Update #5 - Shenzhen Part 1
Firstly, sorry this update is a little later than planned. It’s been non-stop since we got back from Shenzhen, mostly because while we were out the we set an ambitious date for our pre-production run. We’ve been working flat out on things like packaging design, designing and making test jigs and adding test routines to the firmware, and we’ve got a lot done over the last couple of weeks. There’s nothing like a deadline to focus the mind!
We’re writing this update at Heathrow Airport, on our way back to Shenzhen for the pre-production run next week, so as you can guess, we managed to get everything done in time. Well, almost everything – we’ll be doing a few final firmware modifications on the flight!
This update is going to be about Shenzhen, and T2 though. Shenzhen, in case you don’t know, is very close to Hong Kong; in fact, it’s only about half an hour from Hong Kong airport by a very efficient ferry service, so getting there is very easy.
Our first day started bright and early with a quick visit to Victor’s office, and then straight to the injection moulding factory. The factory was a really interesting mix of high precision manufacturing, very ‘industrial’ processes and highly skilled, almost craft like processes. In one aisle, you have tools being spark eroded to very high degrees of accuracy; over in the next aisle you have the moulding machines heating up plastic and squirting it into the moulds under tremendous pressure, and then right next to the boss’s office finished tools are being hand polished.
The picture above is a spark erosion machine. Spark erosion is a very high precision alternative to milling – instead if using a tool to cut away the metal it’s eroded away by sparks from a copper electrode. The electrode is specially made in the shape that you want to cut - you can see it in the centre picture (this isn’t our tool, but the same process was used – it would have been great to have seen our electrode but they go off for recycling once the tool is made). The accuracy does come at a cost though – the machine cuts about 1mm per hour, so this is one reason why injection moulding tools take time to make.
While we were at the factory we saw some of our parts for T2 being moulded. It was the U section, which was particularly interesting as it has a moulded in metal threaded insert. The operator manually loads the brass insert onto a pin on the tool before each cycle. Here’s Durrell checking some of the parts that have just come out of the tool.
We left the factory with a few complete sets of the injection moulded parts for our first test assembly. They looked good, but there was a problem with the arms. The plastic comes uncoloured, and the colour is added as a powder; the powder needs to be ‘baked’ before its used and the moulding people though this had been done, but it wasn’t. The parts came out looking pretty awful, and potentially weaker than they should be so they scheduled a re-run for the following day.
Next, we headed to the assembly factory. For the scale we’re working at assembly is quite low tech – people at a row of tables using hand tools. We think it’s likely to be about ten ‘stations’ (or assembly steps) for Line-us, but there will be a few jigs and fixtures to help with assembly (for example there will be one to make sure the rods are inserted into the base so they are perfectly aligned).
There will also be some test jigs – one to test the PCBs and one for a final test once Line-us has been assembled, and a calibration jig. Some of the jigs are being made in Shenzhen, but the more complicated ones, and in particular, any that require test routines in the firmware have been made by us in London. We’ve ended up making four in total, the calibration jig being the most complex and I’ll talk about the making of those in a future update as it’s quite interesting.
We were really looking forward to day 2 – now that we had a full set of moulded parts, albeit with imperfect arms, we were ready to do our first proper test assembly. We’d had a go in London with the T1 parts, but the fits were too loose to get a working machine so this was going to be an important milestone.
We did notice some issues with the parts – the U mouldings were slightly warped still (although better than the T1 samples) and the bearing fits were a little loose. Mostly minor problems that could be easily fixed, although the warping was more of a concern as it is not obvious how to completely correct it. Having said that, we did anticipate that it could be an issue so had thought of a few options. We do have an ‘if all else fails’ plan that involves building up the servo bosses at one end a little so that they are aligned (which is the important part).
We ended up finding about a dozen issues in total – probably about what we expected so when we finally had two complete machines we were keen to try out a drawing. We tend to use the chicken drawing for testing – we’ve drawn it so many times now we know our good chickens from our bad chickens. Unfortunately, the two machines we made drew two of the worst chickens we’ve ever seen. Needless to say, this was a bit of a disappointment! We spent the afternoon disassembling, swapping, reassembling, measuring and scratching our heads.
You might remember from when I talked about tooling that it’s much easier to add material to the part than to remove it. This means that holes are generally made a bit deeper than the need to be to allow for adjustment. It turned out that because of this the arms were fitting lower down on the servo shafts and were rubbing on the servo body. This was very difficult to spot – in fact it was because one of the servos was getting hot that lead us to the problem. After trimming some material off from the bottom of the arm we had a machine drawing very nice chickens!
We can’t talk about Shenzhen without mentioning Huaqiangbei electronics market. It’s said that 80% of the world’s electronics are made in Shenzhen; whether it’s really 80% or not I don’t know, but it’s certainly a lot. It seems like manufacturing, and particularly electronics is a way of life in Shenzhen, and the electronics market is a good example of this. Imagine walking down, say, Oxford Street in London, and seeing all of the usual shops you’d expect: shoe shops, clothes, cafes, and then next to all of those a shop window that looks like this:
Those are reels of electronic components in a shop window! A reel usually contains a few thousand components, so if you’re buying reels you’re making something in volume, and the fact that you can buy them on the high street tells you what Shenzhen is all about.
The shops you see on the street are the tip of the iceberg though – there’s even more action just around the corner at the markets. Literally hundreds (maybe thousands) of small stalls, each specializing in a particular component; this one was just selling crystals, which were on display in china bowls.
As well as components, there are stalls selling power supplies, cables, tools, displays (one of the market had a whole floor selling touch screens) and anything else you can think of. As you can imagine, I did quite a bit of shopping!
So that covers the first part of our trip to Shenzhen, and I’ll have to leave it there for now as I’m almost out of battery. We’ll publish another update covering the second half of the trip either this week or next, depending on how things go. But we hope you enjoyed the this update and do get in touch if there’s anything you’d like to know.
Rob and Durrell