Open 3:2 Project - Liberating the Surface Studio Monitor
Like many '90s babies, I was brought up in the golden era of monitor aspect ratios. It's the aspect ration of Over the Air TV, VHS tapes and the ever glorious Sony Trinitrons of yesteryear - the 4:3 aspect ratio. Unfortunately, this aspect ratio has largely went the way of the dodo with the presence of 16:9 panels dominating panel production. luckily, more square formats are showing up in peculiar devices made by the tech giants of today, but the displays are often locked down or incredibly inconvenient to reuse for other applications. This project involves reusing the 4500x3000 LCD assembly of the Microsoft Surface Studio as a dedicated Display Port (DP) or Thunderbolt monitor.
As you can imagine, I have a real soft spot for square-ish monitors. My first real experience with high resolution monitors was a 2048x1536 20in monster of a tube made by Eizo: The Eizo FlexScan FX-E7s. It was a tremendous yet glorious beast that was thrown out at the curb. It was a wonderful resolution to behold, especially at the turn of the century, but the monitor suffered badly from ghosting. The only monitors worthy of replacing that CRT were 2560x1600 WQXGA 30in lcds introduced later in the 2000s. I purchased and repaired many of these over the years, but for some reason WQXGA LCD panels manufactured before the 2010s had high failure rates. Most of the issues stemmed from poor bonding quality of the chip on flex and poor solder quality between the LCD flex to the rigid PCB. A common and near permenant solution was simply to add 1/16in of adhesive backed foam to the chip on flex or the solder bonded area near the edges of the display, as shown below. Since LCDs at the time consisted of a stamped steel clamshell construction, this worked swimmingly.
When the surface studio came out and I saw it first hand in a Microsoft store, I knew that panel was my next daily driver. Of course, I was not willing to pony up retail price for a 6th gen intel laptop in a proprietary form factor, so I waited it out. In the mean time, I studied the ifixit teardowns and began sifting through datasheets.
Why this monitor, Why now?
It's quite simple actually. 4:3 ≥ 3:2 > 16:10 > 16:9 > 21:9, always and forever. I find that widescreen monitors just don't cut it for productivity. Sure, 16:9 is commonplace and there are hundreds of affordable IPS high frequency monitors now, but I value productivity greatly. It just feels like there is something lacking with the 16:9 desktop experience. Newly designed 2560x1600 LED backlit monitors are being produced and are a modern alternative to my aging Dell U3011 monitors, but I'm looking for something new, something that's not just a repackaged product with a brand logo embossed on it.
The studio monitor embodies a new frontier in desktop monitors. it's a 28in multi-touch panel. For desktops. That's frankly incredible. While the touch screen would be a nice feature to have, reverse engineering the chip responsible for touch may be impossible if it's an undocumented Microsoft branded ASIC or exclusive IC. Even if it's not, there are other legalities such as distributing a touch driver for the monitor and tearing it out of the surface studio OS build. If that's the case, I could only document the driver build process and not redistribute the driver. It's not a very friendly territory to be in, especially with a company that spends more on litigation than some countries GDP.
It is substantially thinner than most monitors on the market which is a major boon to work space optimizing folks like myself. On top of that, the panel looks really nice in person and has a dpi scaling of 1:1 to the native windows dpi. This means no rectangular, stretched pixels found on some 21:9 panels. Everything scales nicely and looks correct.
The Surface studio panel is edge-lit, meaning the panel does not have the perfect backlight uniformity found on CFL backlit panels and directly backlit LED panels. Anandtech covered this in their in-depth review. The 28in touchscreen acts like a large glass mirror in the right light. This is of course a substantial problem to overcome, but there are anti-glare coatings available for the surface studio that I am eager to try out.
Since this is a challenging project, I plan on releasing a displayport only version of the monitor adapter first. The DP implementation should be easiest to implement. While edp and DP are similar interfaces, I believe I will need a DP redriver chip to reduce ringing and losses across a DP cable, to ensure a cleaner signal. This is a ~5k LCD so it is technically possible to drive 5k 60hz LCD on Displayport 1.3. It will completely max out the bandwidth of a certified 1.3 DP cable, however from my initial findings, it may require an active DP cable.
