As well as the components kindly provided by Element14, Atmel and Infineon there were several additional components purchased for this project. These were bought as the project progressed or were already in the spares drawer. Where the original supplier was not known the costs were estimated based on current prices. Where more components were purchased than needed the costs were pro-rated. I've tried to include absolutely everything but it is possible that something was missed.
An attempt was made to use as many of the provided parts as possible. Although these parts did not cost me, they are incorporated into the cost estimates to give a more realistic reflection of the cost of the project. Rather than incorporating the entire cost for the workshop kit the individual parts used were estimated from the Farnell site. The exact switches could not be found so an equivalent was used.
|Infineon RGB board||Element14||£21.04|
|IRF520 Power MosFet||Element14||£0.61|
|2* 330K resistor||Element14||£0.01|
|2 * 10k resistor||Element14||£0.01|
|2 * Tactile Switch||Element14||£0.18|
These were specifically purchased for the project, postage is included in the cost and where multiple items were purchased the cost was pro-rated between the items. For the 3D printed parts the cost was estimated by uploading the items to Shapeways and using their material value to provide an estimate.
|House||uksalesmaster on Ebay||£17.38|
|Mini Servo||bgood2010 on Ebay||£2.19|
|LM393 Comparator||D&M Components on Ebay||£0.50|
|1M resistor||D&M Components on Ebay||£0.10|
|2x 5.1v Zenner||D&M Components on Ebay||£0.20|
|Piezo sensor||Electronic Bits and Bobs on Ebay||£0.40|
|Stacking headers||AdaFruit via CPC||£1.59|
|Paint||B&Q tester pot||£0.25|
|Lithium Polymer Battery - 2000mAh||Cool Components||£13.69|
|Adafruit PowerBoost 500||AdaFruit via Cool Components||£13.00|
|Nuts and Threaded Rod||Model Fixing||£4.28|
|Fibre Optics||Fibreopticlightingsolutions on Ebay||£0.99|
Parts that were already available
These were found around in the workshop, either from abandoned projects, multipack purchases or salvaged. The solder was estimated using the method found on http://electronics.stackexchange.com/questions/30671/estimating-assembly-cost
|DHT22||PENGHUIMEI on Ebay||£3.74 (2013 cost)|
|Stripboard / Tri-pad board||RadioShack||£0.50|
|Stainless steel wire||Halfords||£0.06|
|Sugru x 2||Sugru||£2.00|
|Bondloc Structural Adhesive||Maplin||£0.50|
|Heat Shrink Tubing||Proto-Pic||£0.04|
Provided parts: £74.66
Additional Components: £73.65
Things I had: £13.65
Grand total: £161.96
The Enchanted Cottage is effectively a prototype so there's plenty of decision that were made along the way. This has resulted in additional materials use and less than optimal design. This means that there is plenty of scope for cost savings.
The key tasks to reduce cost are:
- Reduce component count
- Replace/Remove expensive components
- Simplify design
Also many of the components were bought in small batches so buying components in larger batches the costs would be reduced and multiple different postage costs could be eliminated.
Reduce Component Count
The Arduino Yún is a good board but its focus is to be a generic prototyping board rather than an embedded solution. Key examples of this are that the ATMega32U4 microcontroller is capable of running at 3.3v same as the Wifi module but the Arduino team chose to run it at 5v to give it maximum compatibility. This means that the board contains lots of level shifters on the board to interface between the two sides. If a custom board was made then this could run at 3.3v and eliminate these level shifters.
Also having the 3 boards stacked together with separate power board means that headers, connectors, spacers, struts are used for the board. So again merging all of these onto a single board would bring the costs down as many of these could be eliminated.
The Arduino Yún has already been mentioned, that is one of the highest cost items. A "Dragino HE" WiFi module is approx £15 and the is £2.34 when bought in quantities of 10 or above, a few basic components such as capacitors and a crystal are needed.
The Infineon RGB LED driver board is again and expensive component, it was used for the project as it provided a quick an easy approach to providing the output. However it is really over specified for the job and is being run right on the lowest level of it's capability. The approach here would be to take swap this out with a driver using discrete MOSFets to power each of the channels for the LED such as the as that's quite capable of being driven by the lower 3.3 supply. To standardise the bill of materials these could also be used to control the power to the servo and temperature sensor. Again the same MOSFet could be used to turn on and off the power to the WiFi module
The sign for the back of the cottage was made of plastic and most of it's cost was delivery charges rather than manufacturing costs. If the house was being produced in bulk then this could be swapped out with a sticker rather than a sign which is substantially cheaper.
The house itself is quite a large expense so this could be replaced with a hand made version, the wood could also be bought in larger sheets for greater savings.
The current power design incorporates a boost regulator to generate 5v from the 3.7v battery. This is then reduced back to 3.3v by two different regulators, one on the Infineon board and one on the Yún. By swapping to a 3.3v design this could be simplified to be a single regulator that would provide 3.3v. Assuming further savings in power can be achieved then the battery capacity could be reduced.
The magnets were surplus from a previous project so these could be swapped with cheaper ones for the same results.
In general SMD components are cheaper than through hole components and take up less space on the PCB, so this is an area where plenty of savings could be made.
One of the key improvements of swapping to a single PCB solution would be to remove the need for many connectors and small boards. For example by mounting the LED directly on the board the need for a daughter board, connectors and wires could be eliminated. Similarly by incorporating the power supply parts onto the main PCB further connectors and headers would be removed. The stacking headers needed for the Arduino shield approach would also be eliminated.
As mentioned above one method to reduce costs would be to scratch built the weather house, this would also allow the design to be simplified reducing the wood used and number of components. Rather than having a separate front and back these could be integrated together simplifying the design. Rather than hand cutting the parts these could be laser cut so that there was a single sheet with each of the parts that could be popped out. This would reduce assembly costs. Equally these could be pre-printed as per the original house saving on painting cost and
The 3D printer parts are currently screwed together, if the technique using clips could be made to work, or if these screws were swapped with plastic rivets then the assembly would be simpler and the time taken the make the parts could be easier.
The design incorporates an SDCard to hold some of the code, if the Linino side of the design was optimised then this card could be eliminated and hence component parts and set-up steps for that module would be reduced.
The Knock Knock sensor incorporates a variable resistor to "trim" the level detection. If a series of detectors was checked it might be possible to replace this with a simpler resistor network with a fixed level.
Estimated cost savings
Based on rough estimates and the changes above to the project components it should be possible to bring the cost down to approx £60, a saving of over £100 on the original bill of materials. No attempt has been made to calculate manufacturing costs but the above suggestions should bring those costs down too.