1: Introduction

This blog has been written for those who are interested in the development of Globe and the design behind it. I have tried to keep the engineering terms to a minimum leaving it accessible to all, but if you do come across something that you don’t understand please do write a comment and I will attempt to better explain it.
I have also tried to keep the posts honest; where something has gone wrong either by a silly mistake, oversight or just bad luck, I have left it in. If you are still trying to decide whether to back this project then please don’t be dissuaded when you come across something like this.

The blog posts are meant to be read sequentially and to that end I have tried to make sure their publication dates place them in the correct order, but just in case that goes awry I have also numbered them.

If this project meets its funding target on Kickstarter I shall keep this blog going to provide updates on the progress of the project so do keep checking back here and if there’s something you would really like to see included in the final version, let me know!

 

24: All systems go

We have been funded on Kickstarter, which is great! It will be another couple of weeks before any money shows up in the business accounts so in the mean time I can be doing a load of admin.

I will be migrating this blog from wordpress.com to wordpress.org over the next couple of days. Theoretically everything should remain the same, but I expect there will be a few hours of downtime so don’t be alarmed if you try to get on and it’s dead. The main reason for this is so that it gives me a little more autonomy over the site, since I am going to need all sorts of things like pre-order forms, extra pages, on site videos, proper email addresses etc.

For those who haven’t been to the Kickstarter page, the rough action plan is as follows:

Step 1 – Get PCB v0.4 ordered.
This PCB has 3 main difference over v0.3. Firstly, there is no external HDMI input, instead the video feed comes directly from the Raspberry Pi’s GPIO headers in the form of DPI. While v0.3 had a Pi mounted on it by the end, it fed its video via an HDMI cable fed back into the board. The second major difference is some big old slots cut in the PCB. I held off incorporating them before in case they resulted in significant loss of structural integrity (not something you want in a PCB spinning at 900 rpm, and especially not in one you can’t afford to replace). Fingers crossed this one won’t fail, but if it does, it’s a failure that has been budgeted for. The final main difference is that the frame synchronisation will now be governed by a hall effect sensor, rather than by roughly matching the rotation speed to the frame rate of the HDMI source (which is what resulted in the god awful line synchronisation issue which is really really obvious about a minute into the video). v0.3 was actually capable of this, but due to lack of time and slip ring data channels it never got implemented. Doing so will mark the last major revision of the VHDL.

A few other tweaks will include getting rid of the ridiculous 1 meter latch signal trace that for some bizarre reason didn’t seem like a problem at the time, making sure that the PCB is insulated from the shaft (ground loops seemed to play absolute havoc with the USB and HDMI), and switching to a slightly friendlier voltage regulator (the last one I used required about 20 different valued resistor and capacitors). The PCB design is ready to go, so all I need is the money to arrive and then I can submit it for fabrication.

Step 2 – Organise a load of admin and design related things.Logos, a more professional website, a proper Github account, software licenses, some graphic design work etc. While most of this is not urgent it would be nice to get some of this stuff underway early.

Step 3 – Decide which workshop space I will use.
I recently looked round a great maker space called Makerversity, based in Somerset house in central London. It has a good vibe, lots of start-ups, small businesses and other creative types all working alongside each other.

The other option is to work out of the University of Kent at Canterbury, where I have been based for the last 8 years while I did my bachelor’s and PhD. Our mechanical workshop is pretty well kitted out and it would be very easy for me to simply continue working from there.

I’ll let you all know exactly where I end up. If any of you ever end up there then you will of course be welcome to come and have a look behind the scenes, and grab a drink.

Step 4 – Prototype new mechanical design.
The latest PCB will replace v0.3 in the prototype, since its mechanical housing is already in place and working. This will allow me to start working on the new code as soon as possible as well as identify any further problems that will need to be ironed out in V0.5.

The next step will be to create a new prototype of the mechanical side of things, which will require almost a complete redesign. I still haven’t got round to weighing the first version, but I wouldn’t be surprised if it was close to 40KG, mainly due to the steel frame, top, bottom and rear plates, cast iron bearing flanges and squirrel cage 3 phase motor.

