I spoke to John McKerrell, who is another of the Very Interesting People that I share a workspace with at DoES Liverpool. He's created the WhereDial, a location based Internet of Things device, which we've already mentioned in this blog. It's a very cute device, but I'll let John himself introduce it in more detail! By the way, the WhereDial and Bubblino will be at the OpenIoT assembly this weekend (Jun 16-17th 2012) in London, so come and say hello! (Tickets still available)
Hakim: So, for the book, we want to talk about different products, case studies, how they're done, what kind of services they connect to, various things like that, so there's nothing specific on topic. But, I guess the first question would really be "What is the WhereDial?"
John: The WhereDial is a device to reassure your loved ones and your family and your friends. It's an Internet of Things ambient device that sits on a mantelpiece or a bookshelf and tells your loved ones where in the world you are. It will tell them if you're in a school, at work, at a station, or in hospital. And it can even tell you if you're in mortal peril!
Hakim: Well that last thing especially is very clever! But we'll come back to it later. First though, how did you come up with the idea?
John: The inspiration was the Harry Potter books. The Weasley family have such a device, essentially. They have a grandfather clock which has a separate hand for every single member of the family. The hand will point at a description of a place listed around the face of the clock. So it lists school, work, station, all of those places, and of course mortal peril. And it will magically (because we have magic in the Harry Potter universe) update where each family member is. So a friend of mine, Grant Bell had the idea that I should make one of these using my existing location tracking website, mapme.at, that I've been building for a few years and has been a side project of mine, and lets me track my location using many different techniques. And so the idea was I could use that to input my location and then use the location clock inspired from the Harry Potter books to show people where I was.
Hakim: Because the Harry Potter clock is obviously magical and, we don't have magic in our universe... but people have said that Internet of Things devices are like "enchanted objects", so in a way they do replicate magic using technology. So how does the website, mapme.at, how do you determine whether someone is at school or in mortal peril?
John: Often when you check into mapme.at... it has a database of places taken from the OpenStreetMap database, which means that it has access to millions and millions of different place entries, points of interest. So we will have in the database an entry for various stations, for all the different types of places you will be. So when you check in and say "I am here", you will say that you are at a station for example. You can specify that some of these are favourite places. So if your office that you go to regularly, stations that you visit often, they can be marked as favourite places, and then if you're using a more basic method to track your location that only says your latitude and longitude, it can snap your location to the nearest favourite. So we can do it automatically as well.
Hakim: so is there a config stage, where you have to tell the WhereDial what places are?
John: Yes, so the WhereDial is currently in development and this aspect of it isn't completed. But there will be a website that you access. Some of the things will be automatic, so "Hospital" is categorized in our database, whichever hospital it is. Others like "Work" we won't necessarily know where your work is, so it will be fully configurable in that, if you have a particular place - say if you work in a school, you could specify it as being work. Alternatively, you could say that "anything that's an office means that I'm at work." For example if you travel a lot.
Again for home, you can specify exactly where your home is. Also with the privacy settings in mapme.at, you can keep your home as a "private location" but have the WhereDial know that you are at home. The person who's looking at it doesn't necessarily have to know where you live, just that you are home, and therefore hopefully safe.
Hakim: And how do you go about "checking in" to mapme.at
John: There are meany different ways. There is a web interface you can use. A standard one with a map, and what places are there. You can search for locations, find the right one, and check in that way. There's also a mobile website which can look at your mobile device's location and show you places that are nearby to select from. You can also use other services that do similar things, so foursquare, google places, facebook - actually facebook's not live at the moment, but will be for the WhereDial. You connect those existing accounts to mapme.at, and we can pull your location from there. So if you're already using something to track your location, you can continue to use that, and mapme.at will just suck your location in from there.
Hakim: So in a way if was quite convenient that you happened to have a location based API lying around that you could use, that you could target.
John: (laughs) it was. But of course, it certainly takes away half the... well not necessarily take away, but the two sides of the problem are putting your location in, and getting it out. So it does help that I already had a way of handling half of the problem. I guess if I wasn't doing that and wasn't interested in location, I might not have had the idea, or the suggestion to do the clock in the first place.
Hakim: But for example, if you were interested in location, but hadn't built mapme.at, would you have been able to use foursquare or another service to do the same kind of thing.
John: Quite possibly, yeah. It would have been harder to make it as configurable, so for example to say where your home is, and the privacy settings that everyone can see that or not. But yeah, it would be possible to do something similar.
