​​Solar power for off-grid streaming
​​‘Shopping list’ from LS 2023
​​
​​Hello again,
​​ 
​​Here is the shopping list of equipment we used for the solar set up (its Amazon but if you want to get the stuff elsewhere at least you’ll have the references).
​​ 
​​Best
​​Peter
​​ 
​​ 
​​ 
​​ 
​​ 
​​ 
​​ 
​​ 
​​ 
​​ 
​​ 
​​ 
​​
​​
​​
​​This is a working document based on the Cerro Pelón stream installation in State of México, which is an off-grid installation with a 3G router, Pi streambox and low power custom preamps.
​​
​​It aims to capture experiences, calculations, tests, links and develop a guide that can be used in other locations and situations
​​

​
​Cerro Pelón stream - microphone icon on the Locus Sonus soundmap - 18 Jan 2018 23:23 in London

​​

​
​Introduction

​​Rob Mackay took the equipment to Mexico in January 2018 and he and the team installed it at a site in the mountains within the Monarch Butterfly Biosphere Reserve at Cerro Pelón. 
​​
​​
​​Remote recordings made from UK have demonstrated that the stream can convey fine detail with a good ‘binaural’ stereo image that creates a strong sense of place.
​​
​​The stream has received a lot of interest from the network of those interested in the Monarch butterfly and its remarkable migration between sites in Mexico and northern North America, which spans 4 generations.
​​
​​It was possible for ornithologists  to identify various species of birds by listening remotely.
​​
​​

​
​Description of the installation

​​Our aim has been to set up a technical proof of concept in a relatively short period of time, to demonstrate feasibility of a long term off grid stream.
​​
​​The solar side has evolved as we adapted the installation. Eg we couldn’t use the 4G dongles we expected to be using (because the ISP doesn’t support them); so we added a 3/4G Huawei router, which has a quite high power consumption. We initially plannned for 2 streamboxes; but we have focused our resources and equipment on a single streambox for the time being.
​​

​
​Update - April 2018

​​In March, we installed a fully functioning off-grid system, which has run continuously in a number of weeks.
​​In the process of re-installing, the preamplifiers we were using developed a fault, which we are the process of trouble-shooting. This will bring the level and detail of the stream back up to its previous level.
​​We are currently working out how best to test and fix this part of the installation.
​​

​
​Components

​​
​​The streambox is in a weatherproof plastic ‘dribox’ with the microphone capsules mounted with grommets in opposite walls. 
​​
​
​The box contains:
​​
​​On the audio side:  
​​— a Raspberry Pi 3 with 
​​— Cirrus soundcard ‘hat’
​​— 2 x custom preamps
​​— noise reduction circuitry
​​— microphones
​​— audio cables
​​
​​On the power side: 
​​— a 10Amp charge controller
​​—  inputs to the charge controller from solar panel(s)
​​—  in / outputs to /from the controller to the battery (ies); 
​​— outputs to the load (audio equipment and router)
​​
​
​Externally there are:
​​
​​Solar:
​​— 1 x 20W solar panel (semi-flexible)
​​— 1 x 30W solar panel (semi-flexible_
​​
​
​In a large weatherproof box are:
​​
​​Batteries:
​​— 2 x lead acid deep discharge solar batteries, each with capacity of 115 Amp hours
​​

​
​Connections

​​
​​The solar panels are connected in parallel ie the outputs from the panels are both cabled directly into the charge controller. The power of the panels is additive when connected in parallel, so the total power = 50Watts.
​​The panels are rated at 12Volts. This remains unchanged in this configuration.
​​
​​The two batteries are connected in parallel. This leaves the voltage (12V) unchanged. It doubles the storage capacity, so the total = 220 Amp hours.
​​

​
​Calculating sizing of panels and batteries

​​
​​The size of the panels and batteries we are using was based on what was feasible to carry, the batteries available, and ready reckoning of the power requirements of the system.
​​
​​This ready reckoning was based on Grant’s experiences in the UK and various sources of information, especially this site, which give a useful breakdown of how to go about working out the energy requirements:
​​
​​

​
​Recipes / Kits

​​
​​Here we want to be able to recommend simple kits that somebody can go out and buy or assemble for a particular situation.
​​
​​Eg we want to anticipate the likely power consumption of a particular setup and recommend a size of panel and battery that will work in such and such a location.
​​
​​A list of tested and reliable components would be useful.
​​
​​It would also be really useful to have links to the online solar gain calculators mentioned by Benny below.
​​
​​
​​
​​
​​For reference, and until we’re sure what works best, our workings are shown here.
​​
​​The calculations need to be interpreted with caution (see Benny’s remarks).
​​
​​