We will be switching over to a BLDC motor, a shorter, aluminium frame, replacing the steel plates with either aluminium or composite and a host of other improvements with the aim of cutting manufacturing costs and simplifying the assembly process.

Step 5 – Software and smartphone apps.
The core software that will run on the Raspberry Pi will be open source. More than likely I will end up doing the majority of this myself. The plan is to expand and incorporate some existing open source software, such as Xplanet and EarthWindMap to allow a variety of planetary and meteorological data to be displayed, as well as some additional content.

The smartphone apps will probably be contracted out, once I have the specification for interfacing with the main app. If I have time I may do it myself, but budget has been set aside for having it done externally if need be.

Step 6 – Design and prototype PCB v0.5.
This PCB will not be a huge change over 0.4. The main differences being switching over from a Pi Zero W to a Compute Module (or similar). For those who aren’t familiar, the Compute Module is an almost completely broken out Raspberry Pi with a SODIMM connector and are the preferred device for those who want to use a Pi for a commercial purpose. It provides a couple of bonuses over the Pi Zero. Due to the fact that basically every peripheral is broken out (USB, HDMI etc.) we can A. potentially switch back to using HDMI from DPI (reducing noise and potentially freeing up some IOs on the FPGA) and B. Dispense with the USB through slip rings and instead program our FPGA directly from our Pi. This will in turn allow us to reclaim the space at the bottom of the PCB where the cables went, allowing us to increase our resolution up from the 240 of the prototype, up to 256.

In some ways it would make sense to jump straight to this design, skipping v0.4. However, because the Pi Compute is fairly different from the Pi Zero W, I will need to prototype a smaller PCB first to verify things like a Wifi module, voltage regulators, the direct HDMI to FPGA interface etc. which would take additional time, and I really want to get Globe out of storage and back into action ASAP.

Step 7 – Test the damn thing actually lasts and complies with regulations.
Once we have our new mechanical housing and PCB, we should be ready to start testing. I aim to delivery you a really reliable, solid device, and that means it needs to be capable of operating for several hours, every day, for a prolonged period of time. The major things I will be looking at will be the slip rings, bearings and motors. I will be looking to develop 5 or so of these devices, which will also allow me to illustrate the different cladding options. These test units (assuming I haven’t tested them to failure by then) can then serve as demonstration units.

Step 8 – Order the components and build.
By this point I should have ironed out any final kinks in the design, producing any additional PCB and mechanical prototypes as required. This leaves the main task of ordering all the individual components: PCBs, slip rings, power supplies, motors etc. (most of which are coming from China) and once they arrive, assemble, test and package all of your wonderful whirly Globes.

23: Live on Kickstarter again

New link https://www.kickstarter.com/projects/205276705/globe-a-spherical-persistence-of-vision-display

After a cock up with the title (mostly my fault) the project has been re-re launched with its lower funding target. Everything else is pretty much the same, the only major difference is that Globe will have an acrylic display case rather than glass, saving a load of money and therefore allowing me to keep the unit price roughly the same.

 

22: Gearing up for Kickstarter…again

So the first campaign didn’t succeed, but nor did it completely flop. Of the £60,000 target we hit about the £22,000 mark, with a few people dropping out along the way when success was clearly unlikely.

I want to say a huge thank you to all of my backers who helped get the project this far.

There were a few things that went wrong in the campaign, some of which were my fault and some which were just bad luck.