Hakim: So there is some advantage to having control of the API as well as the device?
John: Yeah, and of course it allows you to... I would always have needed to do some sort of configuration... have some sort of website managing the owners of the WhereDial. So it integrates really nicely with the existing website. And means that you don't have to be using foursquare, or google. You could use a number of iPhone clients you could use instead. And if you don't want to use any of that, you could just use the website.
Hakim: So, whereas if it was connected to foursquare, it would want to be foursquare and that's it...
John: This way it's completely agnostic. You can use whatever you want to and it will work.
Hakim: I had a quick scan of your website, and noticed you'd posted about the API. So you have a public API that other applications can use.
John: Actually, for the WhereDial, for the prototype versions at the moment, they do use the standard API. For the WhereDial I'm looking to customize it slightly, just to make it simpler and allow specific privacy settings... although this probably will go back into the main API too. But for instance at the moment you can specify your privacy for groups of people, and for API clients. But generally, it's at a level of showing your exact location, or what town you're in. With this, what I'm planning is that a privacy setting will be "let people see the type of place I'm in and nothing more". So you might be quite permissive about what town you're in, which doesn't say whether you're in a hospital or work, or whatever. But you might be happy to say that you're "in work" but not the location. And for the WhereDial it doesn't care what the location is.
Hakim: It's a tag.
John: Yeah. So the intention is to do a new API that will also work with the configuration so where you set that if I'm at 23 Acacia Avenue, then I'm at home. The WhereDial can just connect and the API will say "at home" and it's as simple as that.
Part of that comes from trying to keep things as simple as possible for the WhereDial itself. Because the computing power in the WhereDial is pretty limited. It can parse full API response. But it's nice if it doesn't have to.
Hakim: It would be quite interesting to talk about how it's built and what's in it. But first of all, the WhereDial isn't the first version of the clock that you've built is it?
John: No. The very first was an actual clock. It was based on a clock that was donated to me by my dad. I don't think it was functioning... but it had been a functioning clock with lots of gears and big powerful springs and all sorts of things. Which I took apart and essentially bolted a large motor onto the back of it which fortunately meshed well enough with the existing mechanism that turning the motor would turn the hands on the clock and give the effect I was looking for.
So I had that, and I put an Arduino board with an Ethernet shield on. Made my own small controller shield that would control the motors. So there was a stepper motor on the back which allowed me to control quite accurately the position of the motor. Controlled by the Arduino with Ethernet shield connecting to the internet.
Hakim: So being a clock that's 2 hands...
John: And yes, 2 hands, still connected in the standard 60:1 ratio. Which on the one hand makes it take longer to update the clock than it necessarily needs to. Because potentially you might need to do a full rotation of the hour hand, which needs 12 rotations of the minute hand. But it's also actually quite a nice quirk, because if the hands would just instantly move and not take very long, you might not notice that something's changed, and then hours later realise that someone's left work and is on their way.
Hakim: (laughs) so I remember that when I first saw that clock in action, we were in our first office, the original Room of Things, and occasionally Bubblino would go off and make noise blowing bubbles, and then sometimes the clock would go off, and at the time the hands were... they were tracking the Mersey shipping originally? So how did you build that?
John: Well, originally it was tracking my location, and my wife's location. Which is great, and is very useful. But it's not so good for demonstrating. Because... you mentioned Bubblino there, which is great to take to a conference, put him on the stage, and he blows bubbles all through the conference, as people tweet about it.
The clock, if I took it to the conference, and even if I was using it at the time for a presentation... well I'm at the conference, so I'm not moving. My wife is at work, so she's probably not moving. So you have this clock that is stood there not doing anything because I'm not moving for the whole day.
So Adrian suggested ... he took it to MakerFaire in Newcastle actually, 2 or 3 years ago. And he suggested we show the ferries instead. Because the ferry... on the Mersey, makes 3 stops every hour and in between that sails around the Mersey. So we updated it to do that, which was relatively simple. It was changing.. on the clock side it was just changing the users it was looking for on mapme.at, and changing the favourites it was looking for. So instead of "home, work" it was looking for "Seacombe" and "Liverpool" and also a special case for if it's not at one of the favourites, it knows that it's in open water.
Hakim: Rather than mortal peril?
John: Rather than mortal peril... that's when it goes to Manchester in the ship canal.
Hakim: So are there 2 ferries actually?
