Need help with low current voltage regulation

Try two diodes in series.

You need to research "voltage clamp"

1)You have a power shortage; and 2) lithium batteries are fussy about charge characteristics. Also, they don't like to be fully discharged; this can cause a permanent loss in capacity. I don't think you can get decent results from a simple zener diode and resistor circuit given the variability of solar cell output. What are the voltage/current requirements for the data logging device? That will play a key role in determining how much charging energy is needed to ensure adequate reserves, and that in turn sets the requirements for the charging circuit. If you use solar cells, information about the bird's habitat(s) would be useful for estimating the probability of adequate energy at all times of the year.

The details so far:

Your battery manufacturer's data sheet shows that the best discharge and cycle lifetime characteristics are obtained if the charge voltage is 3.1 - 3.3 v, and charge current is up to 100 microAmps. At this rate, it takes 96 hours to fully charge the cell. More bad news if you require the full charge rate every day: your solar cell puts out 50 µA (short-circuited, less at 3.1 V) in full sun at high noon, so you need at least two in parallel; four of them if your subject likes to sit low in the trees.

Voltage regulation and charge termination control are strongly recommended. An overcharged Li battery might send Tweety down in flames - there are a number of true horror stories about defective batteries or controllers causing cell phones to explode (China, 2006-2007). Since we're probably already looking at a MOSFET to control the charging, you won't also need a diode in series with the solar battery if the controller circuit can also perform an undervoltage lockout function.

The next challenge is to design a charge controller that operates on less than 25 µA. I looked at a couple of chips by Maxim, but they need as much as 3 mA for the chip while charging the battery. You might have to make something using a low-voltage CMOS op-amp/comparator.

http://www.seattlerobotics.org/encoder/200210/lithiumion.htm has a few notes on Li batteries and chargers that are a starting point, although the author's requirements are not as challenging as yours.

Alternate power brainstorms:

Is it possible to have the bird charge the battery with a magnet and coil mechanism driven by wing musculature? Part of the answer depends on the bird's power output in its normal "level" flight regime - 310 microWatts must not be too large a percentage. Also, weight, unrestricted range of motion and aerodynamics are important limitations. Another consideration is whether the bird will attempt to remove such a device.

Another idea, if the bird lives in an urban environment, is a power scavenging circuit that uses a coil to pick up alternating magnetic fields from electrical lines and equipment. Durability would be an issue since the coil would probably have to be made of #36 AWG or finer to meet size, weight and number of turns specs.

A set of three or four solar cell chips is probably still the leading contender, even with the question of keeping the cells clean for a whole year.

Thanks for the comments MNIce. I don't think the problem is as hairy as you make out. First of all, I don't think it is an undercharging or power shortage problem. The device uses very little current, about 2uA on average. So the battery should go for over 50 days without a charge. Also, I charge the battery fully before installing it. And it does not take 96 hours to charge the battery. That's just what they did for their test.

Yes. lithiums are fussy about charging, but I really can't afford to go all out with a charging circuit. I have to keep this thing tiny and simple. I'm hoping that holding the voltage at about 3V and keeping the current low will do.

Those are some creative ideas regarding alternative power sources, but from my work with birds, I'm pretty sure they would not work. Using wing musculature would require some way of mechanically harnessing the wing movement, and a bird generally won't stand for it. They would preen away any mechanism light enough to work. As for harvesting magnetic fields, these are not urban birds, so no go.

I just tried a 5V LDO regulator and got output ranging from 2 to 3.4V given reasonable light levels. So maybe a regulator will work after all. I'm going to buy a bunch and just plug and play to find the best combo. But I'm still looking for new ideas.

48 µA-hours/day to keep up is more doable - I was thinking you might need average currents on the order of 20-40 µA for the sensor package. I'm looking at the National Semiconductor LP38502-ADJ. It has 25 µA quiescent current and an Enable pin, and is available in a SMD package. It may be possible to use a micropower comparator and voltage reference to do the control functions - I'll sleep on it before working out the details.

The biggest question I have about the part is how it will behave at very low currents. It's designed for 5 mA+, and operating current requirements during the on-state may be an issue.

If that isn't practical, I have the beginnings of another couple of ideas to consider. The most efficient (and complex) may be a switching regulator, provided it can be done without exceeding the weight limit.

How much of the weight limit is used by the Li battery and solar battery? What is the maximum board space available? What regulators do you have on hand? Is the solar battery exposed when the bird's wings are folded, or can we only charge when the wings are spread?

