Solar powered robot - Charging suggestions requested.

AndyGadget

Senior Member
What would be the best battery chemistry and charge technique for a solar powered device requiring a battery of around 5Ah at a voltage in the range 7V to 12V which would be solar charged in the Gloucestershire (UK) sunshine? This will be for a small (about a foot square) semi-autonomous garden robot.
Li-ion, LiPo, NiMh all have their charging quirks and like to be properly cycled without overcharging. The fickle nature of sunshine is obviously a problem here.
Something I've never used but I know is a favourite around here is LiFePo cells, although they get pricy for larger capacities but do appear to have a simpler charge regime.
I've been running a solar powered Picaxe pond pump controller for aeration, circulation and filling for a couple of years and 12V 14Ah lead/acid was a no-brainer here but the weight is a major factor against for the robot.

Andy.
 

AllyCat

Senior Member
Hi,

I'm not an expert, but my recommendations: A ballpark figure for the average solar radiation is around 700 watts/square metre (at right angles to the sun's rays) so say 70 watts/square foot. Best panel efficiency is around 20% so perhaps 14 watts output, but let's assume 10 watts, or around 1 Amp at 10 volts overall.

Certainly LiFePO4 cells are "intended" for multiple-cycle Traction applications, but do have around 20% less energy density, nominal 3.2 volts (so you'd need three in series) and a higher price. Therefore, I'd be inclined to go for two (or three) Lithium Ion/Poly cells in series. I don't believe the charging regime is too arduous*, basically constant current (which will be limited by the sun) until 4.2 volts is reached. Don't allow them to discharge below say 3.5 volts. *Outdoors they present less of a fire/safety hazard. :)

Dedicated Lithium Charge / Protection modules are readily and cheaply available now (or roll-your-own with a PICaxe), so my primary recommendation is to ensure that you use an individual control(ler) for each cell. Maybe "cannibalise" a Laptop Battery, "Power Bank" or plug-in battery of a power tool for the cells?

Cheers, Alan.
 

AndyGadget

Senior Member
Thanks for the thoughts, Alan.

I hadn't fully considered LiPo as it's a few years since I first read up on them and accepted wisdom at the time appeared to be 'charge at 1C - nothing else', but I see now that slow-charging is fairly popular and may extend the cell life. I was also expecting to be using a solar panel giving 12V nominal and regulate down but it may be better to use say a 6V panel and use a boost converter set to output 8.4V which would give a taper charge to a 2S LiPo.

I think this setup would allow charging in cloudier conditions than the step-down approach where much of the time the panel output would be below the regulator voltage.

The main controller will be a 28x2 but I'll also be using a secondary Picaxe as an I2C slave to parse NMEA strings from the GPS and enable the 28x2 to concentrate on the main functions, with the 28x2 requesting just the data it needs. It would be simple enough to use this to also monitor the battery charge and balance.

Spot-on with your solar radiation estimate by the way. My pond controller with around a 12" * 15" panel pushes 1.2A into the 12V batteries at full sun.

Andy.
 

premelec

Senior Member
Is weight a factor for your battery needs? lead acid is still cheaper and heavier... more tolerant of charge and discharge - however with all the lithium cell use good battery management systems are available - more information on how much power you need compared to how much is available would be useful to estimate your battery... there are listings of BMS units for lithium on Ebay and such... what is your peak ampere draw? Up converters usually have more losses than down converters - assuming robot speed is not a big factor I'd use as little battery current as works ok. Have fun!
 

papaof2

Senior Member
My preferred method of choosing what solar panel(s) is to first go to http://www.solarelectricityhandbook.com/solar-irradiance.aspx and enter the exact location to get the available sun hours by month and panel tilt. Seasonal differences matter for some things and 3 hours in December versus 5 hours in May is a major factor in planning (I plan for the full year but the configuration most support December although you might only be planning for 6 or 7 months).

Then I determine how much the electrical load will be and for how long on the worst day. I would consider breadboarding the robot without batteries to be able to determine what the total AH needed is for the fully loaded robot, using random items to represent the weight of the different batteries and the solar panel(s). Having the total ampere hour (AH) need, you can then select the appropriate battery size (based on chemistry and price).

