The Shot Core Brings a Modular Design to an ESP32-Powered Camera Remote Control

We've all been there. A friend, or perhaps colleague has come to us with a problem statement: They have a challenge or task that could be made easier with automation, some electrical wizardry could work wonders for their workflow, so they've called upon you and your knowledge to step up and suggest something suitable that will make their life easier.

Eventually, you convince them that a full bipedal android isn't going to be the most practical solution, but you've got some ideas that could work — maybe some automation could be the way forward if, of course, you can keep the Yak Shaving to a minimum!

If you can keep on task, and stick to coming up with a design that cleanly satisfies the requirements of the customer, then there's the task of making sure that what you design is suitable for use by actual customers — who won't have the "mothering" touch that we often employ when interacting with our creations. Our customer won't have the same appreciation for our (often) delicate electronics, and to be fair, they shouldn't have to.

Mike Rankin found himself in just such a situation, when he was engaged by a company called Wonder Unit. They were looking for a device to solve some of the issues they faced with the development of their Storyboarder system.

Storyboarder (GitHub) is a wireless camera control system, capable of remotely configuring a camera with the predetermined settings and configuration needed for the some 2,000(!) individual photographs used in the creation of a cinematic storyboard. I don't think I'll ever complain about the amount of preparation for one of my projects again...!

Looking at what Rankin was tasked with, I feel he's very nicely satisfied not only the (exceptionally well defined!) product requirements of his customer, including an evaluation of existing concepts and ideas, but he's also accomplished the project in a clean and well thought out way, managing to produce a design that would have absolutely no issue being released to any production line.

Shot Core is the answer to the Storyboarder problem statement, and found within its design are a few neat hardware design tips and tricks to keep in mind when laying out hardware that is eventually going to be dropped into the waiting hands of the general user.

There are also some notable design features that are well worth pointing out — ones that go the extra mile in hopes that they will potentially mitigate some of the damage that might otherwise happen when the that same user inevitably finds their new toy falling out of those same hands!

Comprised of two seperate function PCB assemblies, this little stack fits snugly together, with signals being passed through 0.1" pitch pin-header / receptacle blocks. It's a solid and proven connection technology, and easy to source is always a bonus if you're looking to subcontract production!

Being so robust, it's not uncommon to see split board designs relying solely upon the mechanical retention of these headers, and for some applications or case designs, that can actually be sufficient.

However, if you want real peace of mind for your hardware, it's wise to add some extra mechanical features — for instance, the rigid metal standoffs that we see sandwiched between the PCB pair.

Not just any standoffs, though. These are not only threaded, saving the addition of another BoM line for a pair of M3 nuts, but they are also designed to be soldered down to SMD pads as part of the PCB assembly process, getting soldered in place with all the other parts during the same reflow process, saving further time — and therefore cost — in final assembly.

Their design also has the very desirable benefit of leaving the opposing side of the board they are soldered to completely untouched, leaving a completely flat surface, perfect for a decal, or in this case, a screen to sit flush against.

These parts really clever as well, in that they greatly simplify the assembly procedure — needing only two M3 threaded bolts to provide a captive lock on the mating of the two circuit boards Throw in some threadlock, and you can also skip the spring washers, as Rankin looks to have done!

Once fastened together, it would take some serious, warranty-voiding shock and abuse to unintentionally get these two boards apart!

Have things your way...

There's a modest requirement for a simple UI, so the user-facing board suitably serves as a front panel interface, featuring all that is needed to lets a user easily interact with the device and its settings.

With a SSD1306-based OLED display sitting flush against the front of the unit, all that is left to complete the user interface are three tactile pushbutton switches and a transistor driven piezoelectric sounder.

Rankin knows what he's doing when is comes to driving these popular little OLED screens. Though widely available in pre-assembled modules, he's chosen to sure up the supply chain by taking the capacitors and DC/DC boost converter that normally reside on these breakouts, and has incorporated them directly into his HMI board.

This has a number of advantages when considering design-for-manufacture. Not only does it remove the variability that can be common place when sourcing pre-built modules, it can even save on costs.

Removing the additional board also allows the design to be smaller overall, something further emphasized by the direct-to-board connection Rankin has chosen for mounting the OLED, forgoing the fragile FPC connector, in favor of soldering the contact fingers of the display FPC directly to an appropriately sized footprint — again, reducing cost by scratching a FPC connector from the BoM.

Getting connected

Having laid out the handful of components that make up the HMI, Rankin can now make headway and head on over to design the cordless controller conveys the commands sent to it, on to the camera it is connected to.

A few years back, such a device this would have demanded quite the design effort. But, ever since Espressif went and released their WiFi and Bluetooth-capable, Arduino and PlatformIO-compatible series of ESP8266, and more recently, it's steroid-jacked sibling known as ESP32 parts to the world, well, things have never been the same.

With unit sales for the combined ESP product range now exceeding 100 million units (that's the 2018 figure by the way...) it's safe to say that it certainly looks like this part has gained notoriety in both the mass production and maker crowds alike!

The ESP32 is the most recent, "mature" part to be offered by espressif, and we've seen before that Rankin is no stranger to these little powerhouses. When he's not using them to crank out complex graphics at fluid frame rates, he's got them stepping through their paces, having them calculate the control and feedback for some seriously smol' stepper motors.

No matter what the project, Rankin has shown that the ESP32 is up to task, and with its dual-core Xtensa processor, clocked at 240MHz taking pole position of a formidable feature set, these parts sure do appear mighty tempting when pitched against project requirements.

