Disappointingly we are having some issues with some of the boards from our first stack of eChook Nano v1.1’s. So far the quality of boards (v1 and other projects I have worked on) from dirtypcbs.com has been great, their cheap and fast service always delivering good quality boards without issue for my needs.

I was initially a little displeased when I received this latest batch, 12 eChook Nano boards, the first kit board we would send out for schools to test. I had hoped the schools would mostly be testing our documentation not our board quality!

The registration of the silk screen is quite poor. This means that the silk (black lettering/numbering) was not nicely centred on around the components. While this is annoying and leads to a less professional looking product I didn’t think of this as a major concern as it would not impact functionality and the board that was put together showed no signs of any issues and passed all of our tests.

Looking in more detail at some of the other boards received (work have recently purchased a microscope for my lab) I noticed that it wasn’t just the silkscreen that wasn’t on centre. The pictures below show the extent of the problem quite well, with the plated pad area (shiny metal) clearly not centred in the area where the solder resist should be removed (the black area). The solder resist on these boards is the white stuff on top, this resists solder and makes a PCB much easier to solder together as it will stop the solder flowing, in our case down the tracks and potentially in to other components. Specifically here, because the board has a large ground plane pour, it will prevent a solder bridge of the PCB pad to the ground plane as the ground plane is hidden below this protective white layer of resist.  This is important as (for all track apart from the ground pads), a short to ground will result in either incorrect readings of the sensor data, or worse, a damaged PCB.

Unfortunately…this hasn’t happened in this case. You can see in this lower image the tip of my multimeter probe, just skirting the edge of the solder resist. In this case I am checking around the 12V battery pin. Around the very edge of the white region I was finding my multimeter continuity test was  showing this region as connected to ground, not good. This means that the ground plane is slightly showing through the solder resist. If I were to solder this pin it is likely that this would be bridged and cause a short between the 12V input and ground.

Can this all be blamed on the manufacturer of the PCB? Simple answer, no! Although the offset here has resulted in an issue I (as the PCB designer) could have taken steps to reduce the likelihood of this occurring. What I now realised I should have done is reduce the size of the area that is removed for the solder mask, effectively making the solder mask (white stuff) bigger and closer to the plated pad (a smaller black area). Doing this without changing the clearance of the pad area to the ground plan, this would have the effect of covering up a larger amount of the ground plane so even when the registration is poor between these parts of the PCB it is less likely to leave exposed ground plane showing.

An annoying blog post to have to write but a valuable lesson either way and certainly something that has added to my learning of the finer details of PCB design. The team will be modifying the gerbers and getting some new boards in to rectify this issue as soon as possible so we can get the kits out there!

We’ve finally got our (hopefully) final prototype boards together! The eChook Nano v1.1

These boards have a different layout, different DC-DC Voltage regulator with short circuit protection and reverse polarity protection, a lower component count and design changes to make them easier to solder. We have also differentiated the power connector to all the rest so that it is far harder to accidentally fry the Arduino! The functionality remains the same as the original prototype boards.

Thank you very much to the teams who gave us feedback on the initial boards 🙂

We are also re-organising and re-writing a lot of the documentation to make it easier to follow. The build instructions for the eChook board are complete, I’m currently writing up the section on connecting the board to the car and instrumenting the car.

We are providing these eChook boards in kit form, for students to build and program. All instructions are provided in the documentation. We’re still selling them for £25 posted for the kit shown below. You will need to purchase the LEM HAIS 50-p current sensor(~£18) separately. This is cost price – we’re not aiming to make a profit.

*Kit Photo missing a diode and length or ribbon cable that are also included.

Alternatively if you are lucky enough to have PCB making facilities, all the files are hosted at https://github.com/eChook and you can make your own 🙂

In case you haven’t worked out from the lack of us being at races so far this year, 3galoo has not been progressing smoothly.

We’ve had numerous problems with the wheels and driveshafts at the rear of the car, and I thought it’d be a good idea to record what we’ve learnt so that other teams out there don’t repeat the same mistakes.

Without further ado:

Bearing Fits

We had the hubs and bearing spacers for the wheels machined by a local engineering shop. Having checked my drawings with one of the technical specialists at work, I had good confidence that the drawings were correct, and most importantly, correctly toleranced.

Sadly, the parts we received back were not up to scratch. The recesses within the hubs for the bearings were too small for us to fit the bearing, although not by an amount that we could measure with a set of calipers. It wasn’t until I sent the parts to my friendly technical specialist that we realised that they were less than 1/10th of a mm too small – this was enough to completely prevent us from fitting the bearing.

Foolishly, we wrecked one hub by trying to turn out the extra hundredths of a mm with our bench lathe which really was not up to the job – we went from not being able to fit the bearing, to it falling immediately back out.

This all put a massive damper on progress, and enthusiasm – as the first thing that we’ve not done ourselves, for it to go this poorly was a bit of an eye opener. We know for next time – make sure our tolerancing is documented properly, make sure the person doing the machining understands and agrees what is required of them, and if possible, get the bearings to them that you want to fit!

Shaft diameters

Our next problem was with the shaft that we bought to use as the rear axles. To our calipers, the shaft we bought was a constant 18mm diameter the whole way along, so we couldn’t understand why it would only fit part of the way through the holes that we needed it to, and why it slopped around at other points.

Again, a measurement with a higher resolution tool showed us that the shaft was definitely not a constant diameter along its entire length. We were advised to instead purchase Cold rolled bright mild steel, which has a much more consistent outer diameter, and has really reduced the amount of wobbliness at the back of the car.

From this we learnt how much difference an imperceptible change in diameter can make, and therefore the importance of using the right materials.

Tolerance Stack up

The design we came up with introduced a lot more ‘critical tolerances’ than it needed to, meaning that many dimensions on many parts needed to be machined perfectly for the design as a whole to work.

The redesign at the rear end used a shaft that we knew would fit well into the I/D of the bearing with no further work, reducing the number of ‘critical characteristics’ by two.

This approach of designing to minimise the ‘critical characteristics’ of any given part makes the entire system more robust, and easier to manufacture, which is a big win!

Hopefully this has been of some use to someone who’s looking at building their own car!