John: There are 3 ferries, but only 1 is ever active. Which is a problem for that clock as it had 2 hands... so it could show 2 ferries, but if they were not the ones that were active it wouldn't do anything. And even if it was one of them, if it was on the hour hand it would take longer, and not look so nice.
But it also had "moored" which is where they'd generally be when they're not active. So you'd generally have one that was moored, and the other that was bouncing around the stops.
Hakim: So how did you get that data into mapme.at?
John: That comes from the AIS system...
Hakim: AI for ships?
John: Yes... I can't think what it stands for... I imagine S stands for ship or shipping. "I" could be intelligence. Basically, every single vessel of a certain size on the sea and larger rivers has a transmitter that transmits its latitude and longitude on a regular basis. And then there are hobbyists and interested persons around the world who listen for this data and submit it to websites, such as ShipAIS.com.
So actually Adrian again set up a script that would pull that location for the 3 ferries off ShipAIS.com and put it into mapme.at. So from that moment on we then had 3 users on mapme.at for each ferry. And they update continuously whether or not there's a clock looking at them.
So you could look at http://mapme.at/where/snowdrop for example to see where Snowdrop is. So you just had to change the favourites and the user and it worked!
Hakim: So you had this clock that had 2 hands... that's a bit different from the design of the WhereDial isn't it?
John: It is... but actually there is a step in between there... Partly this was inspired by the ferries, and the problem of there being 3 ferries and only 2 hands. So I needed a clock with ore than 2 hands. And the Weasley family clock has, I forget, 8 or 9 hands, because they have a large family. And also somebody else has done a similar device that he called I think the Magic Clock and he did 4 hands on his clock.
Hakim: So you had to compete.
John: Exactly. I think I was first, but he'd one-upped me. So I had to at least equal, if not go better than his. So I started working on a four-handed clock. Which has various complexities, because if you want all four hands to be like a traditional clock, they all have to be on the same axle. But then the axles need to turn at least in different rates. But for this clock I wanted to have them separately controlled.
Hakim: Otherwise for four hands it'd have to turn round and round...
John: And also because of some problems I'd had with stepper motors along the way I decided to try using servos instead. So I did spend... most of a year developing that, working on that in various amounts of spare time. I managed to get a version of it working, powered by servos, largely laser cut parts, which is interesting for a reason I'll come along to later.
Hakim: It might be interesting for me and possibly for other people who don't know much about motors... what are the differences between a stepper and a servo.
John: A stepper motor you control by telling it to move by a number of steps... and you can move it by a number of steps in either direction, but you have no feedback to be sure that the motor has turned by the amount that you told it to. So if for some reason there's a jam and it gets stuck or something like that then the motor may be in a position that you're not expecting it to be in. And there's no way of really knowing that.
With a servo... it has a built in potentiometer, so that will tell you what position the motor is at as you're also telling it to go to a position. So a servo should be more accurate. You can say "go to 45 degrees" and it should go there. They tend not to have a full range of motion.
So a stepper motor is just a motor... it will turn continuously. A servo will only move, perhaps only 90 degrees of movement. The ones I use went 180 degrees. I'd actually aimed to get continuous motion but they sold me the wrong thing unfortunately.
Hakim: So if you've only got 180 degrees... and a clock tends to have 360...
John: Yes! Fortunately there is such a thing as gears. And because I thought I had continuous rotation motors I had bought gears to go from an outer axle where the motor was on to the inner axle where the hands were on. And they were similarly sized gears. So I didn't need any gearing, or so I thought... but as it turned out, with only 180 degrees of movement, I just used some gearing, with a smaller gear on the central axle with the hands, and a larger gear on the outer. So 180 degrees on the outer gives me 720 degrees of the central axle. So I can move the hand anywhere I want. It does mean that unlike a real clock, the hands would sometimes go backwards, and sometimes forwards, because I can't continuously go forwards. But these are the quirks that we have in our devices (laughs).
Hakim: How much about gearing did you know before you started on this project?
John: Well actually I hadn't used gears in a long time... but I used to use Lego... Lego Technik. So I'd used gears before. I'm not doing anything particularly complex anyway, so I was quite happy doing it.
Hakim: So from that step you went to the WhereDial. And one would imagine that having done two hands and moved onto 4 hands that this one would have... eight!
