I'm testing Magnesium-Copper (Mg-Cu) and Zinc-Copper (Zn-Cu) galvanic cells in a 0.16M solution of Hydrochloric acid (HCL). This solution of HCL approximately matches the PH and HCL concentration in the human stomach (though conditions in the stomach can vary). The Mg-Cu cell initially produces a voltage (electric potential) of 1.46V, however, the surfaces of the Mg strips rapidly corrode and the voltage plateaus at 1.28V after one minute.
The Zn-Cu test is a little more complex. I had already tested the Zn-Cu cell before I started shooting this video. The residual HCL in the thread and plastic mesh surrounding the Zn and Cu is functioning as an electrolyte. This is why there is a significant electric potential before the cell is even submerged in HCL. When the cell is submerged, the electric potential drops to about 0.5V. Why? This particular Zn-Cu cell is actually three Zn-Cu cells in series. I have separated the cells with plastic flaps which are supposed to act as ion barriers (essentially they insulate the cells from one another). The plastic flaps function well enough when the cell has residual HCL on it, but when the cells are submerged they short across one another.
Electric potential of Zn-Cu cell bank at various functional levels:
1.45V: Ion barriers working, three Zn-Cu cells in series.
0.76V: Center cells shorted, combined cells acting as a single cell.
From a practical perspective, designing a galvanic cell (or bank of cells) is extremely easy. They can be treated like batteries because they are like batteries. For the purpose of this log (and project in general) I am going to describe everything in plain language instead of scientific terminology. If you want to deep dive into electrochemistry Google is your friend.
1) Pick anodes and cathodes that maximize difference in reactivity. The metals (or the like) you choose will determine the voltage produced by your cell. To determine what voltage your anode and cathode will produce, consult a reactivity table like below:
You want to pair a reactive metal (negative electric potential in the chart) with an unreactive metal (positive electric potential in the chart). The maximum voltage produced by your galvanic cell will the sum of reactivity.
For a copper-zinc cell the ideal voltage will be 0.34V(Cu) + 0.76V(Zn) = 1V
Lithium is the most reactive metal easily incorporated into a battery, which is why it produces such high power density batteries.
2) Pick a highly acidic electrolyte. The actual voltage produced by a galvanic cell will never equal its ideal voltage. How close you get depends largely on how good your electrolyte is. The role of the electrolyte is to transport ions. In general, strong acids are the best electrolytes because, in general, they can pack the most ions.
3) Maximize the surface area (and probably mass) of your annode and cathode to maximize available current. Although the voltage of your cell is determined by the reactivity of the annode and cathode, the current supplied by the cell is determined by the surface area of the annode and cathode exposed to the electrolyte. You want an electrolyte that can transport as many ions as possible, but this is still limited by the availability of ions (annode/Zn) and places to put ions (cathode/Cu). More ions means more current.
4) Although voltage of a single cell is limited by annode/cathode reactivity, voltage can be increased by placing cells in parallel - just like batteries. However, cells have to be isolated otherwise you will get a short between cells. To put it more specifically, an ion barrier between cells is necessary if they are adjescent in an electrolyte.
Pretty cool stuff.
In all the cells I've built I have sandwiched multiple strips of metal for each anode and cathode (in the above picture you can see how I've sandwiched Magnesium and Copper) in an attempt to maximize surface area, and therefore maximize available current.
Now that the smartpill is wired up on a breadboard, it’s time create an app so we can log data wirelessly (and eventually through the abdomen). The big question is how to do this using Bluetooth BLE, a widely available standard that is not geared towards intermittent split second connectivity. Just connecting (“pairing”) over BLE can take seconds, while we want to turn the smartpill’s radio on for a fraction of a second. Luckily there is a way to transmit data over BLE without a connection: GAP advertisement information. Every BLE device constantly transmits basic information about itself – this is what you see when you scan for available Bluetooth devices. Only part of this information (“UUID”) is necessary, the spelled out name (“macbookabc”, “fitbit123”, “smartpillxyz” etc.) is for user convenience. Instead of using this extra ‘name’ information to advertise the smartpill for potential pairing, I use it to transmit data. Every time the smartpill turns on for a fraction of a second, it changes it’s BLE ‘name’ to the latest sensor values (for now voltage sensing ADC). My app is constantly scanning for new BLE devices (GAP advertisement information). The smartpill doesn’t have to wait for a connection to be created, the app grabs the sensor values from the advertised ‘name’ information the moment the smartpill is detected by the app. This dramatically decreases the length of time the smartpill’s Bluetooth radio has to be turned on which is crucial to minimizing power consumption.
I have used the Cordova based Evothings platform to create a hybrid Android/iOS app using web dev style Javascript. Evothings is, bar none, the easiest way to create apps for BLE devices. Anyone who knows a little JS/HTML/CS can pick it up quickly. Evothings sits on top of the venerable Cordova hybrid app creation framework. I have placed the Evothings project folder for this data logging app in the smartpill GitHub repository.
Accelerometer: Kionix KX022-1020 using SPI interface
Heart Monitor: PixArt PAH8001 green LED PPG
Battery: 40 mAh lithium polymer
Waterproof: IP67
Device size: 18.0*11.2mm
This is a high resolution microscope image of both sides of the main board aligned so that pins and traces can easily be mapped. I currently use this for reference purposes when using the hacked M3 as a development platform.
Curt White