​
​Calculations - bare bones

​​
​​
​​Panel sizing
​​
​​We are now running multiple devices:
​​
​​Raspberry Pi
​​Preamplifier
​​Router
​​
​​The fact that the streambox is dropping out and indicates the panel is undersized. (The current panel is only 20W) We could simply buy a quite big panel. But it makes sense to size it more thoughtfully, since smaller panels are both cheaper and more portable. 
​​
​​So we need to systematically work out:
​​
​​— the total load
​​— the hours of sun in the streaming site
​​
​​<) Load below
​​
​​
​​
​​Insolation
​​
​​Insolation calculations are based on the season of the year with least sun.
​​According to this map:
​​minimum hours of sun daily on average in our streaming location = 4 hours
​​According to this source:
​​The daily average is a little over 6 hours
​​
​​The daily power available is about 5.8kWh / sqm (on average).
​​
​​Again though, we need to size conservatively based on when there is likely to be a lot of cloud cover.
​​
​​Also, our figures are approximate, not based on a detailed local weather report.
​​
​​
​​Seasons
​​
​​
​​AVERAGE MONTHLY WEATHER IN Morelia, Mexico
  • ​​On average, the warmest month(s) are April and May. 
  • ​​Most rainfall (rainy season) is seen in June, July, August and September.
  • ​​Morelia has dry periods in January, February, March, April, November and December.
  • ​​On average, the warmest month is May. 
  • ​​On average, the coolest month is January. 
  • ​​July is the wettest month. This month should be avoided if you don't like too much rain.
  • ​​February is the driest month. 
  • ​​
​​
​​
​​Load calculations
​​
​​Pi 3
​​Determining ‘typical’ load of Pi 3:
​​Pi themselves quote 0.58A
​​Voltage nominally = 5V
​​So power - 5 x 0.58 = 2.9W
​​
​​For how to calculate, <)
​​
​​ Assume 80% efficiency of conversion by the 12V to 5V step-down regulator. 
​​So the draw by the Pi will be increased by 20% or 0.58
​​Total = 0.58 + 2.9 = 3.48 W
​​
​​At 12V, the current will be 3.48 / 12 = 0.29 A
​​So the Pi will draw 0.29 A every hour from the battery
​​Over 24 hours that will be 0.28 x 24 = 6.96 Ah
​​This is the amount the solar panel must supply each day, on average, for the Pi alone.
​​
​​From the map above, we can expect 4 hours of sun on average in the least favourable part of the year (presumably when it’s raining in June - September).
​​
​​So we have 4 hours to input 6.96 Ah each day. This means we need to charge the battery at a rate of 6.96 / 4 = 1.74Amps
​​At 12V this = 1.74 x 12 = 20.88 W 
​​So we need a solar panel rated at just over 20W  *just for the Pi*
​​
​​
​​Now looking at the other components:
​​
​​Router
​​
​​My Huawei router, which is very similar to the one in Mexico and has more ethernet ports so could be more power hungry, is rated at:
​​< 12 W (ie it uses less than 1A at 12V).
​​So the maximum power consumption is 12W = 12V x  1A
​​Here’s a more detailed report:
​​This is roughly consistent with the idea that the average is around 6W
​​So let’s take 6W as the working estimate
​​6W at 12V = 0.5A > 0.5Ah per hour x24 = 12Ah per 24hr period
​​With 4hrs of sun, we need to charge the battery at a rate of 12 / 4 = 3Amps
​​At 12V this = 3 x 12 = 36W *for the router*
​​
​​
​​Preamps
​​
​​The indvidual preamps are rated at:
​​Maximum power consumption = 10mA = 0.01A
​​There are 2, so this gives a total of 0.02A
​​Add 20% for losses: Total power usage = 0.024A
​​At 12V this = 0.024 x 12 = 0.048W
​​At 12V the current = 0.024A
​​This give 0.024 x 24 = 0.576 Ah every 24 hours
​​With 4 hours sun, we need to charge at a rate of 0.576 / 4 = 0.144Ammps
​​At 12V this = 0.144 x 12 = 1.728W *for the preamps*
​​
​​
​
​Summing up
​​
​
​Predicted solar panel input to operate the streambox (subject to testing and confirmation):
​​
​
​20.88 (for the Pi) + 36 (for the router) + 1.73 (for the preamps) = 58.61 W
​​
​
​So in principle we could use a 60W solar panel.
​​
​​Also in principle, it may be sufficient to use the 2 panels we have together (30 + 20W, assuming they are both working effectively, since the weather at this time of the year will generally give more than the minimum 4 hours of sun we have based these calculations on).
​​
​​Longer term, we might use eg 2 x 30 W panels, which would be portable and cost effective. Or we could use slightly bigger ones eg 2 x 40W  to have some extra capacity.
​​
​​
​
​Battery
​​
​​Based on the above, we will be drawing the following from the battery each day:
​​
​​Pi 6.96 Ah
​​Router  12Ah
​​Preamps   0.576 Ah 
​​
​​Total =   19.54Ah
​​
​​Following a suggestion to allow for 1 week (= 7 days)  without sun (in case of long term overcast weather, which may or may not be realistic in Morelia region; I need to check this further), this would amount to a total power draw without recharge of 19.54 x 7 = 137Ah
​​
​​The battery can not be discharged below 40%. So taking 137Ah as 60% of the total battery capacity, we need a battery of 137 / 0.6 = 228Ah
​​
​​This is a large and expensive battery. Note that the router (which was not designed for initially) is the main thing taking up the power.
​​
​​It will be worth doing some real world testing on the load before investing in such a big battery pack in my view. We may also want to consider if we can reduce the power consumption of the router, perhaps by turning off the wifi network (which uses a lot of the power I think) or at least *turning down the range / strength* (if this is an option). Also looking ahead, we can research *low power consumption routers* as alternatives for this application.
​​
​
​For now, we could usefully put our 2 batteries in parallel.
​​
​​I don’t actually know their capacity but perhaps it is around 80Ah, which would give us a total of 160Ah. This would be well on the way to the theoretical requirement and, again, could well work in this season.
​​
​​Benny: I think our batteries are rated at 115Ah (though the labeling is quite unclear) 
​​
​​