John: Yes... so the four-handed clock worked... but had its own issues. So, mechanically, the two-handed clock also had issues. Because though I'd measured the number of steps that it took to rotate the minute hand, sometimes that wasn't the case. Sometimes you'd rotate it by enough steps and the minute hand wouldn't go all the way around.
Hakim: That's the issues you were saying about steppers?
John: Yeah... although I say it's an issue with steppers, I think in that case the issue was after the stepper motor. So the gearing within the clock would sometimes slip and sometimes get caught and so... even with a servo that may not have been solved, because the stepper motor may have been turning 100 times but somewhere further down the line something was slipping and not turning...
But with the four-handed clock, the connecting something onto... servos are generally not made to turn an axle... so connecting an axle onto a servo turned out to be a tricky thing to do. I tried a few different ways. Still haven't got a good solution. The way that I used with that clock just to test was basically to glue it on. That was the method I used... And it's certainly not the best.
I used hot glue and superglue but both have ended up breaking. Because of course with gearing, and with a servo, because there's so much gearing within the servo, they don't have a lot of play at all. If you stop the servo... it has a lot of torque, so if you stop the hand it's going to generate enough torque that the glue will break quite easily. And then also attaching gears to the central axle.. there were four hands, so it needed four separate axles of slightly different sizes, connecting what was essentially a shop-bought gear... I didn't custom-make my gears, I just bought them online. This meant that I had gears all of similar size holes in the centre, and 4 different sized axles to connect them to... which ended up being a bit tricky. So trying to use tension from a grub screw and the gear, and the glue... and then again with the hands on the front. I tried using tension which... probably would work, but of course at the time I didn't have access to a laser cutter to make new hands, I ended up gluing the hands on, so there was a lot of glue putting it together. It's not perfect, and it would have taken a few more iterations, and was quite fiddly to put together.
And also not having huge amounts of time to put into this, that had been put on the shelf for quite a while, and I hadn't been worrying about it too much.
And then I came across an article on Hack a day, which showed someone who'd made a clock using laser-cut pieces... essentially a laser-cut backplate with some feet, a motor placed in the top. The motor would continuously rotate, very slowly. It had a gear on the front, and a dial around it with the hours of the day on it. And as that dial turns, you read the time from the position of the dial and the numbers on it.
And so Adrian saw that and realised that it was a very good, simple, design for a clock. Something that could be easily repeated. So making 100 would not be a lot more difficult than making 1. At least.. it was a lot less effort to make each one. So it didn't seem like such a scary prospect as making 100 four-handed clocks wit lots of glue that would probably break...
And it looked like something that would actually be simple enough that I could build it and potentially sell it in a very short space of time. So I used a friend's laser cutter (Aaron from Oomlout I cut one myself, and played with the software... fortunately the software for the existing location clock was quite compatible. All it was doing was turning a stepper motor a certain amount to a certain position. So this version again uses stepper motors, for the continuous motion. So I only had to tweak the existing software a it, put it into the new design, and had a functioning WhereDial very easily.
And so I put it up for sale! I set up a website and tried to do a Kickstarter style "If 50 people order this device I will start production and start selling it" and launched that. And that was actually relatively successful. In a fairly short space of time, I had 25 people willing to give me £100 or similar in alternative currencies to have one of these devices. And so I had to seriously look at what needed to be done. Some of the things that needed doing were finding a place for the Arduino to sit. So the existing design was simply a backplate with feet. And I don't know what the original designer did with his Arduino. Maybe it was in a separate box. But I wanted to make the Arduino part of the unit... make it a whole device. So that involved designing a box that would fit into the back, and be part of the device, so I worked on that.
Hakim: Is it still Arduino + Ethernet shield like the original prototype.
John: No. these days it's moved to being an Arduino Ethernet board with builtin ethernet. Again it does need a controller chip for the stepper motor. This time slightly different stepper motors, so slightly different chip. There's a Darlington array, a ULN... P23 maybe... so this chip takes input from 4 pins from the Arduino to control the motor and set it to go forwards and backwards etc.
And that will essentially be all we need. There was a little bit of feature creep as I realised that a "reset" button might be useful, because the dial might fall off and you might need to replace it. And you need an initial state for it... for example with "Home" on the top. So when you press reset, it'll assume that it's starting in home position and then ping the internet to check where to move the dial to... maybe an LED to have a bit of feedback to know it's doing something.. And then I thought if we have 1 button we could have another one that you use to check in... so that if you're at home and you get home, it might be useful to be able to press a button to say that "I'm currently here!"