  1. I didn’t do a very good job of illustrating what Globe could actually do. Although I tried to make it clear that Globe is essentially a spherical monitor connected to a Raspberry Pi, I think my assumption that most people who would be interested would have technical background was false. I received several messages saying “can it display xyz?” the answer being a very positive yes. If it can be displayed on a regular monitor then it can be displayed on Globe, it will just be wrapped round the surface of a sphere. For every 1 person who messaged me though, there were probably several others who simply left the project page because of this lack of clarity. I will do my very best to better explain that for the campaigns reboot.
  2. Globe didn’t gain a huge amount of traction in the tech sphere. While Globe did get a feature in a couple of the online tech sites (digitaltrends, make.de) and on a radio show (FM4.at), more or less it came and went without much interest.
    While I did try to contact a number of other places all my emails went unanswered. This was probably partially my fault and partially just bad luck. The journalists and editors of some of the major blogs and magazines are no doubt inundated with messages from people such as myself, only having time to open and read a very small selection of the most interesting emails. I will need to revise the  content of my emails if I am to stand a chance of getting through to them, and to be more persistent in my approach. However, at the end of the day, a certain amount will come down to luck.
    What is interesting is that you can see that where Globe did feature, backers followed. For example there are 10 backers from Germany, double that of any similarly sized Western European countries so it just goes to show if you can get featured, it really helps.
  3. The campaign reward structure was poor. I was torn between whether to offer early bird rewards in the first place but eventually decided to include them. The danger with early bird rewards is that they are, in some ways, unfair to later backers. By offering discounted prices to the early backers, you are essentially pushing the cost of their discount onto those who come later. While it is nice to try and encourage backers to get the ball rolling early on, the negative sentiment it breeds is probably a bigger concern. For the next campaign there will be no early bird pricing. I have however modified the reward slightly to allow tier to be sold at close to the original early bird 1 price. By switching from a glass case to an acrylic one and offering only oak veneered MDF (the same as the prototype) I can save enough money to allow for a lower price.
    Ideally I could offer a lower price target, something to sit between £1 and £550. Unfortunately I don’t have a design to put on a tshirt (which resulted in virtually no one backing it) and I certainly can’t make Globe for any cheaper. I will however have a bunch of unused PCB panel space (since Globe is round) so I could potentially offer something made out of that.
  4. I set my funding target too high and tried to include too much into the project scope. For those who have read through my blog, you have heard me express interest in exploring certain designs, such as a flex PCB strip design rather than a circular disk & inductive slip rings. It was my intention to use some of the funding to explore these things but in hindsight it would have been better not to for a couple of reasons. Firstly, most people didn’t know about this extra stuff I wanted to do, having not read the blog and me being careful not to promise it in the Kickstarter page (in case the designs turned out to not be possible). Therefore the larger budget may have seemed unjustified. Secondly these things unnecessarily extended the project timeline which was likely off putting to potential backers. In reality Globe in it’s current form is very close to being a ready to ship device, and it’s that which people were backing. By removing this extra development from the project I can bring the funding target down and bring the timeline up giving the project a better chance of reaching its target.
    It was also my original intention to secure dedicated premises in which to work, giving me slightly more flexibility over my work space. In reality, procuring dedicated premises should not factor into the minimum target of a Kickstarter campaign, instead being something that can be done if the campaign does especially well. I have recently been to look round a great maker space in London, which, if the new campaign goes ahead, will be where I do my final prototyping and assembly.
  5. People tend to back projects that look like they are going to succeed. This is partly in relation to point 4 and partially just a general observation. I saw throughout my campaign some of my backers cancelled their pledge as they saw that, based on the time elapsed and funding reached, the project was unlikely to reach its target. While, in my opinion, cancelling a pledge is a bit of a shitty thing to do, I can understand that people don’t want £550+ tied up in a project which is probably going to fail, when there might be other awesome projects they want to back with that money instead. If they were actually cancelling their pledges, how many people decided not to back based on the same reasoning? It becomes somewhat of a vicious cycle, people don’t back because it doesn’t look like the project will succeed…because people don’t back. Hopefully most of my backers from the last campaign will back again, meeting the funding goal very swiftly. This means that for future backers, who know that the project is successful, will be more inclined to back, however I am certainly not relying on this. The project will be possible even if the new target is only barely met.
  6. My presentation in general was a bit crap. Unfortunately there isn’t much I can do about this one. At the end of the day I’m an engineer, not a graphic designer or any such person with artistic talent. Because I have no money (and at this point I mean literally no money), I am unable to bring in photographers who actually know what they are doing, people who can do pretty graphics or slick videos like you see in other Kickstarter campaigns. This means that all my media doesn’t look as good as it could have and my overall presentation of Globe probably isn’t as direct or informative as it could be. I have tried to tidy up my campaign page, get rid of some of the unnecessary tech waffle and make the whole thing easier for people to understand, but that’s all I can do.

That pretty much sums up what I think went wrong and what I am going to try to do to make the next campaign succeed. If anyone has any other input I am always happy to hear it.

The new project should be going live at 12:00 GMT on Friday (06/10/2017) assuming Kickstarter approves it any time soon.

Still waiting for approval so it looks like it will be pushed either into the weekend or early next week.

21: And we are live on Kickstarter!

Kickstarter Link

Video Link

I have done it, I have spherical  pushed the button. Now all I can do is wait and hope. To my first two backers I want to say a huge thank you for taking a leap and getting the ball rolling.

I finally decided to do away with the different flavors of Globe and instead keep it to a single lower cost device. Anyone who wants any sort of bespoke artistic version can always contact me separately and place an order that way, but otherwise it keeps things nice and straightforward for this first project.

Getting all the media has been a massive pain, mostly because I don’t know what i’m doing with cameras and I have limited access to facilities in which to work. Unfortunately when I just about had the code working perfectly and had switched over to using a proper video camera, the HDMI through the slip rings decided to die. Well, not die as such, more like go into spasm, whereby frames were just starting at random positions. I guess that as they wore just a little bit it was enough of an impedance switch to throw the data off. To be honest I was slightly surprised it worked so well in the first place.

So I was in the middle of setting up the Pi Zero W module and trying to sort out the kinks with that when I realised that it was simply mad to delay the projects release any longer, already being about 2 months late, so here we are. If I get any additional good media I will update the KS page, but if not I think what I have already is good enough to at least illustrate the concept.

If anyone has any feedback on the KS page please do drop me a message on here or there. I am new to this and much better at engineering than marketing.

 

DSC08357

20: Nearing the end

So last Friday wasn’t the day…nor was any day this week, however the end is in sight. It’s been a trying week, soldering HDMI cables onto connectors…again…and again inside of a metal box, chasing down potential shorts and generally getting annoyed with some of the design choices I made long ago.

So far I have tested:

  • Motor and power supply. Not a huge achievement here since it was just a case of wiring the things up and flipping a switch but its always good to know I can hook up mains electricity and not kill anyone.
  • Motor driving the shaft with PCB in place. All went well, nothing flew off and I had the board spinning up to 15hz.
  • HDMI through the slip rings. This certainly seems to work, I have so far tested the standard single colours without any parts dropping/distortion and some basic images (more on that in a second).

So everything is spinning nicely and it is mostly lighting up. There are a few issues still to fix:

  • I’ve yet to implement any sort of techometer/rev counter. At the moment it’s a bit of a mess inside the box, with certain components positions not finalised. It occurred to me if using a hall effect sensor and magnet then the magnet needs to be positioned carefully so that it isn’t spinning past any wires because rapidly moving magnets past wires = electronic mayhem. I am probably going to switch over to using IR rather than magnetic rotation sensing to avoid any problems, but for the time being I can just match the shaft speed to frame rate for ~static images.
  • Still need to calibrate the colours. I’m not entirely sure how i’m going to do this, i’m not sure those standard colour calibration tests will translate over to Globe that well.
  • Need to finish tidying up inside the box, make sure everything is strapped down and then put the sides on. To hide all the badness.
  • There is a slight issue with the offset pixels being offset by about 1cm. This is the result of vblank length and the way the code is written. Dropping the length of vblank down will mitigate the problem, switching from the first coding implementation to the third will eliminate it.

So next week is going to be an exciting one, if all goes well and the HDMI is being difficult I will be able to get some of the first proper pictures early on, and then I shall just need to spend the rest tweaking the gamma values for the RGB channels.

DSC05904
A quick test image. Colours are still wildly off and I need fix a couple of LED channels
2017-08-08 14.19.45
Motor temporarily off to put some rubber washers that David sorted out for me.

19: All coming together

I finally have all the pieces of the puzzle; the shaft is anodised, the laser cut steel plates are in and painted and the pulleys have finally turned up. My aim is to have the Globe prototype completed by the end of this week and the Kickstarter live early to mid next week so I have a huge amount of assembling and cabling to get through. The guys in the workshop have heard the words “today’s the day” so many times now that it’s lost all weight and if my PhD supervisor wasn’t the most relaxed Greek in the world then I would be in some serious trouble.