​
​Remarks on the calculations

​​
​​
​​— when calculating power consumption, you may also want to factor in the charge and discharge efficiency of the battery.  Rule of thumb for lead acid is about 70% each way.
​​
​​— energy generated by the panels is only at their rated power when the sun shines directly at 90 degrees to the panels. In effect, power is always reduced by the sine of the angle between the panels and sun. This complicates 'hours of sunshine' as it becomes 'hours of equivalent sunshine, at 90 degrees to panel'. I think the rule of thumb for UK was assume one hour per day, but mexico will be better!
​​
​​[
​​I was looking at map like this:
​​From that I was working to around 4 to 5 hours equivalent
​​]
​​
​​
​​The best way to check your assumptions here is to look at one of the many 'solar calculators' you can find online, wjich calculate output based on weather databases. They wont have a dataset for cerro pelon, but almost certainly will have data for mexico city if not morelia... input the direction, angle, rating, and location of the panel and see if the websites' predicted monthly output is close to your assumption.
​​
​
​>> Can you recommend a site or two to do this? 
​​
​​
​​Ps. For a really good guide to DIY solar installations (including sizing calculations) check out 
​​
​​
​​

​
​Issues to resolve

​​ 
​​One issue we have is that our (quite good quality) charge controller has a manual re-start. It disconnects the load to prevent the battery being overly discharged (to prevent damage to the battery). But it does not re-connect the load when the battery is sufficiently re-charged. This is a bad design for our system, where we have been running out of power. So we need to source a
​​
​
​Charge controller with auto load re-connect 
​​
​
​This charge controller is available in Mexico from this supplier:
​​
​​
​​I an ordering the slightly cheaper 10A version in the UK and plan to run it as an ongoing test stream to duplicate the Mexico situation.
​​
​​Benny has an alternative approach which does the same thing but allocates the 2 functions (battery regulation / protection and solar charge optimization) to 2 separate devices:
​​
​​Cheaper versions of this, adding a KEMO battery protector:
​​
​​This is the (low quality) charge controller I used to use with did solar panels. It auto reconnects loads. but for an enduring solution you probably want something better quality. 
​​
​​
​​one solution would be to keep same charge controller but use a battery guard for the loads, instead of connecting them bis the charge controller. This kemo one should work okay I think.
​​
​​
​​
​​Kemo products separate the 2 functions: battery and solar regulation, as here:
​​
​​
​​
​​
​​

