First I have no real need for this at the moment but had some thoughts about harvesting some power.
So if I have 1mA 0.2V from a temperature difference, (in general 24/7), how do I charge a 3V battery (2-5V this end is not important just something my AVR can use).
Are there a (efficient) way to do this?
Lots of companies now have energy harvesting chips for precisely such applications!
Try Linear, Analog Devices, et al ...
I was more thinking of a circuit where the AVR could do the job. But I guess that I can copy how they do it, the most important is actually than the transformers are easy to get.
Add
And yes I guess that 1mA 0.2V was a tad optimistic for a AVR, but it actually look like linear has a chip that can do it.
I think this is one of those cases where it sounds easy in principle, but the devil is in the detail.
I suspect that the chances of replicating the performance of a purpose-designed chip are rather slim ...
The main problem is to get the primary side to oscillate then this cam up:
https://www.avrfreaks.net/comment...
and a mega169 have this data.
32 kHz, 1.8V: 25 µA (including Oscillator and LCD)
So perhaps the LCD driver can do the job :)
Other than that an tiny13 use about 40uA active running 128kHz (WD)
and about 10uA in idle, so I wounder what idle and pwm will use.
|
"Experience is what enables you to recognise a mistake the second time you make it." "Good judgement comes from experience. Experience comes from bad judgement." "Wisdom is always wont to arrive late, and to be a little approximate on first possession." "When you hear hoofbeats, think horses, not unicorns." "Fast. Cheap. Good. Pick two." "We see a lot of arses on handlebars around here." - [J Ekdahl] |
 |
You say you are wanting to "harvest" some POWER. By using an AVR wouldn't you end up using more power than you harvest. After all you have to power the AVR.
I'm guessing the dedicated chips that Awneil mentioned are some sort of boost converter, but designed with maximum efficiency in mind to minimise power lost in the conversion process. You cannot get power from nowhere so if these ICs increase voltage you will get a corresponding decrease in supplied current.
Thanks that is more like what I want :)
But I don't like the 1K resistor around the open Q2 when not charging , but I guess that there should be some other way to do it.
Hello Sparrow -
One of the huge problems about "energy harvesting" is the ability to start up, with no higher voltage source to run things. Imagine that your battery is discharged and your 0.2V source is the only thing you have. Then, somehow, the electronic circuit has to start running with only that little bit of voltage. It is certainly not enough to make a bipolar transistor work as there is not even enough voltage to get some base current to get the collector to do anything. You certainly are not going to get a micro running under that condition!
LinearTech has at least one chip that will run with just the very small voltage from a thermocouple. Check out LTC3107 to see the extremes that are needed to get operation at such low voltages.
Jim
I'm guessing the dedicated chips that Awneil mentioned are some sort of boost converter, but designed with maximum efficiency in mind
Indeed.
They also do things like maximum power point tracking (MPPT), battery/supercapacitor management, power-path control, etc, etc, ...
Doing all of that yourself is no mean feat!
So you need a step up ratio (without counting losses) of 15 which means that the 1mA needs to be divided by 15 times (67uA) to keep the overall power of 200uW otherwise you would be creating power from nothing.
Or is my maths wrong?
Your maths are right on, John! However, you are not accounting for the efficiency, or lack there-of, in the converter. Could be anywhere from 5% to 25% loss.
Jim
But I did say "without counting losses" ie converter efficiency. 
But I don't like the 1K resistor around the open Q2 when not charging , but I guess that there should be some other way to do it.
Do a bit of Googling (I provided a suggested link) and you'll find a wide variety of variations on the basic theme. There are some claims of devices operating down to 0.035 V input, but I haven't perused them, nor can I say if minimum startup is that low. I would be sceptical.
I'm guessing the dedicated chips that Awneil mentioned are some sort of boost converter, but designed with maximum efficiency in mind
Indeed.
I would point out that @sparrow2 made no mention of a required >>measure<< of efficiency, only that it should be "efficient". That's hard to characterise. Boosting by a factor of 15 to get 3 V from 0.2 V will bring with it certain unavoidable losses.
