I'm working on a radio sensor (temperature, humidity, etc) which will be powered from a super capacitor and a solar panel.
The solar panel can generate up to 5.7V
The super capacitor limit is 5.5V, but it's better to use lower voltage, e.g. 4.7V.
So required voltage range for my device is 1.8V - 4.7V.
Question: what is the most efficient way to limit voltage from solar panel to 4.7V?
What I've tried:
Zener diode.
I'm not sure, maybe I've selected wrong component (1N4732A) but with this diode voltage drop was about 1V. It is bad.
MOSFET + transistor.
I can measure voltage by the controller and switch off the solar panel when voltage > 4.7V using MOSFET and transistor.
It works but the solution looks a bit overcomplicated and I could have a problem in case of controller hangs (I use watchdog, but who knows...).
A single silicon diode forward voltage drop is about 0.7 V, which would bring the solar panel voltage to a safe value for the supercap (providing that there is always some load on the system).
In any case, the diode is necessary to prevent the panel from discharging the cap at night.
jremington, I don't think that this is the best choice. In a cloudy day when the solar panel generates let say 3V I will waste 0.7V.
To prevent discharging I use a schottky diode.
pwillard, yes.
Also I did a small experiment. If solar panel generates 4.57V (without load):
Without zener diode it can charge supercap to ~4.36V (maybe more, but slowly)
With zener diode only to 3.56V
Did you connect the zener across the solar panel, not across the supercap.
You should use a 5v6 zener and a normal diode (1N4148), or no zener at all.
Schottky diodes leak more, and that's ok for a solar/battery system, but it might not be ok for a smaller capacity supercap.
Leo..
In a cloudy day when the solar panel generates let say 3V I will waste 0.7V.
You have a point, but it is exaggerated. Schottky diodes have forward voltage drop in the range of 0.15 - 0.45V at 1 mA, also "wasting" voltage. As pointed out above, they have relatively high reverse leakage current compared to silicon diodes.
PV cells are best treated as current sources, not voltage sources. The open circuit voltage depends only very weakly on the light intensity. You may find that on such a cloudy day, the current generated is too small to be useful, so consider doing a little experimentation with the target circuitry before deciding on a final design.
Working voltage (MPP) of a solar cell is about 2/3 of it's max open circuit voltage.
So if you want to charge a 5volt cap under any light condition, you should have a cell with an open circuit voltage of ~7.5volt, plus the forward voltage of the backflow diode.
Me thinks your 5.7volt solar cell is made to charge a 3.7volt LiPo battery (4.2volt max).
Leo..
First of all thank you for such an active discussion.
I completely agree that schottky diode have to be moved, it was just a quick test with a stable light.
Also I got the point about schottky diode disadvantages (reverse leakage current).
I use 95x95mm 0.77W 5.5V solar panel however, I would like to try smaller one, e.g. 53x30mm 0.15W 5V (haven't arrived yet). Maybe with that SP voltage limitation won't be required, but I have to test will it generate enough power or not.
Module power consumption:
sleep - 13µA
run - hard to measure, but theoretically 3-13mA (during data transmission)
With 1.5F supercup charged to 4.7V it will definitely work more then 12 hours (tested).
I will test how a single silicon diode behaves in my schematic and get back to you with results.
13 uA seems pretty high for sleep and I suspect you could do much better.
AVR processors draw about 100 nA in power down mode, or 6.5 uA with just the watchdog timer running, and if peripherals like temperature sensors are turned off, they won't draw anything. See https://www.gammon.com.au/power
Note that the reverse leakage of the 1N4732A is 10 uA at 1V.
Sorry for the late response, it took some time to get all components for testing.
I've ordered a few solar panels:
53x30mm, 5V 0.15W (30 mA)
68x36mm, 5V, 0.3W (60 mA)
100x28mm, 5,5V, 0.22W (40 mA)
plenty of zener diodes for different voltage and BAT54 schottky diode.
After some testing I decided to use 5,5V, 0.22W solar panel, 1N4733A (5.1V) zenner diode to limit voltage and BAT54 schottky diode to prevent discharging.