The second revision of the monitor adapter will feature a dedicated thunderbolt chipset, mainly because it is possible to pass through DP lanes via thunderbolt 3. luckily, because the surface studio is not 6k, there is some bandwidth left over for USB 3.2 lanes through the dp cable, meaning a USB 3.2 hub will be possible to integrate into the monitor. A similar example is the Apple XDR monitor thunderbolt implementation. From my research, Apple implemented thunderbolt in a peculiar way. I'm fairly certain they used an Intel titan ridge TB3 controller, but they also have DSC (display stream compression) support, which isn't stated in the titan ridge datasheet. The XDR display would have saturated a normal DP bus, which is why thunderbolt was apple's only option. However, this still may not be enough bandwidth, which is why apple's tb3 cables are active.
What is interesting is that a lot of current day GPUs have DP 1.3 and DP 1.4 compatibility, but many don't have the DSC specification implemented into the Displayport IP block of the graphics card.
To my knowledge, the only GPUs that have DSC are AMD Navi based cards. This is why folks testing the XDR display are only getting usb 2.0 speeds out of their XDR type-c hub, unless they use it with the 16in MBP. You need DSC support to reduce the required bandwidth of the monitor, freeing up the USB bus.
I am currently reverse engineering the eDP interface on the surface studio monitor, but I can't seem to find the exact datasheet. I am referencing similar panel datasheets from Samsung in the mean time. If anyone here has any tips or tricks on finding oddball LCD datasheets, let me know! I am looking for the Samsung LTM282RL01 datasheet.
On to the build. I came across a batch of prototype surface studios on the inter-webs a few months back and I simply couldn't help myself. The panels were most definitely from prototypes, considering there was cutouts in the back of the milled aluminum housings and the seller was located nearby Redmond, WA. Gosh, the local Goodwill's near the Microsoft HQ must be true goldmines. Anyways, the cutouts were obviously not present in the final production monitors but provide easy access to a populated FPC connector on the main control PCB on the back of the monitor. This is definitely a location that is in need of investigation with careful probing. I reached out to the seller a few weeks after the sale and asked some questions, but it he claimed to know nothing about them and did not have any other parts associated with the prototype panels. Bummer.
Some other obscure properties of the monitors were milled slots in the monitor back. It's as if early designs of the surface studios implemented an antenna slot design in the rear of the panel housing. Besides the various changes in geometry, there were engraved debug numbers, most likely from different builds, and various stickers from supplier quality checks. The mandarin characters are a dead giveaway.
On to the teardown. Ifixit's excellent and thorough teardown is available for perusal here, but I'm going to compliment that here and dive a bit deeper into the chips used, the PCB layout and the overall function of the computer with some low level block diagrams.
The mechanical liberation was quite simple. The studio uses standard 10-32 threaded holes to mount the display to the base stand. A nice caliper, ruler, CAD and a few hours yielded a very appropriate monitor mount solution: a 5052 folded bracket. Due to the ubiquitous nature of Dell monitor stands in industry, I designed a simple bracket with slots that mate to dell stands from ~2007-2023. This is an awful lot of stands available for reuse, which is most excellent. However, if one wants to benefit from a rotating stand, the scenery is not as rosy. The 28in beast of a monitor hones the monitor stand selection down quite considerably. The Dell UP3017, P2717H, and P2719H stands are most suitable, and whatever newer derivatives produced.
For giggles, I had the monitor mount adapter laser cut from white lexan. Technically, a little even heat should yield a part of similar geometry as the 5052 part. My main concern is the amount of flex because of the off center nature of the original surface studio threaded hole locations.
The goal of this project is to resurrect an otherwise excellent monitor from a fire of 1000 deaths by ewaste. I am designing every aspect of this around ease of use and low cost of entrance. To help spur adoption of this adapter, the implementation needs to be straight-forward and as little effort as possible. Thus the documentation and procedure needs to be as simple and quick as possible. It's beneficial to reduce ewaste and to use as much of what is available as possible. Thus, it's wasteful to drill and tap additional holes in the monitor if that means someone else is put off from doing the conversion process. Only original mount points are used because of this.
Want more? Here's a behind the scenes look at my workspace and some of the images that did not make the cut to be included in the write-up: