Journal: Sculpt

* To view all journals, click here *

Journal entry: 10 Nov 2017

IMAGE: Using a drone for altitude testing of the barometer in one of our prototypes

Mac Jones: In these last two weeks, we've been finalising the technology that we want to demo to the judges. Our system is GPS-based, which means that it may not work well indoors, so we’ve made some software changes to cope with that.

Hardware-wise, we’ve had a few things to wrestle with. As we've been bringing more devices online, we’ve had a couple of software bugs that pass old data around our mesh network, giving false location readings, but we think we have a solution. We also managed to blow up a battery and a GPS unit by accidentally shorting stuff out – you need to be really careful when packing lots of components into a small space! In better news, we’re very excited that our latest software changes have increased the unit’s radio range to far beyond what we’d managed in our initial tests. 

The final pitch is coming along well – we’re onto its 4th iteration, and are now focused on refining it down to fit into the allotted 20 minutes time slot.

----------------

Journal entry: 27 Oct 2017

Image: Rethinking the case design

Mac Jones: This week, we've been working on version 7 of our case. After some input at the recent (excellent) Bootcamp, we went back to the drawing board on its design. We’d like to give a particular shoutout to Jesse, who really broadened our thinking on that front. He made us realise that, with some work, it could be made smaller and more wearable for different scenarios. So, we binned a bunch of old work and started again, now we're really happy with the results.

We've just assembled all our circuitry in the new case, but now our compass issue is back – yet again, it has very weird ideas on where north is! We're currently trying to work out where the interference is coming from, so that we can mitigate or calibrate against it.

We've learned that hardware is really hard – we're already on our third type of microprocessor, second type of GPS and third type of screen. But, now we're really happy with the combination. And we’ve starting to work on our presentation - judging day is rapidly approaching!

----------------

Journal entry: 12 Oct 2017

Image: Emergency repair kit

Mac Jones: This week, Team Sculpt has been focussed on our IP strategy, thanks to the awesome team at Ellis Terry. They've really opened our eyes to the way IP can be leveraged from both defensive and offensive positions. It turns out that IP strategy is really complex, so it's great to have experts to guide the way.

On the technical side of the project, we're still coding software till late at night, squashing software bugs as they appear. We're now focussing on the user interface – our goal is to make it so intuitive that our users don't need a manual. We’re learning though that it's really easy to make software hard to use (just add lots of menus and buttons) …and really hard to make it easy-to-use! 

----------------

Journal entry: 1 Oct 2017

IMAGE: So many bits, so little time.......

Mac Jones: These past two weeks have seen us wrestle with our digital compass. We’d been getting strange results from it, which we needed to fix, but in order to do that, we first had to understand exactly what was going on.

It turns out that the Earth’s geomagnetic field can fluctuate between locations, even within the same country. This declination can make an ordinary compass read a different "north" in the South Island than in the North Island, and it’s largely due to magnetic variations in rocks kilometres underground. These fluctuations have an impact any time you’re trying to determine a location, like we do in our system, so we need to calibrate against them in order to get accurate results.

The great news is that we’ve managed to write some code that automatically performs a calibration step immediately after power-up. And importantly, it then saves those results in permanent storage, which means that it does not have to be done every time you want to use the system. Where last week our compass would happily jump around by 45 degrees, this week it's much better behaved.

Elsewhere in the team, we’ve had a key member sell his home unexpectedly quickly. So, he's compressing his development time around packing boxes and finding a new home!

----------------

Journal entry: 20 Sept 2017

IMAGE: Yellow is the new black!

Mac Jones: We have been busy finalising the hardware components we will use, so that we can finish the design of our enclosure. There are so many hardware options, each with their own cost-benefit implications. We're carefully analysing these "costs", which are not so much monetary costs, but rather costs in terms of power and size. Some modules that may have lower energy requirements or higher sensitivity come in a larger form factor, which would make our device larger than we'd like. On the software side, we are now trying to develop the user interface so that it display’s useful, visual information (rather than pure test data).

Our new LCD screen is looking great, and we're working on different display options, focussing on how to display a lot of data in succinct and useful ways. We’ve 3D printed a test enclosure, which means we can physically hold it - that really makes it simpler to iterate the design.

It’s not all been plain sailing though. The new GPS module that we ordered from China should have included a compass… but it turned out NOT to. Also our 3D printer's heating system failed and required a quick MacGyver fix! But we've been enjoying experimenting with different plastics in the printer – each type of plastic varies in terms of durability, environmental impact, ease of printing etc. It's been a real learning curve to determine the best plastics for the job.

----------------

Journal entry: 5 Sept 2017

Image: 3D printer in action….and when things go wrong…

Mac Jones: For the past two weeks, we’ve been working on the software communication protocols – namely, how to get all the devices talking to each other. We have found a reliable way to share data across the mesh network, and so far, we’ve tested this with four devices. We're now working on some new antenna designs and repeater software to optimise range.

We’re also trialling a few different combinations of CPU board, screen, and GPS, in order to find the optimal platform on which to build. We've had to go back to the drawing board on our screen technology because we’ve found that while OLED screens are nice and bright indoors, they’re useless in bright sunlight! Colour TFT screens perform better outdoors and will give us more options… but we still need to test battery life.

We're also spending a lot of time reaching out to potential customers and meeting them to better understand their needs. Time is flying by – if only we had another few months and maybe no ‘real jobs’ to worry about!

----------

Journal entry: 22 Aug 2017

Image: Programming up our GPS modules so that they can listen to Russian and Chinese positioning satellites, alongside the American GPS constellation

Mac Jones: We are developing a wearable location beacon that provides enhanced situational awareness in remote environments. In this past two weeks, we’ve been focused on two primary areas – 1) talking to actual customers to confirm that our assumptions on the problem we're solving match their needs, and 2) looking at the technical challenges around transmitting and receiving radio signals.

We purchased a second-hand 3D printer last week and have managed to get it working. The next stage is to design the first iteration of a wearable case that can hold our device electronics and battery pack. An important lesson we learned at Bootcamp is that, despite being easy to get into, 3D is a discipline that takes years to get really good at!

Up to now, we've been using open-source mesh radio software, which, because it's already 'packaged', comes with the benefit of saving time. But we’ve recently had a few setbacks with getting it to do what we want – it’s a balancing act! So, we’re currently considering whether we should write our own mesh radio subsystem instead.