The only other numbers mentioned are a 1 mA, 0.2 V source from a temperature differential. That amounts to 200 uW. At 1.8 V, a picoPower device like the ATmega328P running from the 128 kHz oscillator in active mode draws about 33 uA, or about 60 uW. That's a significant portion of the available 200 uW, assuming 100% efficiency in boosting 9-fold from 0.2 V to 1.8V. I'd guess that 20-50% would be more typical, so available power would be in the range of 40-100 uW, the lower end of which is below the active mode draw. Idle mode draw (sufficient for PWM output) is about 8 uA, or about 15 uW.
But some other math might be helpful. What kind of 3V battery are you expecting to charge? Something like this tiny rechargeable LiIon CR2032 has a capacity of 40 mAh. That amounts to, roughly, 150 mWh, or 150,000 uWh. Even if your thermo-electric source can be boosted with 100% efficiency, and even if the charging efficiency were 100% , it would take 750 hours to fully charge the cell. That a full month.
So I guess the question is, what is the end goal?
|
"Experience is what enables you to recognise a mistake the second time you make it." "Good judgement comes from experience. Experience comes from bad judgement." "Wisdom is always wont to arrive late, and to be a little approximate on first possession." "When you hear hoofbeats, think horses, not unicorns." "Fast. Cheap. Good. Pick two." "We see a lot of arses on handlebars around here." - [J Ekdahl] |
|
I know that 1mA 0.2 volt are pushing it, with about 2.5-3V and a efficiently of 30% (the hole system), what give about 20uA, and a tiny13 in idle, should use 10uA so it's not impossible.(using WD as osc)
And if we say 2mA 0.3V then it's not a problem.
Then there is the problem with start up voltage, but it looks like the Joule thief can start very low whatever that is. (and perhaps play with transformer parameters, or even have different parameters if micro don't run so it easier oscillate but perhaps not so efficient )
I looked at the LTC3107 but before you are there it will not be cheap.
Add:
If I use a Joule thief the chip don't need to do any PWM,
And as a battery I was thinking of 2 AAA or perhaps a super cap will do.
So first test would be to have a LED on for 1 sec every min, and see that the battery voltage still go up.
I dare say that even for low capacity NiCd or NiMH 'AA', perhaps 800 mAh, you're looking at upwards of 2 years to deliver a nominal full charge. Without doing the math, I'd guess that 10 uA would be at least an order of magnitude lower that the self-discharge rate of a typical cell. It would be like trying to walk up the down escalator.
|
"Experience is what enables you to recognise a mistake the second time you make it." "Good judgement comes from experience. Experience comes from bad judgement." "Wisdom is always wont to arrive late, and to be a little approximate on first possession." "When you hear hoofbeats, think horses, not unicorns." "Fast. Cheap. Good. Pick two." "We see a lot of arses on handlebars around here." - [J Ekdahl] |
|
Can I ask, if the power source is 200-300 uW, what is the value in having 2,400,000 uWh of storage?
|
"Experience is what enables you to recognise a mistake the second time you make it." "Good judgement comes from experience. Experience comes from bad judgement." "Wisdom is always wont to arrive late, and to be a little approximate on first possession." "When you hear hoofbeats, think horses, not unicorns." "Fast. Cheap. Good. Pick two." "We see a lot of arses on handlebars around here." - [J Ekdahl] |
|
But when you put it up, it's not a problem to expect that they are charged when you put it up.
The main thing is that they never (whatever that is 5 years or so) get low, and perhaps the leak current in the battery is one of the bigger problems.
And tests needs to show if 1mA 0.2 mA is minimum so if AVG is 2ma 0.25V it's .
My question was just to see if it actually was realistic.
And as i said in #15 the first thing is to make this test:
So first test would be to have a LED on for 1 sec every min, and see that the battery voltage still go up.
where the battery probably is a big cap. (and perhaps only on 1/50 of a sec), just to get a feel for it
But I don't have time this weekend.
Add:
And if the osc stop running perhaps the AVR give it a little kick with a port pin that normally is an input.
I have been looking at the net and it looks like you actually can get a joule-thief to auto start at very low voltages if JFET's are used, about 10mV for J310's and down to 5mV for 2SK170.
So it's not at problem with my 200mV, and there will even be some headroom to make it efficient ;)
