The last few weeks have been extremely busy, and as promised, we have a lot of new information to share! This is a long one...
To manufacture the first collapsible display, we have been managing eight key component suppliers (screen, screen sub components, fabric, sewing, retractable arms, custom optics, protective cover, and projector) for a variety of components that have never been made before, including the first wrinkle mitigant rear projection screen, the first collapsible display mechanics/retractable arms, and the shortest throw pocket projection optics on the market. Additionally, we have been working very closely with our main contract manufacturer on the final assembly of SPUD and manufacturing of the projector part of SPUD (electronics, firmware, projector, and chassis). We have also been working very closely with our certification labs on country certifications. All of it has proven to be very difficult (as you have seen through these updates, not a single component was easy!), but we have resolved all major issues. The optics issue continues to set itself apart as the most complex, difficult, and frustrating issue we have faced. Though, as of this week, we are at a point where we can see some light at the end of the tunnel!
This update includes:
As a reminder, we are currently exploring two potential resolution paths in parallel: (1) Improving the existing lens design manufacturing process until we can consistently meet our tight error requirements, and (2) validating a new design that accommodates the shift of the most shifted projectors.
Option 1: Improving the Existing Lens Design Manufacturing
When we initially manufactured our pilot units, we discovered an issue where the projector beam could be shifted more than the projector manufacturer specified. That was a problem because our existing optics system was not designed to accommodate the most shifted projectors. We were able to correct for most of the beam shift by designing a new adjustable lens mount (pictured below), but it still meant that we could only afford to have a very thin margin of error for manufacturing the front lens itself. This has put a pretty large burden on our supplier to be able to consistently produce that front lens at an acceptable quality.
For a quick illustration of the process we have been dealing with: Injection molding is a manufacturing process for producing parts by injecting molten material into a rigid frame (i.e. the mold). It is ideal for mass production, as it is a relatively quick and repeatable process. It is often used to create the plastic housings of consumer products, including our own SPUD chassis:
In previous updates, we talked about a potential solution that involved first molding the lenses, then diamond turning (precisely cutting) those lenses to achieve our error requirements. We knew going in that this would be a much more expensive and slow-moving process. After creating some initial samples, our supplier decided that they could not tie up the machine for production quantities given the time and cost they would have to absorb (and the need for other customers to use the machine). To put it lightly, it was a very frustrating experience, especially since it was the supplier that originally proposed the idea as something they were willing to do. Therefore, we had to focus on improving the injection molding process.
Injection molding the lens (the largest lens in the 3 lens assembly and bottleneck for our project) is difficult for 3 key reasons: (1) It has a very large and abrupt variations in thickness, which makes it extremely difficult to mold; the more uniform a part, the easier it is to mold. (2) The lens needs to be retangular to minimize SPUD’s collapsed size (and it can be since the projected image is a 16:9 rectangle), and non-circular plastic lenses are inherently more difficult to mold since they are less uniform/symmetric. (3) The errors on both surfaces of the part need to be controlled to approximately 30 microns or less, which is very difficult with the thickness variations and the rectangular aspect ratio. To address these issues, since April, the supplier has been trying varying molding process parameters as well as design parameters, including different molding techniques (injection versus compression), different mounting options, and lens modifications (removing material from the thickest parts):
New, thinner lens (left) versus original PVT lens (right).
They have made measurable improvements over these last 2 weeks, and the latest samples we received last week have been the best to date: it appears that the combination of (1) reducing the thickness and (2) making the mount adjustable has improved the manufacturability of the largest lens with the injection molding process. Additionally, through experience, we have also gotten better at correcting the errors through manual adjustment of the projector and the lens. That said, the parts are not perfect, though the resulting image is better than the vast majority of the pilot units (except the pilot units with the most centered projectors). The supplier will to try to further improve the quality, but it may be difficult to improve beyond the current quality.
Here’s what we’re actually seeing. Note that when the pictures below were taken, we had not made any adjustments to optimize the projector or lens position, and those adjustments will further improve the performance. This is a good baseline for the type of performance to expect:
With the diamond turned lens (left), everything appears in focus. With the injection molded lens (right), most of the image is in focus, but “Find and Select” in the top right are slightly blurred. This lines up with the thickest part of the lens (inset).
What this means: These latest lenses are very close to the specs that we need. However, we don’t know if our supplier can improve the lens any more than they already have. The defects correspond to the thickest part of the lens, which makes sense since that is the area where is it most difficult to obtain the optimal plastic flow. At this current level of quality, we plan to make manual adjustments to optimize the image on each individual SPUD.
Within weeks of first discovering the lens manufacturing issues, we decided to start working on a redesign option in case we could not improve the molding process enough on the existing design. The main goals of the new design were (1) to accommodate the most shifted projectors without needing an adjustable lens mount and (2) to optimize manufacturability for molding. Dr. Aaron Bauer, an expert in the field of freeform optics design with an Optics PhD, redesigned the system from the ground up. The new design is now final and we have been trying to prototype the design. We are working to create prototypes, but the initial samples have had multiple problems, including scuffs and cracks:
The prototype delays have prevented us from validating the new design. If we do reach the point where we can validate the design and everything looks great, we have a supplier that is very capable of injection molding for production. However, this would come at a significantly higher cost, and a longer schedule requiring new molds and another round of validation. We already know the largest lens is a difficult part since it is similar to our current largest lens (i.e. there is risk that the new lens would perform no better or worse than the current lens).
We have reached the point where we think option 1 has more merit than option 2, since the option 1 quality has improved and option 2 comes with an uncertain (though certainly longer) schedule, uncertain quality, and high cost. Given the latest results from our lens supplier, we believe we now have the ability to move forward with option 1. The necessary optimization to the lens and projector will add additional labor costs to our assembly process, but it puts us on the shortest path to release. The quality we have seen this week gives us a baseline, and we expect the manual adjustments to each individual SPUD to further improve the image quality. Note that while we do expect to make improvements through manual optimization, there is a real possibility that we have already achieved the best results from this method, and that SPUD may have minor errors (like shown above) that could impact the image.
Option 2 is still on the table for now, at least until we receive some usable samples back for validation. If the initial results are amazing from the start, we may reconsider moving forward with option 1. But as it stands today, it doesn’t make sense to take on the the additional time and cost with so many variables and unknowns still in play.
The rest of the system components are not very interesting, in a good way. In short, all other components are on track for Mass Production (MP):
A lot has happened over the last few days, and we are still finalizing a new schedule. To give you an idea of where things stand, here’s what we do know:
There’s a whole list of other tasks that will need to be completed before we can release, but this optics related list represents the critical path of our remaining schedule. Assuming we move ahead with option 1 as discussed, we will be aggressively pushing to release this year. If for some reason we are forced back to the option 2 path, that would likely push our schedule into next year.
As always, here are some shots we’ve taken over the last few weeks:
We can’t thank you all enough for helping us start this project and supporting us through the ups and downs. We know that this lens issue has been especially frustrating (it certainly has been for us), and we are forever grateful for your support and encouragement! Your early support has been critical in bringing the first collapsible display to life, and as we move into this next “released” phase of our journey, your continued support, feedback, ideas, and general communication will be an essential part of our decision making process moving forward!
We are extremely grateful for your support!
- Justin, Leonard, George, Alex