Down Mt Kenya and into the office: picoBRCK learnings.

We’ve been back from Mt Kenya for a week now.  Although many expedition posts are about the exhilarating adventures we have had, the reason we continue to do these trips is because of the field learnings that come about about; some are expected some are unexpected.

Batteries are the weakest link.  

This was the first BRCK expedition that was human and not car powered.  That meant not only did we have to carry everything ourselves but we also had to find ways to charge all our gizmos along the way; we did not have a 12V socket anywhere nearby.  This meant that throughout the trip most of us had solar panels strapped to the top of our packs.  Luckily the equatorial sun at that altitude is quite strong (ALL of us got sun burnt on this trip) and so the solar panels worked fairly well though we did discover which panels perform best.

Trying to catch the last rays of sun.

Additionally, from the drone to the phones, from the Go-Pro cameras to the picoBRCKs almost all our gear operated off of lithium ion batteries; only the weather station used a lead acid battery.  A review of battery technologies will reveal that although lithium based batteries have many advantages, one of their significant shortcomings is their poor performance at low temperatures.  The temperatures on Mt Kenya regularly get below freezing and so we were continually packed our phones and battery packs under our jackets in an attempt to keep them warm.  Nevertheless, all of us experienced watching our phones go from 20-30% charge to dead within a minute as the phone tried to predict the charge on the battery in cold conditions.

All things considered, I think we did fairly well keeping everything powered.  However, it emphasizes the unique challenges of the IOT market we are targeting.  Our IOT environment is not inside the home or industrial facility.  It is not even inside an urban area; picoBRCKs are and will be scattered in some pretty harsh environments and that means that we need to make sure our power source is up to the challenge.

GSM is NOT an IOT backbone.

We are stuck on this one.  All our current picoBRCK customers want to take advantage of the incredible cellular coverage that we have in emerging markets.  However, cellular communications is intensely power hungry.  In fact the GSMA specification itself actual specifies that the equipment should produce a 2A pulse for over 1mS.  In the world of IOT, where a few milli-amps is significant and things are usually measured in micro-amps, 2 amps is huge.  It affects the size and chemistry of battery we can use (which is no small issue, but too technical for this post), it affects the type and quality of FETs that we can use in the power chain as we try to deliver and then turn off power to the modem in the picoBRCK, and of course it affects the longevity of the battery between charges and the requirements of the charging system.

the picoBRCK circuit board: A full 25% of the board is taken up by the power & battery management circuitry, much of which is needed to power the SIM800 cellular module (top left). The empty space in the top right is reserved for the LoRa module.

In the short term we will stick with GSM as our communication backhaul for picoBRCK and overcome those inherent challenges.  But the one technology that is popping up over and over is LoRa.  Although it has similar range capabilities as cellular communication, it achieves these distances with much less power.  We still have much to learn about this technology but the picoBRCK is already LoRa capable and I look forward to testing it out.

Another benefit to LoRa is that it is free while each GSM based device requires a SIM card with credit that needs to be monitored and topped up.  The actual cost of the data here in Kenya is fairly reasonable but on top of this is the logistical and human costs associated with managing that aspect of the infrastructure.  There are many services that will do this for you of course but these have their own expenses which prevents any IOT solution from truly scaling in this environment.

Anecdote

Our time on Mt Kenya illustrated both these issues beautifully.  We mostly had network connection around and near Point Lenana.  However, the situation with Fender unfolded mostly on the slopes of the mountain so as soon as we could see the lights of the surrounding towns I tried to start making arrangements and informing people about the situation.  At this point my main phone ran out of battery and I had no credit on my secondary phone.  It took some “Please call me requests” and other fiddling while we were hiking down in the dark to communicate with people in Nairobi.  This is not the first time that I have been in situation where either power and credit have failed. Any option to eliminate one of these potential failure points, such as by using LoRa instead of GSM and simplifying the power management circuitry, will greatly increase the overall reliability and up-time of an IOT ecosystem.