UPDATED: Weave Modular Robotics Problem

[WaterPig Master]

[Formerly "Return to Robotics"]

Hi there everyone,

I joined this forum several years ago, at the age of 12, trying to make robots. I made the classic complete beginner mistake of trying to do too much, too fast, and not understanding any of it. 4 years later and I'm back in the game! Taking things more slowly and understanding all the circuits in my current robot, Weave (photos attached).

So called because his shell is a woven wood bowl (from a Chinese supermarket in Liverpool), the base is plywood. Homemade spacers hold the circuit boards and shell onto the base, and a lego subassembly holds the Copal gearmotors, lego gears and wheels on the bottom.

Currently the only circuit is a MOSFET H-Bridge Motor Driver (driven by three 4427s that Maxim was kind enough to send me free samples of) and power regulator, built in Dead-Bug prototyping style. It's a bit rickety but works well, and was good soldering practise for the other two boards required, a 38kHz IR obstacle detector board and a board with the brains (a PIC of some form) on. They're coming later.

I look forward to exchanging knowledge and experiences on SOR!

Thanks,
Barnaby

[Soeren]

Hi,

Is the chassis made of a car tire?

I like the alternative material choices, but you might need to pour some epoxy over the loose component wires to hold everything firmly in place, or it may shake 'n' break apart.

[WaterPig Master]

Is the chassis made of a car tire?

Nope, it's plywood, sprayed black and lacquered. A car tyre would make an… 'interesting'… chassis!

I like the alternative material choices, but you might need to pour some epoxy over the loose component wires to hold everything firmly in place, or it may shake 'n' break apart.

Yep, I've had to repair a few breaks already, I will probably set the circuit when I've got the brain made, and confirmed that the driver works alright. Part of the reason I'm taking a 'real modular' approach is so that I can swap out this motor driver with another one (I have some SOIC HIP4081s ready) in the future, without changing any programming.

Thanks,
Barnaby

[Soeren]

Hi,

Nope, it's plywood, sprayed black and lacquered. A car tyre would make an… 'interesting'… chassis!

My thought exactly, but even knowing it's not, it still looks like a car tire to me (when you grind down a tire getting below the outside rubber, it will look like this (the cheap and/or old type without steel wire). Might be some kind of aliasing in the pic though.

Yep, I've had to repair a few breaks already, I will probably set the circuit when I've got the brain made, and confirmed that the driver works alright. Part of the reason I'm taking a 'real modular' approach is so that I can swap out this motor driver with another one (I have some SOIC HIP4081s ready) in the future, without changing any programming.

That's my favorite approach and it's very easy to trouble shoot too - but you could still do that on PCB (which is quite easy to make at home, as long as you don't need super thin traces).

[WaterPig Master]

Hi Soren,

That's my favorite approach and it's very easy to trouble shoot too - but you could still do that on PCB (which is quite easy to make at home, as long as you don't need super thin traces).

As a matter of fact, I have just started etching my own PCBs, and am in the process of developing a product that will make the entire process easier.

I am hand drawing the traces on with a sharpie (they work quite a lot better than Staedtler markers — thicker ink), then drilling with a little german dremel like thing. So far I have made a single sided 38kHz IR board for this robot (Works really well, photos coming to the projects board soon), and also a double sided PCB for a logic probe (hand drawing that was tricky). You know those little sets of plastic drawers that are useful for components? I've decided to make some little utilities that are contained entirely within them. Started with a logic probe (that includes an 8 bit counter — to count pulses, or help me roughly judge the frequency of a signal). I'm thinking of doing an LED/Battery tester too.

Thanks,
Barnaby

[WaterPig Master]

Hi there everyone,

My robot (Weave) has progressed! I have made the 555-timer-based 38kHz IR emitter/detector board (2 emitters, 1 detector — explained later). The board had some spare space, so I added three generic NPN transistor switches for general purpose use later on in the project.



