Q2 Update

Dear readers, it’s time for another update !

Since last time, I’ve focused on two projects, the LED strip controller and of course M-Bot. My high-level update is that I’ve made a new version of the LED strip controller which still needs to be tested and programmed, and M-Bot is now alive and kicking, though is still missing its head.

LED strip controller

I’ve fully redesigned the board with some advice from an electronics engineer. The main things I wanted to add were safety features, as users will always find a way to make mistakes. I’ve added an over voltage protection in case someone plugs 24V instead of 12V, a reverse voltage protection in case someone plugs it backward, and overcurrent protection in the form of a fuse that will blow up if something goes wrong, and it is sadly not resettable at the moment. (First picture is v1, second picture is v2)

I also changed the MOSFET from a P channel to a N channel; they both function in similar and yet different ways, they both act as gates that are set ON and OFF by setting the gate voltage. P channels are before the load and N channels are after. They have different “chemistry” which makes the N channel more efficient [Rds(On)], so they heat up less. In my case I calculated more than 100 degrees of temperature rise for a P-FET, and 30 for a N-FET, this means the components can be smaller and cheaper.

While testing I’ve found an issue that the LED Strip requires some time to switch off, because of that I’ve recently invested in an oscilloscope to debug electrical issues.

Above is a screen capture from the oscilloscope, the ON time of the transistor gate is tiny (blue) which sets the MOSFET gate low powering the strip (yellow) by letting the current through. But it takes a long time for the MOSFET voltage to drop back to 0V.

Here is the schematics, and the reason we can see this behaviour is most likely due to R34 value, 1MΩ, a value of 10-20kΩ is probably more appropriate here, so the gate reverts back to 12V faster and blocks the current flow.

I’ve looked into manufacturing these devices and went through the EU regulation to understand what is needed, and wrote an article on my blog, you can skip the next part if you already read it.

I need to engineer a product compliant with all the regulations to be able to sell it in the EU, and the EU wrote multiple “harmonisation” documents for products. The regulations explain things such as : a power supply needs to have minimum efficiency while powering or not powering something, the goal of that regulation is to save energy on things plugged to the wall that don't do anything. It’s not too burdensome to actually apply these regulations, they are well written and to the point. However you need to find all the regulations that may apply to your product, and argue why some are or aren’t relevant to your application. So going through the definitions and exceptions of every Harmonised Standard.

(CE symbol stating the compliance with EU regulations)

Once you’ve engineered that device you need to write a bunch of documentation and keep it for 10 years as the manufacturer. I’ve not found an example of what is required or standards to follow, so it’s pretty broad. Then, time to manufacture your product in compliance and keep track / record of how you complied with the regulation. For EEE products, the main regulation is ROHS, which regulates harmful substances within an electrical appliance, things like lead or cadmium.

Also you need to be part of a WEEE group, which is a recycling chain, you need to have a contract with as a seller of EEE goods.

(WEEE Symbol requiring the product to recycled)

This is to me the most annoying part, as it requires working with private for profit entities. You also need to report TONS of products put on the market to the government, which is actually a simple process. The fact that you need to report tons of products really shows the barrier of entry of a seller of electronics goods in the EU. Also every member state (MS) is different and you need to sign a contract for every country and report the tons of product put on the market for each MS, so it makes it quite expensive to sell your product across borders.

Also, take what I wrote with some salt, I am no expert, and I also skipped many details in order to write something as concise as I can.

M-Bot

M-Bot is now fully sentient and intends to take over the world, thankfully it is not yet very coordinated. It is moving, all axes are functional, even though it’s still quite brain dead.

(This is M-Bot’s brain, with a servo and a stepper motor, sensors and LEDs. I am doing brain surgery instead of finishing this newsletter !)

I’ve worked on the nervous system (i.e. : the electrical system) till now, and just started to program the main microcontroller.

On the mechanical side. The tool head; the most complicated part of the robot; has been made. I had many issues, and it took many weeks to machine and to build. A true learning experience. The main point I am taking away from this project is that some parts are better bought than built.

You can learn a lot from making parts and many have to be made, but some are easy to buy, and this is something I learned the hard way. For example, to hold my 8mm shaft, I used a nut which I drilled the thread from, and drilled a 2.5 mm hole which tapped with an M3 tap. This took hours to do, and broke many drill bits (and taps). Parts which you can buy for 1 euro each with shipping and next day delivery.

(Aforementioned parts called “Serrure à vis mère” not sure what is their english name)

Mechanical parts often seem easy to make, however the accuracy often necessary to make them is hard to achieve for me in my shop without a mill or lathe. It can take multiple hours to drill 6 holes into 3.5mm thick steel. Making the holes concentric with other parts, in the right position and with the right tolerances; it is so much easier in theory than in practice. In the future I am pretty sure I need to find someone to work the mechanical side with me, someone with a lathe, a mill, and the dozen of other tools required to make them work. This is something that cannot be improvised or done quickly.

(The CNC Controller board I made for managing the tool head location in 3D space)

On the electrical side I’ve wired all the switches and motors, and discovered I had some thermal issues which triggered the thermal cutoff switch in the IC, and burned my fingies with the hot hot ICs. I’ve temporarily solved the issue by under powering the motors, and stuck some heat sinks on them. The issue is due to my buck converters, which are quite underpowered. I did use 3 of them (green bits next to the LEDs), one for each part of the circuitry to try to alleviate this effect. I’ll likely redesign this part of the circuit with my own step down power supply, or simply use an external buck converter.

(close up on the P Mosfet)

I had a small issue with my MOSFET, if you look closely at the schematics it’s plugged in reverse, so the motor is always on. You can see below the zener diode (-⊳≀-) points towards the SPINDLE (i.e. : bottom), so the gate doesn’t do anything, and let the current through. This was solved through “skilled” wiring, and bending of the TO-220 package to fit the requirement.

(Some parts of the schematics for controlling M-Bot that can be seen on the breadboard)

Next steps for M-Bot is to fully test the robot, replace everything that doesn’t work, create something to hold the piece of wood, design and order a custom PCB for the micro-controller. Quite a bit of work to complete the prototype, but I’ll soon be able to try it out ! Right now I’m doing the software for each part, and testing the electronics so I can design the PCB.

Conclusion

It feels like I’ve been less productive this time, I am not sure if this was actually the case or this is because I ran into harder issues to solve. Hopefully I’ll finish one of these projects soon, so I can showcase it.

I’ll soon be back in Munich working with the start-up on firmware, test benches, electronics, and CAD designs, so doing what I enjoy with people I enjoy being around, personal projects will take a bit of a break, though I usually find the time to order parts, and do some work I postponed.

Fox ears, or my bird network project is on hold. I ran into a firmware issue where the microprocessor doesn't have enough RAM to process enough FFT, and the triggering accuracy is pretty bad. Also, I decided to focus my time on the project I can realistically finish and that one is too much for me alone in the time I have.

I hope you’ve enjoyed reading this newsletter, and you were entertained by my work. It is tough to decide on what is worth writing on and what is not. So let me know if there is something I should write more about.