Can I charge a 2.7v supercap, and do a analogRead() on the line voltage, and stop at 2.65 or 2.7v without destroying the dielectric within the capacitor?
I have a setup with tiny board space (20mm x 30mm), and I'm looking for capacitor charging options...
It should take 4.5~5.5v from usb (or charger anyway), charge the cap to 2.7v, then provide 3.3v to the main processor which is an attiny85.
My first idea is to use a ldo (cuz it requires less external parts than buck converter, I think i can omit the output cap cuz I'm throwing all the current to a supercap anyway) that converts 5v to 2.7v, and then send the current to the supercap.
but that's like a waste of energy, cuz there's 1.8v~2.8v * 0.5~2A energy turned into heat.
So I came up with this idea (see attachment).
The picture says it's a attiny13, but I'll probably use a ATtiny5 or ATtiny9 for my purpose...
And inside loop() I just do something like:
usb plugged in, supercap depleted:
1.1. Q1 is off for the moment, 5v goes C1 and get it charged.
1.2. turns Q2 on, U1 reads ADC1 below 2.7v, turn Q1 on, charge up the supercap, line drops to that voltage and slowly rising.
1.2.1. D1 prevents C1 dumping current to supercap
1.2.2. C1 slowly runs out, when it's below 1.8v U1 shuts off (BOD)
1.2.3. U1 off, Q1 off, line back to 5v.
1.2.4. goto 1.1.
my concern is that, does MOSFETs work like an diode when only GS connected? (unlike BJT, i see the gate is not touching the lines...)
because I know that avr pins are like 500ohm restance to ground when unpowered.
so Q2 is there to prevent supercap dumping current through PB1 of U1 to GND.
but if MOSFETs actually WORK like a diode, then it will dump current through PB0 anyway...
It's probably safer to use PWM and add a normal ceramic capacitor in parallel with your supercap (adding a low-leakage capacitor in parallel with the supercap is probably a good thing anyway).
(Apparently this method is worthy of a patent even though an unqualified hacker like me just thought of it in a couple of seconds).
yeah I know I can use PWM, I just want to simplify the whole thing down to a point that's easier to understand, and try study if it's gonna work or not.
maybe I should just get some MOSFET and get started?
I currently just got no transistors on hand, not even a BJT...
I have several 150+ farad caps, I've not put them to use because of the large current drain..
2.7v is not really enough to power an Arduino 2 caps in series to get 5v then regulate it, with arduino you charge via pwm and a power transistor, stop charging, wait a second then read the voltage, check if voltage is correct, if not repeat...
cjdelphi:
I have several 150+ farad caps, I've not put them to use because of the large current drain..
2.7v is not really enough to power an Arduino 2 caps in series to get 5v then regulate it, with arduino you charge via pwm and a power transistor, stop charging, wait a second then read the voltage, check if voltage is correct, if not repeat...
2.7v is not enough th power an arduino, but definitely enough to power an AVR (well even if its running at 16Mhz, avr these days really just having good quality! GJ to Atmel!)
need balancing circuit with throwing caps in series, especially those supercaps (they carry TONs of electrons).
You cannot connect a supercap to 5V rail with a MOSFET, you'll just crowbar the
supply.
The only efficient way to charge a 2.7V supercap from a 5V rail is using a constant
current current-sourcing buck regulator with an overvoltage cutout. These aren't a
very common topology.
The simplest way is to use a voltage regulator with a resistor in line with its input
terminal to limit current, but you waste energy.
Its worth looking for a supercap charging chip designed for the job (they exist for
many rechargable battery types so its not implausible). Some of the energy harvesting
control chips might do the task too, thinking about it.
Perhaps if you say what you are trying to achieve a simpler scheme might suggest
itself? For instance why a supercap at all?
Palatis:
can't I just use a 600mA auto-reset fuse for the current limitting purpose?
I don't really care the "ripple", cuz they're capacitors!
