple – As the LED uses close to 3v of power, I can hijack that to run an ATTiny85. The ATTiny then runs its PWM output to a MOSFET, which controls the LED. I ordered the parts to prototype it, confirmed the design worked, and then sent off for the PCBs. The device is super simple – ATTiny85, a 10k resistor, and a MOSFET, along with programming pins, power pins, and LED pins. This device won’t be on the store soon, as the V3 is taking priority funding, but it’ll be there eventually. I have a bunch of ideas for saber bits and bobs that I’ve tossed back and forth with Michael, so don’t be surprised if something else shows up, too. If you have any ideas, feel free to toss them out. And, if you want something custom built, I’m completely down for that as well. It’s not always cheap – Michael purchased the entire production run – but I’m willing to work with you to see what we can build. Especially if I can sell it later.

That’s all for now. As always, you can hit us up on Discord, on Twitter, and/or  Facebook!

Weekly Update: April 29, 2017

So, this one’s a bit late. It’s also going to be a bit short: Nothing happened. That’s condensing things a bit, but is generally true. I had a tooth out (Stupid wisdom tooth… Maybe the other three will stay in line now!), which ate up my days off. My primary job’s schedule didn’t leave me a lot of time, either, and spring has not played kindly with my head. A bunch of factors led to nothing happening, but oh well.

I have received more parts to test with – Some diodes will let me experiment with the power enable pin and make sure that works as intended. I’ve received pricing and availability on the regulators (High to low) that I sent off, so I’ve picked the one I want and I’ll be integrating that into the design as well. Beyond that, I’ll be chipping away at getting the code back to a functional state, as the new TFT library doesn’t like everything I had in the old one. And since I don’t know which part of the old one it didn’t like, and the code can definitely be improved a lot, I’m just re-writing it all.

End-goal by the end of the week is to have another prototype design ready to be sent off. I’ll also be updating you on a simple side-project I’ve got going for a friend. It’s specific and niche, but I’m sure some of you will love finding ways to bastardize it.

And that’s it for this, ah, last week. See you on Saturday! Or, if you’re impatient, hit us up on Discord, on Twitter, and/or  Facebook!

Weekly Update: April 22, 2017

Oh man, another “late” one!

I’ve done a lot of things this past week. The prototype parts came in on Sunday, and I started fiddling with them on Tuesday and Wednesday. As of now, my prototype consists of a dual-breadboard setup with loads and loads of wires and about 90% of the prototype parts. I’m not using the NodeMCU ESP8266 breakout any more, as I have the ESP12-E module on a straight breakout, connected to all of the other components.

I figured out that the digital inverter I was using for the card detection on the SD card wasn’t quite up to spec – Instead of pushing a high signal with a low input, it went into high impedance mode. Thankfully, there’s an inverter that does exactly what I want to do, made by the same manufacturer and with the same footprint and pinout, so it was an easy replacement.

I also discovered that with the way the digital potentiometer works, I have to hook it up to the cathode(s) of the backlight LEDs, instead of the anodes. This meant a bit of re-wiring for the PCB, but that’s not a terribly big deal. I did make sure the device itself worked, so I don’t necessarily need to have it hooked in to the prototype just yet.

In additional news, I’ve decided to put a simple linear regulator on the board, to drop 5v down to 3.3v, either from a USB or outside source. The device is inexpensive, so it shouldn’t have a significant effect on the price of the device.

Speaking of price of the device, I’ve done some research into packaging! I was originally going to try to be extra cheap and try to get a simple box with a cardboard wrap around it that had the art/design/whatever printed on it. Then that was a bit pricey, so I though using stickers would work. Then that was also a bit pricey, so I looked into full printed boxes, and they’re a little more expensive – But totally worth it.

As far as expenses are going, the V3 badge is a bit pricier than I hoped, but ultimately this helps me a lot. Once the design is worked out, I can simply e-mail the manufacturer and have them produce a bunch. And with a printed box, packaging, shipping, and storing is a lot easier for everyone involved. Overall, it saves me a ton of time – And as they say, time is money. It’ll mean the turnaround time for conventions and the like will be a lot lower, allowing me to go to a lot more – And the more conventions I go to, the more sales I have, which means I can build up money reserves, which means I can build up stock, making me able to go to more conventions… You get the picture.

That wraps up this week. As always, you can join us on Discord, on Twitter, and  Facebook. Until next week!

Weekly Update: April 15, 2017

It’s time for another round of Good News, Bad News!

Let’s jump right into things. The first set of Good News is that soldering the QFN chips was actually pretty easy with my reflow oven. I just made the board look a bit nasty with the flux. That, combined with a bunch of hand-soldered parts, had me connected to the chip fairly quickly.

It went downhill from there, however. There’s an issue with the circuit board design, and the result is that the ESP module can’t be programmed. That is more than a bit unfortunate. I’m still tracking down the exact problem, and the prototype parts for that have been ordered and should be here on Sunday. Hopefully, I’ll have it squared away fairly quickly and will be able to put in the modifications I need to get the board working. I also noticed a few other issues that will need to be addressed with the board, so those will be fixed from there.

In more Good News, MintShard on Discord mentioned a speedy TFT driver for the ESP8266. Testing it out showed some great improvement to the programmed badges – The Adafruit libraries took about 750ms, give or take, to display a badge. After a few tweaks to the eSPI driver, it was displaying those at about 45ms.  That’s something like a 94% improvement in speed. It wasn’t perfect, though, as BMPs didn’t load well at all. They weren’t positioned correctly, and the colors were several types of wrong. And they weren’t any faster, either, taking about 1600ms instead of 1780.

