Mini Robot Cart – Part Two
This document will list the major parts of the project and discusses some of the design decisions made. The pics show how the parts fit in the box. The aim of the project was a small device, controllable via a web page with some ability to make it autonomous in the future.
Note: When documenting the parts listed here I found it distrubing that some are no longer available. But this is understandable, if something does not sell, why stock it and in the case of the manufacturer, a new better version will likely generate more sales.
Raspberry Pi Zero WH (Wireless with Soldered Headers) SKU: CE05324
Raspberry Pi Camera Board v2 - 8 Megapixels SKU: CE04421
Raspberry Pi Zero Camera Adapter SKU: CE04763
Raspberry Pi Camera Case - Black Plastic SKU: PRT-12846 (no longer available)
Pimoroni Explorer Phat for Raspberry Pi Zero SKU: PIM137 (available direct from Pimoroni)
298:1 Micro Metal Gearmotor with Extended Motor Shaft SKU: POLOLU-2208
Magnetic Encoder for Micro Metal Gearmotors, 12 CPR, 2.7-18V SKU: POLOLU-4761
Pololu Micro Metal Gearmotor Bracket Pair – Black SKU: POLOLU-989
Adafruit Mono 2.5W Class D Audio Amplifier – PAM8302 SKU: ADA2130
Pololu 5V, 2.5A Step-Down Voltage Regulator D24V25F5 SKU: POLOLU-2850
NiMH Battery Pack: 7.2 V, 900 mAh, 3x2 AAA Cells, JR Connector SKU: POLOLU-2235
Pololu Wheel 32x7mm Pair – White SKU: POLOLU-1088
Pololu Ball Caster with 3/8" Plastic Ball SKU: POLOLU-950
Jiffy Box - Black - 83 x 54 x 31mm CAT.NO: HB6015
27mm All Purpose Replacement Speaker CAT.NO: AS3002
White 5mm Cree LED 45000mcd Round Clear CAT.NO: ZD0290
To connect the wires I have found the following kits to be invaluable. Highly recommed buying a crimp tool, even though it is rather expensive. The tool I bought had two positions on the die and suits the connectors listed. The current tools stocked by Core Electronics have three but is more expensive. (approx $50 to $70)
JST connector Kit (2.0mm) SKU: FIT0299
JST connector Kit (2.54mm) SKU: FIT0255
I use a hot glue gun to hold some parts together. It is not ideal and the glue eventually loses it sticking ability over time. But it is easy to add and remove parts and change the design.
These projects begin with an idea and get added to as they go along, essentially the device here is a prototype, a proof of concept. With that in mind, the design is not pretty in some respects and the location of some items is not ideal. The pics were taken after the project was built and working; they are not intended as a ‘how to guide’.
Raspberry Pi Zero and Camera
The size of the Pi, Linux operating system and WiFi were major reasons in chosing the Pi Zero. The position in the box allows the ribbon cable to run flat between the Pi and Explorer board. Placing the camera on the front kept the height of the cart low. From the pics it can be seen the top of the camera box is padded out with nylon washers. The camera is pointing up too high if the box is flat against the case.
Another disadvantage is the view angle of the camera is too narrow when close to objects; a wide angle lens would be better.
Pimoroni Explorer Phat
Sometimes I see a new product or one I think is interesting; I buy it, only to have it sit in the parts bin for some time. This is the case with the Pimoroni Explorer board. I think it is an excellent interface board for the Raspberry Pi Zero. It is the same physical size, handles 5V input/output, has 2 motor drivers, 4 darlington drivers, 4 digital inputs and 4 analog inputs. It is no longer listed on the Core Electronics web site but is available direct from Pimoroni.
Battery, Power supply, On/Off switch and Charging
The Pololu (5V 2.5A) power supply is an excellent product; small, stable and outputs more than enough to power the cart.
The battery (7.2V 900mAH) was chosen mainly due to its size. Various other batteries were considered and rejected.
Charging the battery was a problem. I had built other projects were the cover had to be removed to charge the batteries. After a few times this becomes a pain. Using a jumper header with long pins on both sides protruding through the case solved the charging issue. It also allowed the cart to be powered from an external source when developing code, conserving the battery charge. The header is glued in place with superglue making it quite robust.
Two different size wheels were tested with the motors. The larger size being better, but the motor shaft did not extend far enough for the wheel to be attached without rubbing against the box. The smaller wheels work ok but can slip on some surfaces.
The idea of the castor wheel was to make it easier to turn the cart. It also reduced the size of space needed for the non driven wheels. In practise the castor wheel works well on smooth surfaces but not on rough ones or outside on a concrete path. It tends to pick up small bits of gravel which make it hard to turn. Use of a metal castor wheel might have been better or a brush system to clean the wheels as they rotate. Little bits of gravel can end up inside the cart which is not good.
The motors were chosen to investigate how Micro Metal Gear Motors perform as well as how a magnetic encoder works. The motor shaft needed to extend beyond the casing, not all motors have this feature. I wanted the cart to move slowly, previous carts had moved too fast. The potential gear ratios were 298:1 & 100:1, lower ratios were deemed to be too fast. I chose the 298:1 motor; in hindsight the 100:1 might have been better as the cart moves very slowly. I have not tested it with the 100:1 motor, at almost $25 each these motors are not cheap in my opinion.
Placing the motors on the removable panel made it easier to fit and test them and allowed easy access to other components in the box.
Working through the physical design with available parts, coming up with something that fits together and works nicely is reward enough in itself. Measuring how the motors perform, how far they rotate for a given time, how much current is used. Counting the pulses from the encoders. What sort of battery will work, how much power does it need. And then building the design from a blank jiffy box to something that moves and does stuff. It is the part of the design process I like the most.
Next and final part will cover the software.