The most expensive case would be if you had to build a hybrid module using bare chip dice to make the weight and size limits. I think we can avoid that...

I just ordered a bunch of SMT regulators (microchip MCP1700T-3002E/TT, Seiko 628-812C50BM-G, TI TPS73150DBVTG4...). The one you mentioned, MNIce, is not among them, but the ones I chose have really low quiescent current (~2uA) and low dropout voltage. Some of them also have an enable pin. Some of them also are "cap-free" meaning no external components are needed (big plus for me). As you mentioned, There's no telling how they will behave at low current--I'll just have to try and see. As for switching regulators, I don't think I have the real estate for that. I don't know of any switchers that don't need some big caps and an inductor.

There really isn't any weight to spare, the battery and solar cell take up about 40% of the total weight (last prototype was .7g total). But there is also some serious potting involved to seal the whole thing up and a harness to hold it in place. The rest is processor, sensors, memory, RTC, board, etc... The size is about 1cm x 2.5cm. And the tip of the device (where is solar cell sits) is always exposed. So as I said, there is plenty of power available; its just a matter of making it palatable to the battery.

I'll get back to this after I have tried the LDO regulators. Thanks again for chiming in.

ooh, an interesting job .. i have a few thoughts.

first thing seems to me a full blown regulator is an overkill, you really need a very low current shunt regulator to limit the battery voltage to 3.3v (or whatever the maximum is if lower).

something like a TL431 programmed for 3.3v directly across the battery might do the job, would beed to do some experiments to find the idle current of the circuit .. from looking at the data I am not sure if it would be 500nA or something less. I see 500nA quoted as an off current at 36v but no data on lower voltages. there may be other bandgap or precision references better suited to the job.

I also wondered if you had looked at maybe using a supercap instead of a battery, though the battery at 0.13g is pretty light anyway, not sure of supercap weights, its not specifed in the panansonic data I just looked at.

certain LEDs (non phosphor types?) can be used as photcells, as can any diode if the die is exposed.

if you want a run of more than a few dozen, after prototyping the circuit could then be built using die forms of the ICS, built on a ceramic substrate and potted in thin clear epoxy for the solar power side of things. tyhe size of the dies for non-power devices is generally failry insignificant compared to the package size. getting low volumes of the chips in die form may be tricky.

Hello:

One reason few, if any, regulators are going to work here is because they all require a minimum load current. Your load is essentially an 'off' condition to them and they will not reliably regulate. You need to look at the minimum load current requirement, not the quiescent current of the regulator, that has nothing to do with the load or regulation, that is an operating condition of the regulator.

I think your best bet is to use a simple FET voltage regulator circuit. This can be designed to operate at microamp currents with proper component selection. First, you'll need a constant current 'diode' to bias the regulator FET, this is simply (in this case) an N-channel FET with a resistor between the source and gate, an FET with a low gate bias voltage is preferable. Look up a constant current diode and you'll see how it is done. If the gate is in the 0.2-0.5 volt on range, you'll only need a moderate resistance for say 10uA. No, you won't find a commercial diode with this low a current.

The current diode goes from the 'regulator' N-FET's drain to the gate, a resistor goes from the gate to common, bypass this resistor with a capacitor. The source's output is the regulated voltage. This resistor adjusts the output voltage, with only 10uA of current through it, select its value for the necessary output voltage.

Lithium batteries are notorious about charging, you may have to go to a different coin battery type to ease the charging requirements. Lithiums do not like the 'trickle' charge method. A NMH battery may work better for you in this instance.

I'll try to post a schematic for you, I'm having a bit of trouble with the idiot program to do this. I'll see if I can give you some suggestions for components a bit later.

Here is a possible starting point for a low current voltage regulator, you will have to do a little empirical work to determine specific resistor, capacitor values as the FETs bias points are not identical. You can use SMT if needed to keep weight down. There is nothing critical about these components. The FETs are cheap and should work well with microamp currents. You may have to bump up the load current slightly if the regulator doesn't hold real good but a few more microamps shouldn't be a problem. I think the nickel-metal hydride battery will be a better selection, they are not so picky about charging.

Oops!, I just noticed that I drew R1 in the wrong place, it should be in series with the source terminal, not the gate, sorry about that.

Ok, I just modelled wiz's circuit using 2n5916 fets and its ideal for you.