I poked around a bit, and found the author of the driver also had a JPEG Decoder. And boy, is it great. The V2/V1 aren’t capable of running the code to do it, but the ESP8266 on the V3 is, and that’s one of the reasons I chose it. Testing showed it was much, much faster as well. Two JPEG images I tested it with displayed at 509ms and 265ms. As JPEG images are compressed, the speed will vary from image to image, but that’s still about a 70-85% improvement in speed. Additionally, being JPEG compatible makes it a lot easier for end-users to put their own images on the device. I haven’t checked to see if there are restrictions on JPEG image types like there are on the V1/V2 BMP restrictions, but I haven’t seen anything indicating such.

I’ll still be looking in to seeing if I can fix the BMP display with the new driver, but that’s honestly a low priority. Right now, I’m mostly concerned with figuring out the issues with the board and resolving those as quickly as possible.

As always, you can join us on Discord, on Twitter, and  Facebook. Hopefully, next week will have some updates on figuring out the board issues, but I’m not going to guarantee anything.

Weekly Update: April 8, 2017

Greetings!

Some important updates today, although not the ones I had really wanted. There’s a little hiccup that I’ve come across, but although it slowed me down, it won’t be terribly problematic.

First up: I received the first V3 prototype boards! They look fantastic in the new black silkscreen I chose, and I immediately noticed some a problem: The outline for the screen is slightly off where it should be. This makes it not line up properly at all with where it should be. Only by a few millimeters, but at this board’s tolerances, that’s a lot. Additionally, the milling that made room for the backside of the screen’s flex connector is a bit too small. It needs to be a bit taller, and nudged a little bit outwards. Things like these are why you prototype. Additionally, the buttons are a little smaller than I expected. This may need to be addressed by moving them a little further away from the board, but I have to assemble one completely first to see.

And speaking of assembly, the hiccup. Soldering the QFN chip will be a few types of interesting. It’ll be possible, for sure, but it won’t be easy. These connectors are tiny! I’m going to try hand-soldering one chip, but I’m expecting to have to use my reflow oven. Unfortunately, this is a very critical component, so I can’t just skip it and assume it works. I’ll be heading to my lab shortly to start working on these, and odds are I’ll be posting about it on the usual channels as soon as I get it working – Provided I haven’t completely screwed up in the design.

So long as everything works together, I’ll be able to get all of the components working together code-wise, and be able to adjust the PCB for a better fit. If things need to be moved around, I’ll go for  a second prototype. Otherwise, I feel comfortable enough that I should be able to order the first batch if it’s only the screen needing adjustment.

As mentioned, I’ll probably be posting results of testing as they happen on Discord, on Twitter, and possibly on Facebook.

Weekly Update: April 1, 2017

Hello again!

This week was, for the most part, uneventful. Early in the week, I received the ADCs and a benchtop power supply to test/break them with. I have to say, I’m loving the MCP3021 library that I found. It has a built-in function to return the voltage of the ADC, and since I’m using it to monitor the battery, this is perfect. It also means that updating the code to include it involved only six lines of change.

After ensuring the device was working, I took it over to the power bench and found out how high I could take the ADC pin before releasing the Magic Smoke. Turns out, it’s a little over 18v, more than 3x the recommended operating voltage of the device. Plugging the device back in, it surprisingly still “Worked” – Although the ADC pin was fried, it would still generate spurious data occasionally. After soldering another ADC together to the testing board, I put the ADC pin on a constant 3.3v with the input voltage for the ADC disconnected. At that point, I left it on and went to do other things – Played a bit of Mass Effect: Andromeda, along with helping my father set up protection for some seedlings so the birds don’t tear them apart. About 8 hours later, I pulled the device from the power supply, tested it, and it still worked like a charm. So, thankfully, the ADC circuit doesn’t need to be modified to protect it.

In other news, I’ve decided to leave PCBWeb behind for future designs. Due to the amount of work I’ve put in to the V3, I’ll still be using it to finish off this design (And probably several revisions), but I’ll be teaching myself KiCad to replace it. Now, I’m sure you’re going to ask why, so here’s the answers:

First and foremost, PCBWeb lacks a number of features that are incredibly useful in designing PCBs. While it’s easy to learn and use, it lacks advanced things such as being able to cut holes into a circuit board like what is needed for the ribbon connector on the V3 screen (So the screen will sit flat), and its outline is either an all-angle design or a circle. There are no options for rounded corners, of all things! There are numerous other things which I’ve never needed, but I’m sure I’ll find useful.

Beyond that, and the direct impetus to switch, is that they switched from a DigiKey Library to an Arrow Library. Before I go much further, I’ll say that it’s not the library switch that irritated me, it’s how they did it. There was zero warning – I even searched their website, twitter, and Facebook for any hint of it and found nothing. Additionally, this update meant that the 250+ part library that I had hand crafted needed to be updated with Arrow part numbers. Manually. And when I went to pull DigiKey Part Numbers from them, they had all been removed. Now, the Manufacturer’s Part Number was still there, but in many cases DigiKey has multiple products for the same item – Different packaging options, along with things like buying the whole reel vs individually. It’s nice to know which one of those I had selected.

Hopefully, early next week I’ll be getting the first prototype boards of the V3 and will be able to assemble and test them. As always,  feel free to join us on Discord, on Twitter, or on Facebook.