Crowdfunding an ESP32-Based Phone

Last time we took a look at the WiPhone, back in the middle of last year, the ESP32-based Voice over IP (VoIP) phone was closing in on full funding on Kickstarter. Since then it’s been a bumpy ride for the project towards shipping to its backers. That’s to be expected, as scaling from a single prototype to a full scale production run isn’t easy. Everyone runs into problems with manufacturing, and how you handle that as a startup that is going to decide whether you succeed or fail, not the problems themselves.

Measuring 120 x 50 x 12 mm, and weighing in at just 80g, the WiPhone comes in two versions and colours. The basic edition of the WiPhone has a polycarbonate frame, while the more expensive WiPhone Pro comes with an anodised aluminium frame.

However, uniquely perhaps amongst the many phone GSM-based self-build handsets available, the WiPhone is a VoIP handset. Built around the Espressif ESP32, which provides the onboard Wi-Fi support, the phone has 4MB of PSRAM and 16MB of Flash memory, along with a standard 3.5mm audio jack, and an internal micro SD card slot. It uses a 2.4-inch screen, driving the 320×240 pixel display over SPI, and has a micro USB connector for charging, serial communication, and programming.

WiPhone is expandable through daughter boards, and the whole back of the phone is a replaceable panel that accepts a standard 1.6mm thickness PCB, which you can use to add more functionality. Connection is via 20 pogo-pins, arranged in two 2×5 pin blocks, on the back of the main board which offer UART, SPI, I2C, PWM, ADC, and GPIO.

We recently sat down with Ben Wilson from HackEDA, the folks behind the project, to talk about their journey from prototype to product

You started shipping evaluation units ahead of schedule?

We shipped the first evaluation unit in June of last year (a pre-production unit). I believe the first units I’d consider production quality went out in November (2+ months after we said we’d ship), but I’d need to go back through the data to get an exact date.

… but the main shipment of units hasn’t gone out yet? Why is that?

Really it’s not just one issue, more like there are a host of things that come up when you move to shipping higher quantities vs. carefully making a few bespoke units. Schedule for hardware projects is one of those things that never works properly but is still needed to keep forward progress. We aren’t interested in shipping bad hardware, so we’ve delayed shipping rather than ship something we aren’t satisfied with.

Additionally, as we built more and did more testing, we noticed more things that could be better or are actually something people are going to need for the device to function correctly. And as ubiquitous as phones are + how much time has gone into optimizing the user experience, appearance, etc. of the expensive phones people are accustomed to using, the par is pretty high for this type of device. Most phone design teams are orders of magnitude larger than ours… we do the best we can to design it right as we go because we can’t afford the extensive user testing, custom test equipment, expensive prototyping, or supply chain verification that a typical smartphone would go through.

And finally, we’re a team of almost exclusively engineers, so we can’t ever let anything go without noticing 5 new details that could be optimized.

So it was a lot of fit and finish issues? And right when you thought you were done, more fit and finish issues?

See above, but also a list of issues, listed roughly in the amount of delay they’ve caused:

— The adhesive for the front cover looked like it was a solved problem, then we started noticing bubbles in some of the longer term tests. Now we need to find and verify a new adhesive for all the clear-front phones.

—  PCBA supplier can’t keep the defect rate down so we switched suppliers multiple times.
 —  The frames have a lot of smaller features and it has been difficult to get parts that consistently pass all our QC checks.

—  Tolerance issues surface, what fit fine when we made 5–10 pieces no longer always works when we make more. So we have do dig down and find the root cause and fix it. For example, the thickness of the PCB can vary +/-10% and remain in the supplier’s spec (tightening that would mean significantly longer lead times and much higher prices, plus the number of available suppliers goes way down). But aligning the USB connector to the hole in the frame looks like shit if it varies by that much.
 —  A few smaller issues, needed to add some sound holes so the loudspeaker isn’t muffled, changed the back cover thickness so that people can make 1.6mm thick daughterboards and the connector doesn’t interfere, etc.
 —  Some issues with color and texture consistency.