I am about 75% of the way there; the individual panels are assembled and just need bolting to the frame, the shaft needs inserting through the bearings, pulley and slip rings, the HDMI and USB connectors putting on everything and a final tightening of all the nuts and bolts. It should be possible to get all of that done in a day so I am tentatively going to say…today’s the day.

In other news; my company is now registered with Companies House which is cool and i’m in the process of applying for a business bank account. I don’t want a large amount of money flowing through my personal bank account and you need to have your bank details sorted before your Kickstarter goes live. Depending on how long it takes HSBC to scrutinise my application and say “but you’re a broke PhD student, what could you possibly need a business bank account with, at the point of application, virtually zero investment for?”. There wasn’t a tickbox for “I’m opening this account so I can proceed with my KS campaign and hopefully I will reach my funding target, but I could also far exceed it but it’s hard to tell…oh and it could also be a total flop”.

So that’s about it for now. As soon as I have the prototype working, some pictures and videos of it working I will make this blog accessible.

2017-07-25 13.33.16
Black anodised shaft
2017-07-27 09.50.28
Frame with top and bottom plates on
2017-07-27 09.54.24
Read plate from outside
2017-07-27 09.55.03
Rear plate inside, just need the ribbon cable from the VFD to control panel

 

18: Kickstarter work plan

So depending on how the Kickstarter campaign is going/has gone this blog may be more or less redundant, but I am writing it on the assumption that the campaign has reached its target and the project is going ahead (after all one needs a solid project plan before asking people to provide funding).

As an electronic engineer I can handle a sizable amount of the design work myself, but certain parts will need to be contracted out. If the project significantly exceeds its funding target the development timeline should shift up 2 – 3 months as more work can be contracted out.

The project can be partitioned into 4 main sections:

  • Mechanical: This includes the mechanical housing of the project, the central shaft design and the PCB mounting ring.
  • Software: This encompases the Android & IOS apps, the software that runs on Raspberry Pi and the web interface.
  • VHDL and Firmware: The VHDL is more or less all in place, mostly just needing tidying up and a few tweaks making here and there. The firmware for monitoring and protection will need to be written from scratch.
  • Electronics: The visible electronics in the display modules, the Raspberry Pi and FPGA control circuitry and the power transfer electronics in the slip rings.

 

Mechanical

Work description

The mechanical work that needs to be undertaken is not trivial. It can be viewed as three main components: the main box housing, the central shaft and the PCB mounting ring.

Central shaft

In the prorotype the shaft was made out of a machined piece of Aluminium, it was fairly heavy, expensive and over-engineered.  The shaft needs to switch over to an easily produced plastic design. The shaft will also need redesigning as it will no longer be housing a flat PCB but instead accommodating the PCB mounting ring.

PCB mounting ring

The PCB mounting ring will be a hoop, either made of metal or plastic upon which the flex PCB is mounted. It is important that the hoop is strong and accurate in its construction, if the PCB mounting holes are out by half a millimeter the result will be that when the disk spins rather than interlacing you will suffer from superposition. It will also need to be strong enough so that when spinning it does not deform or warp.

Main box housing

The box housing for the prototype was made out of steel box section, with steel plates for the bottom, top and rear and veneered panels on the other sides. While not overly expensive this method did result in a woefully heavy device; this needs to be cut down by at least 50%, ideally 75%. The design can either move over to aluminium, or simply us thinner pieces of steel.

Work plan

The majority of the design work this side of the project will be completed externally. While I will provide the specification for the various mechanical parts it really needs completing by someone who knows exactly how to design something to be both economical and reliable. The aim will be to have the design work commence in the fourth quarter of 2017, once the general specification of the wireless power modules and PCB are complete.

Software

Work description

Globe requires being as accessible to the end user as possible. While almost all of you who have made it to this blog are tech savvy and completely comfortable with the task of remotely connecting to a Raspberry Pi and putting whatever content you want up, a lot of those who haven’t come over here probably aren’t, and that’s okay we will have them covered too! Globe is aimed at more than just us nerds who want a really neat toy.

  • Firstly we need software to run on the Raspberry Pi; a simple local server to interface with the web & smartphone apps.
  • Secondly we will need the web app to allow users to control the content being displayed remotely.
  • Finally we need the Android & IOS (and maybe Windows if anyone actually wants it) apps to allow content to be controlled directly from ones phone as an alternative to the web interface.