There's a ESP32 variant that's especially hard to say no to. One, highly-integrated QFN packaged part is all you need to set up a full IoT/WiFi connected product, and that's the ESP32-PICO-D4.

We've previously reported on this part; a system-in-package that takes a bare ESP32 MCU Die, an additional lump of flash, the required 40MHz crystal oscillator and a slew of decoupling capacitors, and wire-bonds them all up into a single, easy to integrate QFN48 SIP, with a footprint area of just 7mm x 7mm!

At just shy of £3 (Qty 100+), this is likely the most expensive BoM item, but given this single part caters for a significant part of the functionality of the design, it's not going to be hard to optimise that much further!

All that is left to to do now to this board is to provide a debug and configuration interface, over which to initially allow the ESP32 serial bootloader to take receipt of a binary file to be loaded into the FLASH memory. This is provided by a Silicon Labs CP2102 USB-Serial converter IC.

This part features a few GPIO lines that can be toggled via the handshaking of the USB serial port, which enables the automatic reset and initialization of the bootloader without any need for user intervention.

Reducing the risks of radio

The antenna with its AC coupling capacitor are the sole RF components that are required externally to the PICO-D4. With anything RF-based, it pays to stick close to a verified design. Ask anyone who has ever had to deal with the infuriation of impedance matching, or specifying special stackups, with specifications that satisfy the SNR requirements of a signal suitable for a stripline.

Espressif have done a lot of legwork in making their product as designer-friendly as possible. From full hardware design files for the development boards and modules themselves, Espressif have also provided a wealth of reference documentation in the format of application notes, example BoMs and much, much more.

Using the same antenna that espressif themselves favour for their D4-based boards not only gives a pretty good chance that the design will have a satisfactory RF performance, but it also means that the design can closely match a known, FCC-certified layout such as 2AC7Z-ESP32PICOKIT, for example.

This can be a huge benefit if you've got plans to sell your gadget. Emulating a known reference design as closely as possible can hugely simplify the certification process required to place the product on the consumer market. This won't remove the barrier of certification, but it can potentially give you a bit of a leg-up and over when it comes to it.

The only way I can see that Rankin could further simplify this module would be to perhaps substitute out the PICO-D4 in favor of the newly released ESP32-S, a single-core variant of the ESP32 that also offers a built in hardware USB interface.

However, good design is a balance of all the factors, and the fact that the ESP32-S is currently not offered in a SiP variant, quickly renders moot any cost savings to be gained by dropping the CP2102 - as you'd then have to provide the required FLASH memory chip, and take up the extra design effort of the more complex layout - both costs that the PICO-D4 helps avoid!

Frugal FR4 Faceplates

A final nice little trick Rankin has employed further files down costs, supplier listings and likely, lead times, all in one fell swoop! Put that way, it almost sounds too good to be true - it's very reminiscent of the project management triangle we've all seen, with it's usual assertion that you can only ever have two of the holy trinity at any one time.

Often, the front facing decals and overlays we see on many products are formed of die-cut, adhesive-backed sheets of plastics, with lettering and logos laid down using some relatively exotic inks and printing processes, and when it comes to manufacturing, anything "exotic" tends to associate with "expensive". If not for the die-cut tooling alone, even the most detail devoid decal can have unexpected costs associated with production.

Fortunately, there's a supplier already on Rankins list that can produce rigid, screen-printed front fascias, complete with the features and holes required for the components that make up the UX.

Oh, what is more is that they can do it quickly, cheaply, and to a very high standard. Any guesses as to who it is?

For those who haven't figured it out, let me fill you in — FR4 PCBs make for an excellent frontage for an enclosure! For all the reasons listed above, and then some. If you've been keeping tabs on the #PCBArt hashtag, you'll be familiar with just what can be accomplished with the right mix of copper, silkscreen, solder mask, and selective plating.

Keeping things simple, yet effective, the final front plate not only looks fabulous, but it also gets fabricated and shipped over in the same parcel as the electrically functional PCBs that make up the rest of the Shot Core unit. We can even see potential to use the OLED window as a adapter board for the "hot-shoe" connector, leaving it in place with some mouse-bite tabs, ready to be popped out and populated at the time of production!

Some closing comments...

From what I have seen of his output, I feel that Rankin has shown time and time again that he understands the constraints and challenges that can come with designing for manufacture (DFM).

His designs demonstrate balanced optimization and rationale, evident throughout all stages of the project. Most importantly (in my view) and unfortunately, all-too-often overlooked, is the clear , mutually agreed definition of the customers needs - not only their problem, but the way in which it is to be solved.

Feature creep can often be the demise of a poorly defined product specification, and not only does an agreed problem statement do away with this, it also sets a clear path for the project structure, and can keep things on track by identifying potential production problems before you've even started prototyping!

It can also give you something to cite if there are any differing opinions at a later stage. This can potentially save you a lot of money in the event of a customer who doesn't feel your solution, for whatever reason, satisfies their requirements.

Finally...

This has been a bit of a lengthy article, and if you're still reading, thanks for sticking with me!

I hope you have found some useful tips in reviewing some the design tricks Rankin has put into practice while prepping his Shot Core product for production. Maybe you knew some of them already, but perhaps you've picked up something new that you want to try out on your next design?

If you're not already following what Rankin is up to on GitHub, I'd advise you give him a follow there. If you're more tuned into Twitter, here's his handle — go and hit him up!