Cell Characterization

I experimented with reducing the ESR of a single cell before trying to scale up to a full battery. I standardized on zinc and copper electrodes (zinc roofing strips with copper tape on the back side), and 1.8" circles of potato. I used a mandoline slicer to get consistent thicknesses, and a biscuit cutter to create circles from the potato cross-sections. Some of the experimental results are summarized below - the open circuit voltage was almost always 0.80V, so I am only reporting the progression of ESR here:

• 500 ohm - 3mm slice, no treatment

• 60 ohm - 3mm slice, salted - ESR increases significantly after a few seconds

• 990 ohm - 2mm slice, no treatment

• 750 ohm - 2mm slice, boiled 3 minutes in tap water

• 115 ohm - 2mm slice, boiled 3 minutes in salted water - fragile

• 75 ohm - 2mm slice, boiled 6 minutes in salted water - very fragile

• 95 ohm - 5mm slice, boiled 6 minutes in salted water

• 80 ohm - 9mm slice, boiled 6 minutes in salted water

• 40 ohm - 9mm slice, microwaved 30 seconds - good performance when hot, ESR increased after potato cools down

• 1300 ohm - 6mm slice, microwaved 30 seconds, stored in saltwater for 2 hours (estimated time from home to summit)

• 160 ohm - 6mm slice, microwaved 30 seconds, stored between dry paper towels for 2 hours

• 130 ohm - 3mm slice, microwaved 1 minute, stored between dry paper towels for 2 hours

• 100 ohm - mashed potatoes - very messy

I settled on 3mm thick slices microwaved for 1 minute, as a balance between good ESR and making a mess in the kitchen.

A study on boiling potatoes is widely cited (e.g. this article). I suspect that similar to boiling, microwaving breaks down the structure of the potato and allows for better electron flow. I don't have an explanation for why the 2 hour saltwater soak made performance worse.

Module Characterization

Cells were combined into modules with an open circuit voltage of roughly 12.8V. I ran the below discharge curve over a 12 hour period with a 1kOhm resistor connected (higher resistance would be better), and assume the module would have continued running beyond this arbitrary 12 hour period. The initial spike in voltage represents a sag in voltage over the first few seconds of loading, and I assume the shape in the middle is related to the module falling over on my desk during the test.

The following calculations assume that the module's voltage went to zero immediately after the data stopped capturing. There was likely more energy in the potato, but a microwaved potato becomes unpleasant after sitting at room temperature for 12 hours...

The module has an energy density of approximately 1 Wh/kg, which is less bad than I expected at the beginning of the project. An AA battery is roughly 300 Wh/kg.

My local source for small quantities of potatoes is $1.50/lb, which puts the potato battery at $1.8/Wh. An AA battery is roughly $0.3/Wh, so again the potato battery is not so many orders of magnitude worse than an AA battery. This ignores electrode cost (which could be reused between batteries after sanding down the zinc), and the roughly $0.10 of grid energy I put into microwaving the potatoes in each module.

Assuming a typical 160 calorie potato (160 kilocalories) and a human's ~20% efficiency turning food into mechanical work, it would be more efficient if I ate the potato and pedaled a generator.

I paralleled 4x modules of this same design for the activation.

The Rig

To keep the scope of this project manageable, I used my KX-2 as the receiver and connected it to a homebrew transmitter and TX/RX relay assembly. There was a battery for switching the TX/RX relays, but 100% of the energy radiating from the antenna came from the potatoes that were on the shelf at Trader Joe's just 24 hours earlier.

The VFO has a 3.3V regulator in order to keep a constant power supply for the oscillator. It is a straightforward Clapp oscillator with varactor + potentiometer control, and sits in the open air (more on this later...).

The VFO drives a buffer, then a Class E power amplifier tuned for approximately 100mW at 10V input. Output power drops closer to 50mW after TX starts. A tactile switch in series with the source of the power FET acts as the key for the rig. Finally, there is a low pass filter.

I used my QRP Power Board between the potato battery and the transmitter. By reconfiguring the buck-boost control IC to operate in burst mode and using its precision enable line, the buck-boost keeps the input voltage regulated to roughly 6V using hysteretic control. The controller is "sleeping" when the input voltage is lower than the enable threshold, so the overall efficiency of the system is relatively high. This 6V input regulation point corresponds to the maximum power point of the potato module, meaning the bulk capacitance on the transmitter input charges as quickly as possible.

The below oscilloscope trace illustrates the input voltage (blue) and output voltage (yellow) of the buck-boost converter when fed from a high impedance source and charging a large output capacitance. The output voltage is nearly at the 10V regulation point at the left-hand side of this trace, just to illustrate the burst-mode behavior with a high impedance source.

The Activation (Coyote Peak, W6/NC-399)

The at-home prep went well, the drive was quick, the hike was pleasant, and the battery assembly on the summit was faster than expected. I put up a 40m EFHW on a 7.2m fiberglass fishing pole and confirmed acceptable SWR. I started calling CQ around 19:50 UTC and only had a little bit of success. Jeff (AA6XA) was on the lookout for my weak signal, but we didn't have any luck at first. I called Jeff with 500mW from the KX2 and he gave me a 339 report, so I knew the ~10x lower power of the potato rig was definitely going to be a challenge - but the antenna was at least working. Eventually he heard me from the potato rig and it turns out the VFO had much worse drift than I had expected; he reported that I drifted through his passband in the time it took to send my call sign.

I was able to improve the drift a little bit by preventing direct sunlight from hitting the VFO. Then, I found that preventing board flex helped stabilize the frequency. And then, creating a shield with blue tape helped even more on this relatively windy day. Eventually I got spotted via the Reverse Beacon Network (below), and Cody (KJ7SYX) was able to see my signal on the Half Moon Bay SDR roughly 40 miles away.

More or less, I failed to activate the peak with the potato and my homebrew transmitter. I attribute this failure to VFO drift and low power, but the potato power system worked as planned! I gave up on the 40m potato rig after a couple hours and made some quick 20m CW contacts to claim my 1 point for the day.

Many thanks to the supportive hams who tried to help me make a contact under potato power.