Work plan

This work can be started almost immediately after the KS campaign ends. The requirements are unlikely to change and Globe is, at the end of the day, a big old display connected via HDMI to a Raspberry Pi. The developer doesn’t need to do any fancy interfacing or know what happens after the Raspberry Pi. While it’s the sort of thing I could cobble together myself, it would be a lot more professional and refined if completed by someone who really knows their way round app development. Therefore I shall find a reputable looking freelancer who is up to the job.

My aim will be to have initial prototypes ready by the new year with a round of updates tweaks and changes by the end of the first quarter and another round by the end of the second if required. I have also budgeted for a post launch update of the software to accommodate any feedback from backers.

Firmware

Work description

The FPGA already accepts a 1080p signal, it simply drops every 4 lines (and then a bit) to divide it down to 270×1920. As shown in one of my earlier posts it is capable of handling the 1080p data rate so only a minor alteration needs making there. Mostly this code just needs to be tidied up, optimised and thoroughly tested to ensure there are no bugs left in it before it ships. While I will do all of this myself, I may liaise with someone who is more familiar with publishing professional VHDL to ensure it is as readable as possible for any future development.

Work plan

The work on the VHDL doesn’t need to be completed until the prototype hardware is ready for testing. Since it doesn’t need a huge amount more work doing to it it will exist as a background task to be completed throughout the project. I will however carve out a couple of weeks for it at the beginning of April just in case some major unexpected work does need to be done.

Electronics

Work description

The electronics can be viewed as two separate sub projects. The first is the display electronics: the flex PCB upon which sits all the LEDs and drivers, the FPGA HDMI interface and the Pi compute module. The second is the wireless slip rings: the coils, the resonant converters and power regulator modules.

Display circuitry

The display circuitry will, in some ways, become slightly easier to implement than in the prototype. Because we are switching from a spherical disk PCB to a curved flex PCB the placement coordinates return from polar back to cartesian, all the LEDs will have the same X as will all the other components and their Y locations will change incrementaly. The FPGA and HDMI interface overall design is complete so the only completely new design work will be for the compute module.

There will need to be two different versions of the display PCBs; the 540 LED version and the 1080 LED version, but the other electronics can remain the same across all models.

Slip ring electronics

The slip ring resonator electronics will be adapted from a Wurth Elektronic reference design. Theoretically not a huge amount will need to change, it will just need to be thoroughly tested to ensure that the power transfer is stable while rotating and the cooling is sufficient at high power. In addition to the slip rings I will also require an equally rated voltage regulator to drop the receiver coils 24V output down to 5/3.3V without loosing too much energy along the way. Theoretically it would be possible to just configure the resonator circuitry to work at the lower voltages BUT the coils themselves can only handle so much current before saturating. I may be able to use different coils (or even custom ones) depending on whether we exceed our target but for now I am planning on sticking with 24V and dropping it down with a regulator.

Work plan

I will handle the design of all of the electronic circuitry, bringing in an additional team member if funding far outstrips the target to help speed up the production schedule.

The most pressing task will be to start prototyping the slip rings. A reference design unit will be aquired as soon as possible after funding to allow me to test the power transfer quality when spinning at 900RPM and to evaluate what sort of aditional cooling will be required to reach the higher power limits. The aim will be to have a prototype of the slip rings completed by the end of November to provide plenty of time for thorough testing and modification if need be.

The second most pressing task is the Raspberry Pi compute module. Theoretically this shouldn’t be an overly complex task, there are plenty of reference designs out there and I am only using a fraction of its capabilities. I will start of by building a prototype PCB with just the bare essentials: an SODIMM slot for the pi to sit in, a wireless chip, voltage regulators and an HDMI jack. Once this is completed I can move onto designing the single Pi – FPGA PCB. I will start work on this straight after funding and aim to have it completed by the end of October.

The third task will be the LED display rings. These should be a straightforward task, because the electronics are mostly the same as the prototype, simply scaled up for more LEDs, but I will need to liaise with my PCB fabrication house to make sure that everything I design meets their capabilities. This should only take about 4 weeks to complete so I aim to start early December and have them completed early January.