I like Excel and I usually make up a spreadsheet listing the devices that make up the load and an estimated number of hours (or fraction of an hour) each device will be in use on a given day. Then you can compute the total amps and hours and size the battery's AH to properly handle that - no more than 50% discharge for AGM, no more than 80% for lithium (although some would not take lithium down to just 20% SOC). Example: I have a garden shed that's used only a few times a week, generally 20 minutes to take things out for use and another 20 minutes to put things back. Based on two days a week use in winter and threes days/week in summer, a 15AH AGM battery is adequate for the 1.78amps of LED strip lighting in the shed and a 30 watt solar panel on a PWM charge controller can keep up with that usage. The PWM controller I'm using has adjustable parameters and one of those is the load shutoff voltage. If the battery is below 50% SOC, the load is turned off to protect the battery. Along with that, the power switch is a spring wound timer so the lights can't be on for more than an hour without manual intervention. My design is to have lighting at the flip of a switch and for it to operate as reliably as commercial AC power - OK, I'm a perfectionist, but the shed lights have been working fine for months. http://www.jecarter.us/files/My-12Volt-Solar-Shed.xls

If you need to drop the voltage from the solar panel to the battery, a buck converter is almost always more efficient than a linear regulator but a little research might find a solar charge controller that would work for your configuration of lithium cells - a few of those are even fully adjustable for min/max voltage, etc.
 

AndyGadget

Senior Member
Weight is a factor, that's why I ruled out lead acid, even though it would be best suited to the solar charging.
I'm using worm geared 6V motors which flat out will move it at around 1 metre in 5 seconds with the wheels I'm using. It will have 4 wheel drive and each motor takes approx 50mA off-load, 180mA high load.

Getting a bit over-complex for me there Papaof2 :) but the website and spreadsheet you link are very interesting. I'm basing this on my solar charging experience with the pond pump controller I mentioned above.

In several of his SF books, P.K. Dick mentions in passing artificial (robotic) creatures which have ended up roaming the wastelands and the concept has always appealed to me. This is just intended to amble around the garden, it's main behaviour being to bask (recharge) in previously remembered sunny spots based on time of day, GPS reading and UV sensor, only exploring further when it has the energy. It will also have PIR detection and will interact with warm bodies, although I feel that following The Cat around and kicking him out of his warm spots might just be a bit unkind. Moisture detection is in the plan as well with a 'hutch' to lurk in on wet days. (The hutch could possibly be a recharging station also, powered by a lead / acid battery.)


Andy.
 
Last edited:

AllyCat

Senior Member
Hi,

PV panels are fundamentally "constant voltage" generators (each "pane" is a forward-biassed slicon diode around 600 mV, all connected in series) where the available current is proportional to the "light level" (solar radiation). The trick is to draw as much curent as possible without the terminal voltage falling significantly (remember power = volts * amps). Switching-mode converters or power supplies can trade off an excess of input voltage for a greater output current, rather like a (variable wire-wound) transformer or "Variac" (old technology) The clever / difficult bit is to get the duty cycle correct so that maximum power is being extracted at all light levels.

As others have said,"buck" (down-) converters are generally more efficient than linear or "boost" (up-) converters, so I would stay with a "12 volt" panel. This will also give an easier configuration if you want to charge and discharge the batteries at the same time (which I guess you do). But you shouldn't use simple series voltage regulators for charge control, because one cell might need charging whilst others are already "full". Commercial charge-balancing modules are available or simple shunt regulators may be adequate. In the case of PV panels, there's little merit in "winding down" the load current because the unused power just stays in the panel and heats it up (where else can it go?).

Just a couple more general points: Getting GPS accuracy good enough to navigate (consistently) around an area as small as a garden might be quite a challenge. Also, PV panels are primarily sensitive to Infra Red and Visible Light (about 50% each) not UV. But the VEML6075 (I2C) "UV sensor" is quite interesting as it has separate measurement channels for UVB (skin-damaging), UVA (burning), Visible (light) and IR (heat) wavelengths.

Cheers, Alan.
 

westaust55

Moderator
Solar PV panels like other power supplies only generate the current required by their connected load.
Overload and for at least most PV panels the short circuit current is only around 5% more than the rates full load current.
Hence while there is wasted opportunity if the max power is not used, there is no current circulating internally to heat up the panels.

Heating by solar irradiation or ambient temperature above 25 degC will reduce the output power.

Rule of thumb is the have the panels facing towards the equator and inclined to the same angle above horizontal as you are degrees from the equator. So park the robot facing the equator with an included panel.
For small installations the power to operate sun trackers tends to negate the extra power produced.
 