How do you decide which suppliers to work with when manufacturing in China?

Many times hardware projects just assume that once you have working prototypes, moving to manufacturing is a matter of adding some zeros to the order quantity. But jumping to production quantities like that almost invariably means now you have a whole lot of product that is out of spec or not working correctly and either gets scrapped, dumped onto unhappy buyers, or needs extensive rework. We have a rule that the supplier has to deliver at least one smaller batch with no significant issues before we go to larger batch sizes. If you jump the gun and order lots of parts too soon it’s really hard to tell a supplier they need to remake them all and not have them turn into a wet noodle (the supplier, not the parts). Also, small batches tend to flush out design or assembly problems that should be addressed before moving to larger orders.

Choosing a supplier in China, at least at the scale we work at, is as much an art as a science. For projects that ship less than 10,000 units, there really isn’t any budget to get fancy with supplier vetting. Doing a site visit and interacting with them can give you a feel for the honestly level and overall capabilities. You also want to pick suppliers that are the right size. Right size means the orders you are giving them are big enough that they need to pay attention to you, but not so big (or not so technically demanding) that they can’t handle them. If we tried approaching one of the giant manufacturing houses they would just tell us to get lost because the order size isn’t big enough for them to do it cost effectively. But if they did somehow accept the order we’d probably get assigned as learning fodder for some brand new project manager just to let them fail with us so they can have some practice before moving up to a real client. The size jobs we have often fit well with the smaller shops where the owner is the project manager for every job that goes through the shop, up to one where there are a few engineer types acting as PMs for the jobs that come through.

So you’ve done a lot of small runs, that must have been expensive?

Yes. We’ve spent well over the amount we raised at this point.

But you now have FCC/CE certifications?

Yes, FCC/CE/IC, and I think the Australia one.

What lessons do you think you’ve learned?

We knew going into it that we didn’t have any real marketing skills and probably should have had someone dealing with that more. We did I think a decent job and raised a respectable amount of money, but really it’s nowhere near what it required to design and manufacture a phone from scratch.

Are you happy with the final product?

Yes, I’ve been pushing the hardware side for quite a while now and it’s just about to the place I feel really good about it. Software is getting there but not at the same point yet.

You’re being held up right now by the Coronavirus?

The Chinese government extended the Lunar New Year Festival due to the coronavirus and told everyone to start back to work on the 10th. We can’t do any production work until at lest then (and it’s looking like it could be longer). It has given us time to put together a first draft of our documentation.

Once the phones are shipping, you have daughter boards to work on?

We’ve been working on daughterboards for quite some time and will continue to do so as and after we ship WiPhones.

What daughterboards are going to be first?

LoRa was promised as some of the backer rewards, and we’ve also added a couple breakout boards that people can add to their Kickstarter shipment.

Others that are approaching the user testing stage: LTE, GSM/GPRS, MEGA battery pack, RC car, RGB array/Sparklepony, prototyping board, and a barcode scanner.

Will there be crowdfunding for new daughterboards?

No definite plans, but yes, there is a high probability.

It’s a tough journey to take a prototype to a final product, but the WiPhone has almost made it through to the far side, and will be shipping as soon as practical given the current conditions in China.

In the face of its delays the project has kept its backers on side with remarkably good communication and regular updates. Something that other crowdfunding creators could learn from, most backers expect delays, and while they won’t be happy they aren’t going to complain too much, unless they just don’t hear anything from the project creator. Being ghosted by the creator is one of the biggest problems with hardware related crowdfunding projects.

Wilson says that he is already planning the next version of the phone that will support encryption, as well as the possibility of providing a service to rapidly customise the base design and make a pseudo-custom device based around custom daughterboards.

It was Chris Anderson that coined the phrase“the peace dividend of the smartphone war” arguing that “…when giants battle, we all win,” and it’s that smartphone war that made building your own phone a possibility. We live in interesting times!