The fourth task will be the Pi – FPGA PCB. This again should be a fairly straightforward task, essentially taking the logic from the Pi prototype board and the FPGA design from Globe’s prototype and combining them onto one board. This should only take a couple of weeks of work and I will start this straight after the first LED display ring is complete, which should have both of them ready for fabrication before Chinese New Year.

The final bit of electronics required will be the motor and power controller. This small circuit will sit down in the housing and will be responsible for monitoring temperature, inertia, tilt and other factors to ensure that Globe is only being run when it is safe to do so. This will only take a couple of weeks to design and will commence once I have decided on the design of the mechanical housing.

As the individual electronics come in they can be tested, modifications made and new prototypes created as need be. This should mean that by the end of March there should be all the components ready for testing as part of a full unit.

 

17: Gearing up for Kickstarter

As of writing this the prototype is nearing completion and I am gearing up for Kickstarter. It’s been a difficult project to plan, partially due to Globe being a fairly complex project (as far as Kickstarters go) and also because there’s still a fair bit of development to be done. As with most things, it is an expensive thing to produce in small numbers and becomes a lot cheaper as volume increases, but either way will require a higher than usual funding target for a Kickstarter.

I have been really torn as to how to market Globe; whether to offer a single advanced device which would be more straightforward to manage but would mean that those who aren’t interested in some of the extra features (more housing options etc.) would end up paying more than they wanted, or whether to offer multiple different tiers allowing backers to choose exactly what they wanted. Based off other similar Kickstarter projects splitting into multiple tiers is the way to go.

The three tiers that I propose are:

  • Desktop – 1920×1080 resolution, desktop, veneered oak cladding, laminated glass casing. Essentially the same as the prototype. Aimed at those who are not going to have the device running constantly on public display ~$800
  • Professional – 1920×1080 resolution, full height,  range of cladding options available, pedestal for unit, toughened glass casing. For those who want the unit to sit on display all day, ideal for advertising purposes. ~$1400
  • Artistic – 1920×1080 resolution, full height, customised solid wood/premium metal cladding, pedestal for unit, acoustic glass and sound insulation, customised content. Ideal for those who want to use the device as an artistic piece in locations where ambient noise is minimal. $3000

The main difficulty with doing things this way is that the prototype device is roughly the entry level version with half resolution. It may be difficult for people to visualise some of the features on the higher tier models and decide whether that’s something they are interested in.

The next problem that needed to be overcome was shipping the damn things. Due to being fairly heavy devices, the entry level likely to come in at about 10KG, the pro ~20 and the artistic ~30-40, shipping will not be a simple issue. The simplest and most cost effective option will be to farm it all out to a fulfillment service. Which one(s) I use will depend a lot on how much funding the project receives and where backers are based, so I will wait until after the campaign to make a firm choice on them, but I have tallied up the rough costs of getting things distributed. At the moment the list of countries I will ship to is fairly limited, mostly EU/north America, but if there is popular demand for another country I will certainly look into finding a shipping solution.

The only other consideration are all the various import duties and taxes that would or could be applicable. Because my business will be UK based there will be no import duties for EU orders, but all other other orders will be required to pay whichever import taxes and duties that their country charges.

So that pretty much sums up the last couple of weeks for me, madly trying to plan around all the different logistical realities of making a big old spinney whirley wheel and distributing it around the world. Ideally I will have the Kickstarter go live within a couple of weeks of writing this.

 

16: Mechanical housing

The mechanical housing has been kept as rugged and straight forward as possible; a simple frame 425mm wide, 400mm high made out of inch square mild steel to be nice and sturdy and leave plenty of room for fitting things in.

2017-03-24 16.11.26

The top, front and side panels are laser cut out of 6mm oak veneered MDF, stained in “Terra” which personally I think looks quite good. Depending on funding and opinion I will endeavour to offer a range of stains and possibly some other basic material such as anodised aluminium. Thoughts?

