Thursday, 9 February 2017

Proof of concept hardware (Anton)


To make easier for maintenance and debugging I wanted to have solid soldered joints (something which will not come lose after I will move it between the classes). So I decided to have own PCB with lot of test points exposed so I can connect scope probe to it. In past I made mistake that ground test point had only 1pin and was easy to break, now I used 4-5 pins to make the joints physicaly stronger. Added simple power supply made from 7805 linear regulator, this way I can connect my LiPo batteries or DC power supplies to it, this way it's portable if needed. I had to only give it heat-sink because linear regulators are very inefficient and convert everything to heat.. 


First I tried to 3D print the gimbal for the both servos, but the design files I found weren't good and had very weak walls and broke quickly. I would be surprised if they would hold up to any abuse, 3D printed walls with thickness under 1 mm can't be used for any mechanically abused parts. I could probably design my own, but considering the short time frame and that this is just proof of concept I decided to opt out to full plastic gimbal.


Because I wanted my own PCB there was a problem, the USB connector on top right were too high and could touch the PCB and cause short-circuit.

So I took dremel and separated that part to be sure even if it touches it will not bridge all the pins together, before assembly I put electrical tape over it just to be completely sure.


When I was testing the servo routines I noticed that when motherboard was connected then there was nasty anomaly. Even if the servos were working correctly I wasn't happy with strange step in the waveform. It could be caused by the motherboard, the multipurpose chips have many functions combined and shared on the pins, so sometimes the hardware needs to be configured to be in some specific function, I might not know all aspects of the mbed hardware and the fault could be on my part. Not configuring the output pin properly could be one cause. Other source of problems could be my PCB, I could have use shorter signal paths, I might have introduced some parasitic capacitance as well. Custom etched PCB would be the best solution, but for now prototyping PCB will have to do. So I decided for this proof of concept to abandon the motherboard to avoid me getting into some issues which could waste my time.

I will borrow group member's motherboards to measure them on the scope to see if it's not caused by a faulty motherboard.

After disconnecting the motherboard the servo signals were clean, so for the moment I will have to do without the motherboard.

For sure adding the test points for easy oscilloscope probing paid off immediately. I don't know if it's good sign when I already put into design fallbacks for moments when the hardware will misbehave as I wouldn't  trust it to work well. 


No comments:

Post a Comment