So maybe you tend to use foursquare but your family's house isn't on foursquare, so you can just tap on their WhereDial. So that's another button and another LED and a few resistors. So all of this means that the controller shield gets a little bit more complicated, and so means that... in theory the best way to go would be to get a custom PCB made.
John: Yeah, a big step for me. I've never done that before. I've always soldered my own boards up and gone that way.
Hakim: When you get the custom PCB, can you even get the components from the Arduino on that as well? or will you get the Arduino as it is.
John: You could build your own PCB with all the parts with all the bits for an Arduino Ethernet... but it's even more complicated and beyond my current capabilities... but maybe something to look at for a future model. But doing that could in the long run bring the costs down... but in the short run it would take so much of my time, and so much expertise that it would be more expensive.
And that pretty much brings us to the current state of the project... in that I'm still at the stage where I need to design these PCBs and get the ordered and sent over... but once I can do that then pretty much all the other kinks are ironed out
Hakim: You said that before you didn't have your own laser cutter and now you have access to one.
John: So... for a good few months, when trying to design the box to go on the back I'd have to come up with a design, go to a monthly Makernight that we have in Liverpool and have access to a laser cutter... try it out... put it together and find that one of the screwholes is ever so slightly off, and this tab doesn't quite fit into that hole, and all kinds of little small things that would just take a few tweaks to a design and just passing it through a laser cutter... but by that time my time is up at Makernight and I have to go home, so I have to wait till next month.. So I wait, and make that change, and come back, and find out that these holes aren't quite right for those LEDs...
Hakim: It sounds like when you used to have to compile code by typing on punched cards and then they would get run overnight by technicians...
John: Yeah, it's a little bit similar, because also the alternative is that I could send away and get Aaron to cut it on his laser. So literally I'd make a tweak and send... well email him a file, and he'd send in the post some designs, and I'd try them and send more.
But then in around January, DoES liverpool, where we both work from, got use of a laser cutter, which we manage for Open Labs, and that made things so much simpler because I could make the tweaks and try again, and basically within just a couple of days of playing with it, if that, I had a fully functioning design that all fits together quite nicely, quite tightly... it makes quite a nice looking and well fitting functional design.
Hakim: So in terms of scaling up, having a laser cutter has really helped you do that... but would you then be cutting the shell here at DoES or would you get that done outsourced?
John: Well, once I have the design worked out... I could probably get that done somewhere else... because the laser cutter we have in DoES is a fairly small one, and you would only be able to cut one WhereDial at a time. So it might be useful if I have 100s to make (which I don't at the moment)... it might be useful to be able to do multiple ones, and would probably go to either Aaron's company Oomlout, or to Offcut Liverpool, another place in Liverpool that offers lasercutting services.
Hakim: How long does it take to cut a single WhereDial here?
John: The current timing I believe... it's a while since I've done one, but about 15 minutes. The version before that was half an hour... but we managed to tweak the etching settings and get it a little bit faster. Because the problem is that the cutting doesn't actually take that long to cut out all the parts, but there are various things on there that are etched on, like the mapme.at logo, and the names of the locations around the dial face... and etching is what takes the most time, because it has to go back and forth over the same space to etch the design on.
But there are ways of making that faster... moving the laser faster while having the power stronger... or doing slightly lower resolution, so that you don't get so much detail but it goes faster. And also simple things such as placing different parts of the etching on different layers in your file and turning your design by 90 degrees can make a big difference... it just depends on how the head of the laser moves.. So if you can optimize for that you can bring the time down. A combination of those is how we managed to halve the time from 30 to 15 minutes.
Hakim: Stuff you couldn't imagined would be the case until you started playing with it.
John: Yeah, you doesn't realise there is so much variance...
Hakim: So to do 100 would take about 25 hours... so about about four man-days of cutting if someone's doing that full-time... and how long does it take to assemble a WhereDial? Assuming you've got the PCB...
John: Even with the PCB... still a few hours.. To build the wooden parts is minutes. 10 minutes really. There are some bits that are slightly fiddly at the moment, so I probably will continue to tweak the design a bit more to make it just that little bit easier, but yeah, it's not particularly difficult. I do want to fully construct it because... I'm aiming for it not to be a kit, really. Especially I'm thinking the amount of money that you're spending on it, I think people would like to receive a fully made up unit. Unfortunately due to the shape of it it's probably not going to be completely made up, because then it would just be a strange shape and difficult to package and post.