With 1N4733A I got voltage drop ~0.1V (without load), it's appropriate as for me.
BAT54 has reverse leakage current 2μA at 25V, it also fits my requirements.
Schottky diodes have very rapid increase of leakage with increasing temperature, unlike standard pn diodes,
you can get leakage upto milliamps at the top of the temperature range - this means they are usually unsuitable for very high reverse voltage situations where self-heating leads to increased leakage leads to increased heating, vicious circle takes over and thermal-runaway destroys the diode.
However that's not relevant at this voltage, but is worth bearing in mind before bunging a schottky into any
old circuit.
The problem you will face is that when the cap is discharged it will never let the panel produce power.
The cap will demand a HUGE amount of current from the panel which will drop it's voltage down too low to charge the cap at all.
It might work a little, but it certainly won't be efficient at all.
So if the cap is down to a voltage which kills your project, say 2V, the cap will pull the panel voltage down and keep it down at 2V. The panel might be in direct sun and able to produce 0.2W, but not at 2V, it might produce 0.02W.
As a minimum you will need some kind of PWM controller to allow the panel to be operating at a higher voltage than the discharged cap.
If you are already using an arduino there are example circuits for making a PWM charge controller out there.
The cap will demand a HUGE amount of current from the panel which will drop it's voltage down too low to charge the cap at all.
You are confused, and need to do some reading and/or experimentation. PV panels are best modeled as current sources, not voltage sources.
If charging a supercap that is initially uncharged, the panel will supply the short circuit current, charging the supercap at the maximum possible rate. The supercap will eventually charge up to the PV open circuit voltage.
jremington:
You are confused, and need to do some reading and/or experimentation. PV panels are best modeled as current sources, not voltage sources.
If charging a supercap that is initially uncharged, the panel will supply the short circuit current, charging the supercap at the maximum possible rate. The supercap will eventually charge up to the PV open circuit voltage.
Circuit model for a PV cell:
Yes, but current without voltage is fairly useless. 150Amps @ 0V = 0 W, 0 J/s, 0 Charge/s, 0 Energy
You can't see them purely as a current source if you want any "power" out of them. You have to select the voltage at with they produce the most power, which varies with conditions and light level.
Granted we are talking theoretically here and even a few joules flowing will, as you describe begin to raise the voltage of the cap. However it still stands that the panel at a low voltage, regardless of the current less Joules of energy will flow out of the panel and thus it's being used in efficiently.
If what you were saying was purely and practically true I could connect my 50W panel directly to a flat SLA battery @ 9V and expect it to charge it at around 2.5A (24.5W). Which it might, but if I PWM or buck the voltage down from it's maximum power point of 17V it will charge the battery with 50W and the battery will charge faster.
You can't see them purely as a current source if you want any "power" out of them.
Correct, but the panel will still charge the capacitor at the maximum possible rate. When the capacitor is charged, you can then draw power as needed. That is how these tiny, low power sensors are designed and intended to work.
jremington:
Correct, but the panel will still charge the capacitor at the maximum possible rate. When the capacitor is charged, you can then draw power as needed. That is how these tiny, low power sensors are designed and intended to work.
Yes, but current without voltage is fairly useless. 150Amps @ 0V = 0 W, 0 J/s, 0 Charge/s, 0 Energy
That is only for an instant, current flow charges the capacitor.
The instant you have just one electron difference in charge across a capacitor you have a potential difference.
A voltage develops across the capacitor as current charges the capacitor up.
The bigger the capacitor the longer it takes a charging source current to charge it to the desired voltage.
The charge current stops because the capacitors charged voltage equals the maximum voltage possible out of the PV.
Have you factored in the fact that any PV spec is a laboratory figure, in real life they need perfect conditions to acquire those specs.
Have you tried your project out in the sun and experimented if during the hours of no sun and low light you can keep the capacitor voltage high enough to run the controller and sensors.
Are you going to use a boost converter?
If you are doing any AtoD conversion, power supply stability may be an issue.
Do any of your sensors require a regulated supply for stable and accurate operation?
