A couple of weeks ago, I started getting texts and emails from friends with an attachment to a blog post. A person named Paul had engineered and built himself a lay-down desk, wrote about his experience, and posted it on HN. Creating a lie-down desk is a big project in itself. To make even a static lie-down workstation involves understanding the human body, understanding workflow, building mechanisms to make everything seem simple, and then trying to do the same thing for a large variety of people and their specific workflow. To add, the idea of moving the system from standing to sitting to recline is much more complicated. I’ve been working on this problem for ten years, and it is rare to come across someone familiar enough with the situation to have a really good conversation about it.
It’s the type of problem I call “seductive.” At first, it seems like an easy, clear-cut problem to solve. We all use recliners, we all use desks, we all use chairs...surely, you can just take a part from this and a piece from that and call it done. Well, we tried all that and came up with prototypes that kind of worked, at best. For anyone that has built and tested their prototypes, the problem comes when you have a solution, but you realize it’s so clunky, so hard to use, or just finicky that you stop using that feature, and you keep refining to find a better way. Since not many people have been down this specific lie-down workstation rabbit hole, I was interested in chatting with Paul. We set up a video call and had a great discussion.
Paul had a detailed understanding of how difficult a problem like this is to solve. To be upfront, I want to make sure everyone understands that Paul built a DIY solution for himself, with very limited tools and time. In no way is this a comparison to the Altwork Station, which has ten years of development, five generations of prototypes, and three generations of production hardware, all adding up to 80,000 hours of design and engineering. The creations themselves are apples and oranges, but the fundamental problem is the same, and my discussion with Paul reminded me of the fun we had solving it. In this post, I’m going to touch on just one of the features of the Altwork Station that I find fascinating. Let’s go down the rabbit hole.
What’s So Hard About Headrests?
We have all probably noticed that most office chair headrests are basically useless. If a headrest is comfortable when you are reclined and looking at the screen, then it pushes your head too far forward when you are sitting upright. Easy to observe but a great example of a rabbit hole because come on, how hard can a headrest be. It is important to note that the solution we finally arrived at is unique, and we have been granted patents on these solutions. It is not like we could have gone to some catalog and picked from various designs. It required genuine research, development, and invention. Below I will share the development and testing we did to figure this out.
First, let’s start with the final design. Our fundamental design goal was to have seamless motion of the machine from sitting upright to fully reclined. In this video, you can see how the aluminum arc extends and then retracts while the backrest reclines, resulting in zero need to adjust the headrest.
This headrest moves to where it’s needed throughout recline.
Summary of what we found: A headrest is simple if your torso stays at one angle throughout the workday, but if you change that angle even a little bit, the headrest needs to change too. Further, it turns out that when a body is upright, it prefers to let the spine do the work of supporting the head, so the head is free to move/swivel. As we begin to recline, the spine continues to support the head UNTIL the muscles at the front of the neck become strained. At this point, the headrest needs to move to its farthest forward point of travel (which is remarkably farther forward than you think). As the person reclines even farther, we need to retract the headrest to keep the neck comfortable. The final trick is to figure out how to do all of that for people of different sizes.
How did we get there? Study, prototyping (step and repeat) until we understood what we needed to do, how to do it, and how to make it look good. I include pictures that we have never published before. The old prototypes are ugly (many people considered them scary - the running joke was to say it looked like the exoskeleton Ripley uses to fight the queen in Aliens), but these prototypes were essential to the development.
Our development plan was:
- Prototype 1 (P1): Build something to understand the problem and measure comfort
- Prototype 2 (P2): Incorporate lessons from P1 to prove we understood the problem
- Prototype 3 (P3): Build a prototype with all required features; functional but no effort into aesthetics
- Prototype 4 (P4): Keep the functions, but incorporate them into the intended design silhouette
- Prototype 5 (P5): the final prototypes before production
Let’s start with the second generation prototype (P2) because the first-gen hardly worked at all.