R1 and R2 need to be much higher values to get the efficiency, if R1 is 10M and R2 is 5.6M then waste current is around the 200nA region at the inputs voltages of interest, higher values are certainly possible but cleanliness will become a major issue.

simulated the cell with a series resistance of 80k ohm and the battery B1 at 3.3v with a 1k series resistance ..

so a cell output of 4.5V results in a charge current of 79nA at 3.3V and 0 current at a battery voltage of 3.4v.

the same input voltage gives 24.18uA charge current at cell voltage of 2.5V, the most important bit being that the solar cell is providing 24.34uA .. so efficiency is up over 99%.

You will need to try these values out with real FETs, if you find the cell is still being overcharged then drop R2 further.

Cool. Any idea where 2N5916 JFETs can be had? I tried mouser, allied, newark, and digikey. Is there a viable substitute? Also, did you use a 0.01uF cap?

Thanks again, guys.

Nice work, guys. I had that idea in reserve, but was reluctant to try it because of the necessity of hand-setting the resistors and variations with temperature. Maybe I'm too fussy.

Is 2N5916 a typo? The datasheet says it's an NPN UHF large signal transistor in a 1/4" stud / micro stripline package, and was made by RCA back in 1974.

How would the Fairchild Semiconductor MMBFJ201 (SOT23 version of J201) work? Mouser sells it for $0.22 in single quantity; it has a V_GS(Off) less than 3 Volts, so it should do the job. If this is a one-of-a-kind project, you can buy five or ten and hand-select your transistors. In practice, you would use a potentiometer in series with a fixed resistor and change the transistor if you can't get the setting you want. Once you have the circuit working, measure the potentiometer/resistor and replace with the nearest fixed surface mount resistor.

If you go the JFET route, test the circuit at the bird's average body temperature while adjusting it, and design your mounting package so the regulator circuit's temperature is held more or less constant by the bird. This should prevent issues with temperature dependent JFET parameters.

The point about cleanliness is a good one. If I'm working with resistances above 4.7 MΩ, I avoid touching the insulated parts of the circuitry with my bare fingers. I've heard that salts in the skin can cause significant humidity-dependent current leakage, but, perhaps because of my precautions, I've not had significant problems, even with circuits sensitive to picoamps (nerve interfaces). In low-drain circuits, small leakage currents can have a significant effect on battery life. It may not be necessary to wear latex or plastic gloves while you build the field version, but better safe than sorry.

heh, yeh its a typo .. 2n5196 it was supposed to be .. I used 5spice analysis for a quick simulation and that was the only n-type FET listed in the default library.

I never checked the data .. its pretty dated I imagine.

ok .. the 2n5916 data lists a pinch off of 4v so the FET quoted above would do a better job anyway.

Stoney, would you mind sharing your Spice model? I'm trying to replicate your efforts, but I am new to Spice and can't figure out how to generate model and get reasonable output.

try using black tape to block some sun????

Thank you very much Stoney and Wiz. Your ideas are the sort of new approaches I was after. I am a bird guy and not an engineer, so shunt regulators did not occur to me. A quick search revealed a 3V fixed shunt regulator by TI (LM4040D30IDCKR) that works at 43uA and will require no external parts. I'm very excited to try this. That failing, I will try Wiz's transistor circuit. I may not have enough real estate for all the parts, but we'll see.

I've searched far and wide for alternative batteries. All of the NiMH I've seen are way too big. The battery has to weigh less than .4g, and there aren't many options at this weight threshold.

your short circuit current from the solar array is only 50uA .. your battery will never get charged with an LM4040..

Looking at the data and it will draw current of up to 40uA so it is no good in this application.

you need a nA solution if this is going to work at all. ideally you want a cct that draws absolutely zero until the threshold is reached. anything in the uA range before that point are going to severely cramp your charged time.

The current is actually 100uA. I may have the wrong photo cell part number. There is a small one 4v & 50uA and a larger one 4V & 100uA. I can use either one. Also, I do not need to fully charge the battery, just give it a little maintenance current every now and then. It will be soldered in with a full charge and will have a very low discharge rate.

Does the LM4040 really draw 40uA? I couldn't find that in the datasheet. Is current draw synonymous with minimum cathode current?

Regardless, I will follow your advice and look for other solutions. But I will at least put the LM4040 on a breadboard and see what I get. Thanks again for the shunt suggestion.

Hello:

I am familiar with the LM4040, I assume you are referring to the 3.0 volt version. Its nominal current drain is 47uA with 62uA maximum PLUS your load current. It might work if you can supply the all the current from the photocell. I'll see what I can find in small coin NMH batteries.