​
​Miscellaneous links and info

​​
​​
​​
​​
​​

​
​More working out

​​
​​We have 2 x 115 Amp hour batteries = 230Ah
​​We can use 60% of this = 138Ah
​​
​​We have been using these figures (from above calcs) for the current we expect to draw *in one 24 hour day*:
​​
​​Pi 6.96 Ah
​​Router  12Ah
​​Preamps   0.576 Ah 
​​
​​Total =   19.54Ah
​​
​​In a week without recharge (ie totally overcast), the battery would be drained by 19.54 x 7 = 137Ah.
​​In reality, Benny suggests discharge is only 70% efficient, so the actual discharge would be more like 137 / 7   x 10 = 196Ah
​​We can only use 60% of our battery capacity 
​​So the total capacity required would be 196/6 x  10 = 326 Ah
​​
​​Our battery capacity is currently: 2 x 115 Amp hour batteries = 230Ah
​​This is well short.
​​Of course, the chances of a whole week with no sun may be low. Franco said the last 2 days have had no sun, and the 2 before that were overcast but with some sun. But before that was sunny. I think this is the sunny time of year.
​​
​​Also: the amount of power being used by the router is the main unknown. We have taken an average figure. If, for instance, that figure were double to the full rated capacity of the device (ie 12W instead of our previous estimate of 6W), which is not impossible (ie it would be operating at 12V and drawing a current of 1A), then the power consumption of the router would be double and our consumption figures *per day* would now be:
​​
​​Pi 6.96 Ah
​
​Router  24Ah
​​Preamps   0.576 Ah 
​​
​​Total =   31.5Ah
​​A big jump up
​​
​​Finally, the extra power usage by the router would double the amount of panel required from the panels to match it, from 36W to 64W just for the router
​​
​​This means that our total panel requirement would jump from 59W before to:
​​
​​20.88 (for the Pi) + 64 (for the router)+ 1.73 (for the preamps) = 87 W (to nearest whole number)
​​
​​So we would be looking at closer to a 100W system to give spare capacity.
​​
​​I am still to be convinced this is the real power consumption of the equipment we’re talking about, but it’s useful to be aware of the possibilities
​​
​​This emphasizes, as Benny suggested: we need to test / monitor the actual power consumption of the system.
​​
​​>> Grant to do some parallel testing in London using a close to identical setup with a Pi and a Huawei router, and check the power consumption.
​​
​​>> Find more detailed estimates of solar gain in the streaming site
​​
​​
​​
​​

​
​Next Steps - Working

​​
​​Notes from Rob and Grant’s skype on 22 FEB 2018
​​
​​
​​[22/02/2018, 13:03:31] Robert Mackay: Grant Skype 22nd Feb 2018:
​​
​​
​​Next steps:
​​
​​Buy equipment (do this now, have this up and running before I go out):
​​
​​Add a solar panel. Therefore have 3 panels in parallel. Take another one out? Buy another 50W panel (this would create a 100W system). This needs pricing. We should contact the people who sold us the solar batteries to see if they can sell us a 50W solar panel, and maybe a charge controller. Could get Franco to call them. Could be possible to buy these today and then send them in the bus. Maybe get working by the performance deadline?
​​
​​Change charge controller to have auto reboot. Grant shared the link to a charge controller available in Mexico. $100 approx.
​​
​​Ask Franco to go up the mountain before Huds performance. 
​​
​​We did have issue with the large solar panel at first. But Franco had been using the other panel to successfully recharge the battery. Everything was working well and when did installation. Is there maybe a loose connection somewhere still?
​​Is it possible there’s still something wrong with the TelCel contract? But this has been changed now. 
​​Franco back up the mountain on Friday? We need to sort out paying him. 
​​Router is distributing a local wifi network. This is a major draw on power. But might not be a convenience we can afford. Maybe we could change the router for one with two ethernet ports? That way can plug a laptop in rather than using WiFi.
​​Grant to test the Huawei router he has with wifi on and off to see power draw difference. 
​​Need to get local weather data from nearest weather station. Ask Pablo. 
​​Benny has rule of thumb calculation on 30% efficiency. 
​​Check how much still owing to Franco. 
​​Put out call to people in Canada and San Antonio, if people can chip in for data usage. Maybe kickstarter? Ongoing Open Source public resource. 
​​Website for project (or webpage). Doing html from scratch rather than WordPress. Could potentially have a PayPal link to contributions to the stream. 
​​Could we add a webcam to the stream. Initially we were looking at setting up 2 streams. Or could use light sensitive sonification software which converts the flickering light off the wings to sound. The guy at the pond in up state new York does that. Alternatively Grant’s son Max has worked with Raspberry Pi webcam to monitor cloud cover. Maybe look at this for second location. 
​​The Cumbria Wildlife Trust have a Seal Cam and they’ve been able to get donations. Look at this.
​​