This first photo shows Weave with it's shell on. This is a woven wooden bowl found in a Chinese supermarket in Liverpool, of all places. It's surprisingly nice material to deal with (or maybe that's because I'm a musical instrument maker?), and drills/cuts very easily. It is elevated using some rather roughly made spacers — hardwood dowels with embedded bolts and nuts. There are smaller ones that attach the lower storey of circuit boards to the base.



This second photo shows the motor assembly on the bottom of Weave. It is made of Lego technic, and Copal 50:1 (I think) gearmotors. The couplers are permanent, glued on with Araldyte (2 part epoxy). The motors are low power and efficient, and the only noise generating components are the Lego idler gears. I am willing to sacrifice noise for ease of assembly. Talking of ease of assembly, the motors are lodged in with pieces of veneer. Effectively a friction fit, relying on the flat sides of the motors and the rough wooden surface.



This photo shows the IR board — home designed and etched. In the middle, the power header + Supply caps. To the top right, the 555 with adjustable circuitry. Along the bottom: transistor powered IREDs in heat shrink tubing, with brightness pots. In the middle, the 38kHz detector.

Because this detector ignores any constant frequency, I cannot use them for a logic based robot that has two emitter/detector pairs. So, I am seeing this as a good excuse to build on my PIC programming experience and figure out how to send a set of pulses from one IRED, store the result, then send pulses from the other IRED, and take action.

The IR board mounted on Weave:



That's as far as I've got. I'm currently designing the board for the brains.

Thanks,
Barnaby

[waltr]

That is a wonder little Bot. Very nice work.
Please keep us updated as this has been an enjoyable thread.

[WaterPig Master]

Hello All,

Weave has a brain! And, a brain that can be programmed in place, is highly expandable (on-board breadboard and an entire port of ADCs, software comparators and timer inputs to have fun with!) and control the motors.

The brain is a 28 pin PIC: The PIC16F886. I went for 28 pins as they're a nice size, and offer the kind of expandability I want in a modular project. I was tempted to use a similar PIC18F chip, but I'm more comfortable using the 16F asm at the moment (fewer instructions and features to get my head around) — and, they have built in software/hardware comparators!

This first photo is a shot of Weave, in his near-'complete' state. Note that there is currently no connection from the brain board to the IR board — it's not even receiving power at the moment.



The second photo just shows the copper on the board. I'm particularly proud of this etch — the largest and most complex I've done so far. Uses a double sided board, but only mounted 3 SMD components on the top. All traces and patterns are hand drawn.



This photo shows Weave with the brain board unattached, and the spare battery shown for clarity. The batteries are 7.4v LiPO clusters I got for a meagre £1 each from eBay — they seem pretty hardy and not at all bad so far, but I've only been using them for two days now. No explosions yet… :|



If anyone's wondering, the circuitry on the breadboard is a simple debounced switch. I didn't include one on the board as it wouldn't always be reachable. Currently the button starts/stops the motors. My first PIC ASM program to use interrupts! Took a while to get my head round it, but now the program runs through a series of motor control statements whilst checking the button for a press every TMR0 timeout. It's not greatly efficient, but I'm still learning.

So, next up: Connection to IR board and get it avoiding obstacles!

Although Weave is not a real, sensor-data-interpreting 'robot' yet, I now consider it to be partly finished (It does flash lots of pretty lights, after all ). So, what have I learnt from the project so far? Rather a lot, it seems:

Vastly improved my basic soldering skills.
Learnt basic board layout skills, as well as gained experience etching, and am half way through testing/prototyping a board etching product I plan to release in the near future.
Learnt how to prototype in dead bug style.
How to use MOSFETs and FET drivers — in general, and to control motors.
Memorised details of basic transistor switches.
Become more comfortable using calculations such as Ohm's law. Begun to actually apply them to circuits.
Using PICs! Standard circuits, pull up/down resistors, BORs, interrupts, delays, timer peripherals, developing them using a Mac (A challenge in it's self…)

And assorted other stuff, like what companies will give me free samples (Thanks, Molex Microchip Intersil Analog Devices Maxim!). So, not bad so far

Thanks,
Barnaby