So the front dial will be separate and there'll be two screws that you'll need to fit, but it should be as simple as that. Just those two screws.
Hakim: I do have a couple more questions... so you have advertised on your website which was good to check that there was a basic interest that was worth pursuing. But in terms of getting your initial 100 that was your first goal for your first production run, have you got that already? And if not, what other channels are you looking at?
John: I haven't so far. At one point I was talking about 1000, and brought it down to 100. And when I actually advertised, I only put the figure of 50 on. And admittedly these are fairly arbitrary figures. Just coming up with some figure that means that I'm happy enough that enough people will give me money for this thing and it's worth me putting some time in. So far I think I have 27 orders. So I haven't hit the 50, but I'm past half way.
I'm not a marketing person so I did the bare minimum which was to tweet about it, to blog about it, to mention it on the mapme.at website, I think I may even have done a mailshot to all the subscribers of mapme.at.... so obviously the type of people that are interested in tracking their location, and who would hopefully interested in having one of these.
It's been blogged by other people, but perhaps not as much as I'd like. It'd be nice if hackaday would feature it, but some have... Josette has, for her O'Reilly blog.
So really, on the one hand I should do more of that. So perhaps if I get the PCBs done and the stage where it's basically ready to go, I could do another push with "If you order these, you'll actually get them... very soon!"
I could look at other methods of selling too, thing is that it's not the type of things you could put on google adwords, because people don't tend to be searching for WhereDials. And if they are... I'll come up, and they'll buy one hopefully.. I would consider looking at some sort of novelty gift shop type places or gadget shops. They're likely to be able to order more units which is a good thing but also adds difficulties, and they have certain expectations for... well for safety certification and so on... that selling as commissions on my own may not be such an issue.
Hakim: And you looked into Kickstarter... and the crowdfunding options didn't you?
John: Yes... so kickstarter tended to be the preference, if only because it has so much traction at the moment, so there are a lot of people simply looking at kickstarter to see what they can fund. So you don't just get the platform, you get the community. Whereas that seems to be less so with other websites. Also kickstarter seemed to be the only place... at the time.. that would really do the "hit a target or don't hit a target" completely binary. You either hit it and got your money, or you didn't. And if you didn't there was no cost for anyone. And if you did then great!
I believe that's changed now, and indiegogo also have that style of model now. The problem with kickstarter was that it's US based and you have to be, essentially a US citizen, with a US bank account and a social security number... I believe, to set up a project on that. Which does limit me!
Indiegogo don't have that requirement, but also didn't have the model I wanted. Now they do, maybe it's worth revisiting that.
I did look into using kickstarter anyway, going via a friend in the US as a proxy and having them manage the project. But for various reasons that didn't work. Both human in that my friend's very busy and has his own life to worry about, and also... I guess I wasn't aware of quite how much information kickstarter would want from me. And how much apparent commitment they would want from me before they'd accept the project. So when we did submit, via my friend, I probably only gave the minimum information just to... get an account setup so that he just had to do the minimum, and I could put the video up and fill in all the gaps. But it seemed that kickstarter needed more than just that basic information to be able to approve the project. Because that's what they do... they approve projects, and turn down people that they don't think are suitable for their website... and unfortunately that's what happened.
Hakim: What a shame! I'm sure it'll be their loss. But I think the final thing is... let's say that you get your 100 or whatever orders, and it takes 15 mins to cut it and a few hours to build it, how do you go about scaling that up? Is that you spending a couple of months building these by yourself?
John: I guess some of that depends on margins... if I'm making enough profit to make it a full time job for me, then maybe I will. But my time is probably better spent elsewhere. But I can certainly look into bringing in staff and involving others in the process of building the WhereDials, and that would give me more time to build multi-handed WhereDials and start researching more options.
Hakim: As a general principle, apart from the PCBs which seem to be cost effective only outsourced to China etc., the rest of it you'd prefer to have assembled here in Liverpool? Or outsource that too?
John: I think so... though part of that may simply be that... the original idea is to get something that essentially I can build and make and ship out. And that just makes it simpler for me to manage. And of course, as the numbers go up, it makes sense to get someone else involved. But it would be nice to be able to keep it local. And of course there are plenty of people in China that need jobs. But there are also plenty of people here that need jobs as well. So it would be good to be able to keep that side of things here as much as possible.