Fuse regular or self reset / poly / pptc
Both will max instantly cutting the current to the capacitors, a poly will cool down begin conducting and instantly heat up again.. you need to regulate the current properly and have a fuse to match if charging gets out of control.
When I said use 2 caps in series, I did not mean charge them in series....
charge parallel, discharge series, is also very dangerous.
An empty cap looks like this:
+-+-+-+-+-
==========
-+-+-+-+-+
and - are balanced
An charged cap gets more electrons on the - side, and more holes on the + side, looks like this:
++++++++++
==========
----------
but caps usually have a 20% tolerance, some manufactures even have -20% ~ +80%.
say you connect 2 caps with different caps together and tries to discharge them:
electrons going into the first one, push the electrons away, so it tries to fill the second one.
but it just don't have that many holes, so you end up "charging" the bottom one reversed.
Might be worth investing some time on using 2 caps independent of each other but to use a switching method to use both caps in a charge pump circuit , the efficiency is the only concern
Palatis:
charge parallel, discharge series, is also very dangerous.
An empty cap looks like this:
...
caps usually have a 20% tolerance, some manufactures even have -20% ~ +80%.
say you connect 2 caps with different caps together and tries to discharge them:
...
then the smaller one is gonna go kaboom.
Luckily for us those very same manufacturers put them together in matched pairs:
You should be able to charge one of those up with a 5V supply, no problem.
t85 is really efficient, takes 600~1000 uA, and I can put it into sleep.
but LEDs are really power hungry, I managed to give them 5mA each, that's 15mA.
so total power consumption is like 16mA, but I PWM them, so it should take ~4~6mA.
my goal is to let it work for 3~5 hr, so I need 12~30mAh.
5v supercaps are like all in small caps (max 1F), so if I charge them to 5v and discharge to 1.4v, that's 1mAh.
I'll need 12~30 of them...
Never heard of positive plate of a capacitor as having holes! A hole
is a positive charge carrier existing in a semiconductor lattice and
really requires quantum mechanics to fully explain.
A surplus or deficit of electrons in a metallic conductor leads to net
positive of negative charge in a capacitor plate (and if there is a dielectric
the polarization of the dielectric is important too).
Palatis:
well my toy's got 1 RGB LED + 1 attiny85.
t85 is really efficient, takes 600~1000 uA, and I can put it into sleep.
but LEDs are really power hungry, I managed to give them 5mA each, that's 15mA.
so total power consumption is like 16mA, but I PWM them, so it should take ~4~6mA.
my goal is to let it work for 3~5 hr, so I need 12~30mAh.
5v supercaps are like all in small caps (max 1F), so if I charge them to 5v and discharge to 1.4v, that's 1mAh.
I'll need 12~30 of them...
Tiny85s will work at 2.7V, no problem. I imagine 2.7V will put 5mA through most LEDs. Just run the entire thing at that voltage.
Never heard of positive plate of a capacitor as having holes! A hole
is a positive charge carrier existing in a semiconductor lattice and
really requires quantum mechanics to fully explain.
A surplus or deficit of electrons in a metallic conductor leads to net
positive of negative charge in a capacitor plate (and if there is a dielectric
the polarization of the dielectric is important too).
well yeah, it's just a easier way to explain.
polarized dielectrons are like + facing one side and - facing the other.
and yeah you definitely need quantum mechanics to explain how a graphine works in a supercaps.
fungus:
Tiny85s will work at 2.7V, no problem. I imagine 2.7V will put 5mA through most LEDs. Just run the entire thing at that voltage.
2.7v won't put 5mA through a blue led.
mine's like 2.0v 2.9v 3.2v for RGB, respective.
blue lights with 2.7v, at a very dim level (I mean, VVVVEEEERRRRYYYY dim).
green lights till ~2.3v, also at a very dim level (brighter than blue, tho).
red will light all the way down to 1.5v.
t85? i dunno, it worked till 1.5v at least, can't tell, cuz leds are off...
I'm looking at NCP1400, takes 2.7~0.2v, and boost the voltage to 3.3v (using 3.3v cuz i'm interfacing with USB...).
