July update


Hello everyone! I hope all of you here in the States had a terrific Independence Day (and most importantly, managed to keep all your digits)! I realize It’s been quite a long while since the last update was published so I spent a good portion of this week reflecting on what we’ve been doing recently and I’ve put together an new update for you. A LOT has happened since I last did this and there is a lot to cover, so why don’t we just get started?


So when we last left off FedEx had lost 4 of our 5 main board prototypes but Zack had just received our last remaining PCBA in the mail and was about to do some testing to see if the resistor that had been placed on the board solved the problem. After a week or two of trying to bring the bluetooth module online without any luck, Zack reached out to the manufacturer of our processor to ask some questions. He spoke with their engineers about what he’d tried and they worked together to try to troubleshoot the issue.

Martin email.JPG

After exchanging information and reaching a mutual understanding of the problem, their engineers were stumped. Zack had tried everything that they would have tried and it still wasn’t working. They suggested we send our board down to them to see if they could run some tests on the hardware and see if there was a problem there. Their theory was that that specific Bluetooth module may have been bad or gotten shorted along the way. We were a little hesitant at first to send our last remaining prototype through the mail again after what had just happened, and at one point even considered driving it down to Texas to hand deliver it. Eventually we came to our senses and reluctantly packaged it up and sent it on its way.

Martin email 2.JPG

The unit arrived a few days later and before we even received confirmation that it had arrived, their engineers had already tested the unit, identified the problem, and fixed it for us. It just so happened that there was a microscopic break in the solder of the newly added resistor and the only way they were able to tell was because of very expensive equipment.

With that problem now solved we had the board returned to us. After helping us out with this, their engineers strongly urged us to have more prototypes because doing this development work is hard enough when you have multiple boards, but if you only have one board then it’s impossible to tell if there is a problem with the design of the board or just with the manufacturing of the design without sophisticated equipment. We agreed that it would probably be a good idea to go ahead and do so, but before we did we had our hardware designer incorporate the change into the design of the new boards.

A few weeks later the boards came in and appeared to be in good working order. The changes were made successfully and everything seemed to run as expected. Things were good. That is until we discovered that the Bluetooth module can only run on 5 volt power which it only received when plugged in and not when it’s running on internal battery power. At first glance that seems like it would be a big problem, but really it just means we need to add a step-up transformer (booster) to the board so we can deliver enough power to the module. That does mean we’ll need to make the design changes and produce more prototypes, which will take some more time, but we’ll still be able to do further testing while under external power so this shouldn’t set us back on time.


While all that was going on with the hardware, we were still working with Josh and Keith on the optical design. In the last update we were awaiting the optical design parameters from Keith so Josh could model the design of the lens and lightpipe. A few days after that update was published we received the design parameters, Josh modeled Keith’s design, and returned it to back to Keith so he could run a physics simulation to confirm that the modeled part matched the design parameters. It did and below are the modeled parts.

You’ll notice in the pictures with the blue lightpipe, the lens is different. That’s because the edges of the lens were trimmed since we don’t need them due to the type of camera we use, and mounting features were added so it would fit more securely in the enclosure. After looking at the design for some time we realized how much the lens looks like a duck, which we thought was kind of cool. The duck’s bill is actually an orientation feature which lets anyone assembling the product know which direction the lens needs to point in order to work correctly.

After receiving these 3D models we submitted them for injection molding quotes and manufacturability analysis. Upon first glance it looked like we had a problem. With designing for injection molding it’s important to maintain a (relatively) consistent wall thickness. This is because as the plastic cools, you want it to all cool at the same rate because as it cools it shrinks. If the thickness of the parts varies too much then the thick parts will cool more slowly and shrink more than the thicker parts that have a lower thermal mass. When that happens the plastic can have deformation in it such as sink marks, warpage, and voids.

When we submitted the lightpipe for manufacturability analysis we initially received an automated warning that said because of the non-uniform wall thickness, our part may be subject to sink marks. For some products, such as children’s toys, this may only be a slight cosmetic blemishes but otherwise not be too big of a deal, but for this part any sink marks would likely ruin the optical characteristics completely.


So for a short time after receiving this analysis we were a little worried. Although we had other potential solutions to this problem, we knew the lightpipe would not perform correctly if we changed the geometry of the part, and the part looked like it couldn’t be manufactured correctly without changing the geometry. Fortunately, over the next few days we talked with the production engineers at the company that had quoted the part and they told us that the automated system only looks at the nominal wall thickness to see how uniform it is. If the wall thickness is not uniform then it will kick out a warning, but so long as it doesn’t indicate that a design change is required then we should be alright.


So now, knowing that the parts should be able to be injection molded, we decided to move forward with some CNC milled acrylic prototypes. We ordered the parts and eagerly awaited their arrival. When they were delivered, the parts were cloudy. This was expected. Just days earlier we’d received an Amazon shipment with various buffing tools and compounds so we got to work using our Dremel to get a crystal clear finish on the parts. Here you can see the difference.


If you look closely you can see that there are still small machine marks left over from the manufacturing process, even after polishing. This too was expected and was not something we thought would meaningfully change the optical characteristics.


