Electronics > Projects, Designs, and Technical Stuff

Looking for criticism on Micromouse board design

<< < (2/2)

MK14:

--- Quote from: Frzoen on September 02, 2020, 06:12:25 am ---The response time isn't crucial.

--- End quote ---

I suppose at hugely slow speeds, while moving in a straight line, it might work, some of the time, maybe.
But at higher speeds, when trying to do tight turns, after wheel(s) have slipped, and the position/angles are not where they are suppose to be, maybe not.

Anyway, it should be a good learning exercise, either way.

You have got, what 2 cm of clearance, either side when doing turns ?

Perhaps 18 cm - (wall thickness + Micromouse width + initial position error) = Not that much.

So, even small angular/positional errors, can soon mount up, turns are not the easiest of things to do, especially at speed. So, very slow responding sensors, doesn't help.

If it works reliably 90% of the time, that would mean the Micromouse would crash every 10th turn or movement.

Frzoen:

--- Quote from: MK14 on September 02, 2020, 09:04:57 am ---I suppose at hugely slow speeds, while moving in a straight line, it might work, some of the time, maybe.
But at higher speeds, when trying to do tight turns, after wheel(s) have slipped, and the position/angles are not where they are suppose to be, maybe not.

--- End quote ---

By saying that in second phase robot relays mostly on encoders i didn't meant that we'll ignore distance from walls.
I agree that is't still necessary to close loop of the control algorithm with feedback from VL53L0x.
In previous version of micromouse we're using VCNL4010 proximity sensor (not ideal for distance measuring btw.).
With this (VCNL4010) we're pooling each sensor with frequency of 10Hz and that was sufficient enough. With VL530Lx we wanted about 20Hz measurement frequency witch i think is achievable.

16.4cm is a track width (0.8 per wall thickness) - 5cm of robot width gives us ~5cm on both sides of robot. I think we are good with robot dimensions  :)



--- Quote from: MK14 on September 02, 2020, 09:04:57 am ---So, even small angular/positional errors, can soon mount up, turns are not the easiest of things to do, especially at speed. So, very slow responding sensors, doesn't help.

If it works reliably 90% of the time, that would mean the Micromouse would crash every 10th turn or movement.

--- End quote ---

Slippage of wheels, miss positioning of robot, kinematics, dynamics, control theory and so on was my daily concerns on my bachelors/master studies. I think we've sorted them out. :)

Ps. This is out 4th micromouse robot (if counting only those made on our PCB's not made from arduinos :) ). Below some pictures of older ones.

MK14:
I'm glad you're improving your robot designs, on an incrementally better, each time process, and learning from it.

daqq:
Welcome to the forums! Very nice design.

If I had to nitpick:

The button debouncing circuit isn't all that great - when you press a button you will be effectively shorting a capacitor to ground. The energies aren't particularly high, but you aren't behaving nicely to your switches and the contacts *may* get unhappy. Most of your debouncing can be done in software these days.

Not sure the buzzer needs a series resistor.

If you want to save space, you may want to go for a single multiple output DC DC converter instead of 4 LDOs. Or a multioutput LDO

Wouldn't it be easier to operate the VL53xxx of of +3V3, thereby saving you a power rail and trouble?

Frzoen:

--- Quote from: daqq on September 02, 2020, 09:15:34 pm ---The button debouncing circuit isn't all that great - when you press a button you will be effectively shorting a capacitor to ground. The energies aren't particularly high, but you aren't behaving nicely to your switches and the contacts *may* get unhappy. Most of your debouncing can be done in software these days.

--- End quote ---

I'm aware of that. To be honest i just forgot to put them into the design. There is even issue on git about this "bug" and in next revision i will add those resistors. Still i like to have hardware debouncing. I like to simplify my work later in software (i know that this approach is not suitable in large scale projects, where every capacitor and resistor counts).


--- Quote from: daqq on September 02, 2020, 09:15:34 pm ---Not sure the buzzer needs a series resistor.

--- End quote ---

I followed this article: https://www.murata.com/en-global/support/faqs/products/sound/sounder/char/sch0007 for buzzer circuit.
I wanted to be able to mount both piezzo and electromagnetic buzzers (changing the diode in parallel to the buzzer for the resistor).
Right now this series resistor is not present and his pads are shorted together. This is thing to change in schematic to be more clear about this options.


--- Quote from: daqq on September 02, 2020, 09:15:34 pm ---If you want to save space, you may want to go for a single multiple output DC DC converter instead of 4 LDOs. Or a multioutput LDO

Wouldn't it be easier to operate the VL53xxx of of +3V3, thereby saving you a power rail and trouble?

--- End quote ---


This is my first adventure with VL53xxx and i wanted to operate them in their optimal conditions (datasheet suggests 2v8).
Next time i will definitely go for operating in 3v3 to save me this 2v8 rail and also go for multioutput LDO for 3v3 and 5v rails following your advice.
I want to keep 6v rail separate (it's only supplying motors, feeling right thing to do). Again there is a option to eliminate this rail (6v) and connect h-bridge directly to +7.4V from batteries.
This would enable higher speeds on this same gear ratio but i wanted consistency in operation of motors (removing variable of changing supply voltage on motors for control algorithm).
Pololu 6V motors easily can handle up to 9V of operation (not and advice  :) just an observation) but then they wear out very quickly.

Navigation

[0] Message Index

[*] Previous page

There was an error while thanking
Thanking...
Go to full version
Powered by SMFPacks Advanced Attachments Uploader Mod