We were focused on bigger problems at this stage of development, so there is no surprise that this design didn’t work. At this stage, the lesson learned was that the headrest had to move a surprising amount forward from fully upright to fully reclined. The second lesson was that this pivot point was completely wrong. We were still thinking about the Altwork Station, more like a hospital bed at this point in development. We had started to figure out how wrong that was, but we still thought we could get away with some sort of hybrid solution to make the engineering easier.
This image is actually a rebuild of the same prototype above (P2). It was common for us to cut off portions of the prototype to weld on a new concept for that portion. In this picture, we are experimenting with a mechanism that allows a lot more forward travel. We got as much development out of P2 as we could and moved on to Prototype 3.
The third prototype (P3) was called the “engineering functional prototype.” Its goal was to have all the required features functional, but no effort was put into aesthetics. The headrest design in P3 allowed a tremendous amount of adjustment. Clearly, two sliding bars with a big wing nut in the middle was not going to be received well, but it worked for test purposes. Lessons learned: This is the amount of adjustment we need. The headrest can work well with a cam drive function. Figuring out to then balance the engineering requirements with aesthetics was looking daunting. What we didn’t realize at the time was that we didn’t really understand why we needed this adjustment. We were compensating for lack of knowledge of body mechanics with a large amount of manual adjustment.
The 4th generation of prototypes is when we started to work on the aesthetic silhouette. We knew it would be hard enough to work the overall shape, so we planned to work the surface refinement in the fifth-gen protos.
The designers hated seeing a “spike coming out of the user’s head.” That is a fair view, so this concept was scrapped. You can also see the first attempt at the “S” curve shape to support either the top or bottom of the skull. So we modified it into a gear-driven approach that was low profile and high travel. The plan was to cover the mechanism in the production model. Our attention to detail went so far as to test this with someone with long hair, as we worried it might get caught as the mechanism moved. With this system we felt like we had the solution.
But then we got an idea. The timing was terrible, but we couldn’t let the idea go because it was an idea based on the fundamental structure of the body. It sounded good, so we prototyped it, and it was just so, so much better. At this point, we had spent four years figuring working this out and here we were completely starting over, but it finally just all came together. In this new and final design, an arc bends around the natural pivot of the upper back and neck. The key difference here is that when we started, we thought the skull pivoted around the neck area. Spoiler alert: it does not. Then we thought it bent in multiple places, more like a snake; it doesn’t do that either. There is actually a pretty repeatable pivot point that is a function of height located around the last cervical vertebrae. Our arc is centered at that pivot point, and the user can easily adjust the height.
This is the very first version of the mechanism we used to test the concept. Once we figured the actual pivot out, the only adjustments we needed were:
- Size of the head from the back of the skull to the pupil
- Height of person from the pelvis to the skull
- The tilt of the support. For reasons we are still not certain of, about 90% of test subjects liked the bottom of the skull supported, and 10% liked the top supported. We were able to accommodate this with a simple pivot.
Once we proved that it was the right concept, we widened and flattened the arc piece to decrease the wobble in the pivot feature and improve aesthetics.
Vise grips are not included in the final model.
In the picture above, you can see we used a vise grip to adjust on our early testing. Of course, we had to engineer and test the heck out of the concept to develop it, but the idea was there.
Here is an example of life testing using a milling machine to cycle it. One could argue we spent too much time testing, but our aerospace experience led us to believe there is no such thing as too much testing. We wanted a machine that could reasonably last decades.
Our DIY life testing setup.
The first version of the fifth generation of prototypes.
The fifth generation of prototypes was our last. We also made five of them to be able to work details faster. Of course, we named them “The Final Five” because we are Battlestar Galactica geeks. If you compare the details of the picture above to the image below, you can see many of the differences we made in development. This post is only about the headrest function, so I won’t go into those, but I hope it is indicative of all the refinements we made in that process.
The last version of the P5 series.
I hope you found this discussion helpful in understanding how complex our problem was and how we could get so engrossed in solving it. It’s been a fun trip down the rabbit hole, and if you would like to learn more about the engineering behind our machine, I’d be happy to write another post about another rabbit hole.
Che Voigt is one of the founders and CEO of Altwork. Please email any questions or comments to firstname.lastname@example.org.