Once the parts were polished we did some quick, easy initial tests. We held the lightpipe over a cell phone camera flashlight to see if it transmitted light as expected and we passed the lens over some text to see how it refracted the letters. These tests obviously aren’t very scientific but they’re fun to try out, easy to do, and can give you some insight into what to expect in future testing.


What we didn’t expect, though, was that the LED that’s mounted on our camera board wouldn’t put out nearly as much light as a cell phone does. But that’s just what we experienced, or so we thought. The picture above shows comparison of the lightpipe with and without the LED on. You can see from this that the output from the LED is almost meaningless. It's not until an object gets within about 1/3rd of an inch that you can see light beginning to hit it at all, and it’s certainly not enough to be our only source of light to illuminate the page. So what are we going to do?

After clearing our heads we evaluated our options. We came up with 4 potential alternatives, shown in order above

  1. Swapping for a stronger LED and reworking the lightpipe slightly to fit the larger profile

  2. Create a 3rd circuit board with just an LED, a circuit board, and a connector, attached to a split flex circuit and aimed at the text

  3. Traditional LED mounted through-hole and bent into position and mounted in the enclosure

  4. Moving from a linescan camera to an area camera which requires much less light.


However, as we were weighing the alternatives and deciding which we wanted to pursue, Zack chimed in with some good news. It turns out that the logic in the LEDs was just backwards. We later found out that the reversed logic was because they are wired to power and not ground, so sending them to 0 completes the circuit and lights them up. Luckily that means with a simple change to the code, the LEDs will now shine at full brightness! Phew!

In all, this was a very good thing. We still get to move forward our #1 design choice, and if something were to come up with the lightpipe and it turns out that that design is not feasible, we now have four other potential designs ready to be implemented at a moment’s notice.


By now you may have noticed our enclosure design has undergone some changes. In the last update the outside shell of the enclosure was pretty well finished but the internal features had barely begun to be modeled.

I’m happy to report that the internals have now been fully modeled! There is still some work to do as we test, prototype, and revise certain aspects, but the enclosure is pretty well finished.

After receiving the 3D STEP files from Josh we had a very high quality 3D print made for us. This prototype was produce using state-of-the-art “Multi-Jet Fusion” technology which quite similar to inkjet printer, except instead of spraying ink, it sprays a jet of photo-curable gel that will harden when exposed to UV light.

The reason we went with such a high quality print this time is because we needed it to be able to test the fit and function of the components. Other 3D printing methods aren’t accurate enough to give us good test conditions.

Overall the fit test was a resounding success as you can see from the photos above. The battery laid in place and did not move or rattle. The main board fit perfectly on top and screwed into the enclosure exactly as it was designed to. And the lens fit snugly into its mount and could be turned upside down without coming dislodged.


There were a few issues that need to be addressed on future iterations, however. First, in previous iterations the camera board fit in its mount so snugly that it could be turned upside down and shaken and it would not come loose. With this prototype that was not the case. There was no lateral movement but it would fall out when turned upside down. To fix this problem we instructed Josh to place a rib on the other half of the enclosure to clamp the camera into the mount.

The other issue was that while the lightpipe snapped into place and fit tightly, the rear end of it didn’t extend all the way to the LED like it was designed to. Because of this, a portion of the light emitted by the LED never made its way into the lightpipe which could be problematic. It’s possible that either the lightpipe, camera board, or the enclosure are outside of specified tolerances. We’re still working to test our hypothesis but our theory is that that’s not the case. We believe everything is designed and manufactured perfectly, but the mounting ribs are flexing as we’re putting the lightpipe in and it’s not seated properly. We’re going to make the ribs longer so they deform less when we mount the lightpipe. We’re also going to make an L hook feature around exit of lightpipe so we can put that end in first and rotate the other end into place.

Once those changes are made we’ll have new prototypes made and see if those changes fixed the problem.


Lastly, we’ve been working with an industrial designer named Anna to produce a logo for the new Whispr brand and we finally settled with this design. An interesting fact, the letter “W” in this logo is in the form of a low frequency sound wave, so it’s essentially a whisper!

Whispr Logo-01.png

Since the Gotham font used in the Hidden Abilities logo is so heavy, we wanted to contrast that with a lighter font style in this logo. The lighter font also helps reinforce our brand image of being discreet. It’s also worth noting that the font used here was custom made specifically for this design and completely unique to this logo.

Now that the logo design is complete, Anna will be helping us create the packaging graphics and quick start guide.

Next steps

The next steps are to do move on to functional testing, primarily of the optical system. If we are able to collect high quality images using our prototypes then we should be in good shape. We have some early results and things look promising, but more testing is needed before we can say for sure that we’re fine.

We also still need to undergo FCC testing before we move to production. However, due to new rules brought about because of the battery problems of the Samsung Galaxy S7 we now have to test a complete, ready for production, device rather than just the electronic hardware. That means we’ll have to wait until the unit is ready for manufacture before we can get the product FCC certified.

Once that is done, however, we’ll be able to move to production. We’ve already placed a purchase order for the processors and we’re now in the production queue for those parts. That means that once we’re ready for assemble your Whispr pen circuit boards, the processors should be just about finished us and we won’t have to wait for the full manufacturing lead time before we receive them.

Published Jul. 10, 2019