The rear and bottom panel are laser cut out of 3mm mild steel, since all the heavy stuff (motor, driver, PSU) will be mounted on the rear panel and the feet that will be supporting the entire display will be mounted on the bottom. This means that there will only be 3 “attractive” sides on the display, the rear will have a bloody great heatsink hanging out of it, along with the motor control panel and HDMI, power and USB cables.

If the Kickstarter gets funded then the finalised product will look lovely from all sides (kind of the point of having a spherical display, right?). The motor driver will be significantly smaller with no external heatsink required, no interactive motor control interface and no HDMI or USB cables. All that would remain would be a power cable which would either be a standard kettle C13 plug or a hard wired cable.

For simplicity and sturdiness a bearing holds the shaft at the bottom and top panels and the shaft extends about 250mm up the PCB. Therefore with a little bit of space between the bottom of the PCB (for cables to go in) and the housing we have a total length of ~720mm. The bearings and slip rings have a bore of 38.1mm so we have a 720mm long, 38.1mm diameter shaft, which needs a slot of ~290mm milling down from the top and a hole of 14 data and 4 power cables worth up from the bottom. Along with some other slots and holes that’s pretty much all there is to it. Unfortunately due to its size its a fairly pricey piece, even by buying in a 12.7mm thick walled tube to save on having to drill the central hole it’s still looking like a £250 job (and that’s getting it done in house).

Aside from a pair of pulleys, a drive belt and some mild steel right angle sections on the corners of the box keep things neat that pretty much sums up the lower section.

The upper section is comprised of the previously mentioned acrylic cube which should provide an attractive enough protective casing to curb any desires to see how much it hurts if you poke a finger in while it spins. It also means that if one of the components makes a bid for freedom it should be suitably contained. This is set within a square section of aluminium U channel with a rubber insert, a couple of grub screws providing a little bit of force to hold it…in case some of those who are REALLY determined to poke a finger in decided to remove the casing too.

If the Kickstarter goes ahead I will more than likely replace the acrylic with some form of toughened glass which would become a more viable option if produced in bulk. I would also bump up the safety a bit more, adding a sturdier locking mechanism and various sensors to make sure the device was being operated in a safe manner.

 

15: Cables – to DIY or splice?

The sliprings I bought for this project have 4 power lines (10A each) and 14 data lines. HDMI cables consist of 4 shielded differential pair signal traces (3 data, 1 clock), I2C (clk and data), 5V, GND, CEC, hotplug detect and in some cases ethernet, along with a bunch of shielding. Of these I am actually only interested in the differential pairs, I2C, hotplug and shielding. This means that the total number of the data lines of my slip rings that will be required for HDMI will be 11.

Standard USB cables consist of a differential pair for communication, 5V and GND. Of these I am only interested in the differential pairs and GND, taking up a further 3 of my 14.

DIY

The differential pairs must be twisted together (so that any electromagnetic interference is equally applied to both and cancelled out on the receiving end), wrapped in foil with a drain wire to provide shielding between them and other signals.

The I2C wires and the HPD can be left exposed but then all the HDMI cables must be encapsulated in braided copper, similarly for the USB to try to eliminate as much EMI as possible. A well made HDMI cable can reach lengths of 30 meters (possibly even more), mine only has to manage about 50cm, so fingers crossed even my bodged together cables work. I am still a little unsure how well the data is going to make it across the slip rings…only testing will tell. In the worst case scenario I always have space on the PCB to mount a Pi and run cables from this over.

One thing I did find surprisingly difficult was getting hold of HDMI connector shells. While I was able to find a mini USB shell fairly easily, and I was able to find a solderable HDMI male connector, finding a shell to encapsulate it was next to impossible (which is fair enough really, why the hell would anyone want to DIY an HDMI cable?). In the end I found one on Aliexpress, but it is made of metal rather than plastic and fairly bulky; probably not the best thing to be hanging off the bottom of a board spinning at 12 rps, so I am undecided whether to use it or not.

 Splice

The alternative to making my own shielded cables is to buy pre-made cables, cut them, expose the internal wires and splice them onto the cables already present on the slip ring assembly. This would mean that the cables will be thinner, with shielding already in place so less concerns about signal integrity. I am slightly reluctant to try this method because it requires cutting the slip ring cables, if this method then doesn’t work it rules out attempting the DIY method.