AllyCat

Senior Member
Hi,
there is no current circulating internally to heat up the panels.
Maybe not, but there is (say) 700 watts of energy striking each square metre of panel (or lots more in Australia), which must go somewhere (for a "steady state" system), due to the Law of Conservation of Energy. Some energy (light) will be reflected from the surface of the panel (constant except with angle of incidence for a given panel) and some will be converted (or remain) as heat, which is why the panels are usually mounted with a convection air flow behind.

Then, normally around 20% of the energy can be "exported" as electricity, but if it isn't, then it must be converted to another form of energy. Light, Sound, Mechanical (e.g. expansion against a force), Chemical (converting one material into another), Nuclear (e = mC2) and Electromagnetic (i.e. radio waves), etc. are all "unlikely" (and undesirable ! ), which AFAIK leaves only Thermal (heat). Thus the panel must heat up to a higher temperature, to dispose of the excess energy by either thermal conduction/convection or "black body" radiation. Or did I misunderstand my Physics lecturers. ;)

Cheers, Alan.
 

papaof2

Senior Member
Hi,

PV panels are fundamentally "constant voltage" generators (each "pane" is a forward-biassed slicon diode around 600 mV, all connected in series) where the available current is proportional to the "light level" (solar radiation). The trick is to draw as much curent as possible without the terminal voltage falling significantly (remember power = volts * amps). Switching-mode converters or power supplies can trade off an excess of input voltage for a greater output current, rather like a (variable wire-wound) transformer or "Variac" (old technology) The clever / difficult bit is to get the duty cycle correct so that maximum power is being extracted at all light levels.

As others have said,"buck" (down-) converters are generally more efficient than linear or "boost" (up-) converters, so I would stay with a "12 volt" panel. This will also give an easier configuration if you want to charge and discharge the batteries at the same time (which I guess you do). But you shouldn't use simple series voltage regulators for charge control, because one cell might need charging whilst others are already "full". Commercial charge-balancing modules are available or simple shunt regulators may be adequate. In the case of PV panels, there's little merit in "winding down" the load current because the unused power just stays in the panel and heats it up (where else can it go?).

Just a couple more general points: Getting GPS accuracy good enough to navigate (consistently) around an area as small as a garden might be quite a challenge. Also, PV panels are primarily sensitive to Infra Red and Visible Light (about 50% each) not UV. But the VEML6075 (I2C) "UV sensor" is quite interesting as it has separate measurement channels for UVB (skin-damaging), UVA (burning), Visible (light) and IR (heat) wavelengths.

Cheers, Alan.
Most consumer GPS units are not high resolution (1 meter or less). While there are some VERY accurate positioning devices (the ones used to steer tractors so all the plowed rows are straight and equally spaced), those are also VERY expensive. For getting the location in a garden, there's another post about the possibility of using ultrasonics to triangulate a position in a small area. That might be something to follow up on. Of course, you could build your own "active grid" system with buried wire every meter or so and each wire having a different pulse frequency or pulse spacing. Add in an odometer of some type on your robot and you could compute that you're 0.2 meter north and 0.3 meter west of the intersection of east-west grid wire 4 and north-south grid wire 7. A quick lookup in the EEPROM (or some additional calculations) would tell you exactly where the robot is - and be able to compute the course back to shelter and a recharge.
 

premelec

Senior Member
FWIW there have been PV designs which use a cooling fluid [sometimes air] to use the thermal excess for heating water or house air etc and keep the panels cooler so they then have better efficiency.
 

westaust55

Moderator
Then, normally around 20% of the energy can be "exported" as electricity, but if it isn't, then it must be converted to another form of energy.. . . which AFAIK leaves only Thermal (heat). Thus the panel must heat up to a higher temperature, to dispose of the excess energy by either thermal conduction/convection or "black body" radiation. Or did I misunderstand my Physics lecturers. ;)
Cheers, Alan.
Hi Alan,
Yes, your thermodynamics/equilibrium is good :). The solar panels heat up and radiates the excess energy back into the air as best they can else warm up as some heat is retained due to increasing ambient/surrounding air temperature (until the air later cools).
At the PV panels heat up there efficiency drops seen as reduction in output voltage.
Often quoted is a roughly 0.5% reduction for every °C rise above 25 °C.

As premelec has stated, some installations used forced air of cooling water adjacent to the panel to reduce the temperature.
The idea of putting the heating into a hot water system is good but on that stinking hot summers day (40 °C and above) I think I prefer to avoid further warming the house here downunder.

However other than noting that there is no excess current flowing above that used by the charger/load, not much else towards improving PV efficiency can be economically achieved by the OP for his garden robot other than consider the type of charger as has been mentioned.
 
Top