I haven't read the whole thread so apologies if this sort of thing has already been suggested.

Zener diodes are extremely good at about 5.6V: but down at the voltage you're looking at they're terrible and the data sheets are very misleading (at best). What you want is something which works just like you'd expect a good zener diode to work. What about a precision shunt regulator:-

http://focus.ti.com/docs/prod/folders/print/tl431.html

It just needs two other resistors to make it work properly: look at figure 3 on page 17 of the data sheet.

2½V sits across R2 you want R1 to take you up the other 0.8V

At these very low currents you may want to take the 2 to 4 µA I_ref into consideration.

I know that Electronics Wiz is much better at this sort of thing than I am. But this solution might be a bit simpler.

Hello Randall:

Zeners are designed to work with a minimal current flowing, usually the zener curves will show the two 'knees' between which it regulates. Irregardless of zener voltage, none of them will 'zener' or avalanche at currents below its knee. Motorola published an excellent book on zeners some years ago (Zener Diode Handbook), gave extensive data on how zeners work and lots of curves. Possibly hard to find these days. Generally, it is the misapplication of zeners that causes problems for which they are blamed. They really do a very good job when used correctly.

Looks like the TL431 has a min cathode current of .4mA, which is too high. I'll only ever get .1mA. The LDS431 by Leadis Technology (which claims to be a drop in replacement for the TL431) might work. Min cathode current is 18uA. Guess I'll give that a shot.

While the 431LP does have a TYPICAL current of 18uA, it can be as high as 40uA and don't forget that the programming resistors are going to draw current as well. There are usually limits on the values. The LDS431LP data sheet is rather lacking on some details. Check my earlier posting on the lithium battery, it changes the voltage requirements.

Morning folks:

I've read over the latest postings, you can try the 2N5196 circuit but two things; this dual FET is going to be comparatively expensive, way more than the other single FETs and you really don't need matched FETs and secondly, I have high misgivings about operating FETs at nano-amp currents. SPICE is notoriously bad at predicting circuit operation at these levels and many FETs do not operate predictably at nano-amp currents either. You can certainly try the circuit, no harm there.

You seem to have some overhead current capacity available from the photocell. My original circuit only uses 10uA, the current that goes through R2 is the only operating current. If necessary, it could be reduced a bit further (perhaps 5uA) but again, you are getting closer to iffy FET operation. You can try the circuit without the capacitor, just check for oscillation, but with moderately low resistances, you might get along without it or a smaller capacitance. Easy enough to find out on a bread board.

I would not worry too much about temperature variations, I don't think they will be bad enough to cause problems for the battery or other circuitry.

Speaking of which, try to do your breadboarding as close as possible to the actual layout, if it isn't too much inconvenience to work with. I'm still looking into the battery situation, will post if I find something.

Unfortunately, your lithium battery's charging requirements may complicate things beyond being useful. They are very picky about charging characteristics. If you don't follow the correct charging curve, they won't last very long.

Brid0030:

Read this source, they explain lithium charging without too much tech babble. You might just get away with lithium. It would appear that after initial charging to full capacity (and I recommend the 4.10 volt level to be on the safe side), we could get away with a 'trickle' charge of 4.10 volts without causing damage to the cell.

batteryuniversity.com

This does present a problem, you noted that your photocell only puts out a peak voltage of 4.5 volts on a good sunny day. While the battery certainly should have enough reserve capacity to power your circuits during the less than peak current periods, the problem is the approximate 0.4 volt overhead for the regulator.

I think that 4.1V charge level is for batteries rated at 3.7V (a standard for digital cameras and perhaps other things). The datasheet for my batteries recommends a charge voltage no higher than 3.3V. 3.1 is nominal. So there should be reasonable overhead from a 4.5V source. I didn't see much reassurance in the article that trickle charge was OK. Trickle charge above the rated voltage can cause lithium plating and battery failure. My hope is that keeping the trickle charge at or below the rated voltage (3v in my case) with low current <50uA will allow the battery live a long and fulfilling life.

I should have checked your specific battery type, there are many different chemistries in lithium. Yours does have a difference voltage spec. I didn't have time to look it up earlier as I got a call and had to leave, sorry, should have waited to comment.

Yes, there should be sufficient overhead at 3.1 volts, now to the 'problem' with trickle charging. Lithium does not accept trickle charging as mentioned in the article (just different voltage applies). Now, how to implement some form of intermittent charging as required? Assuming we go for ~3.1 volt nominal (which leaves some room to work with for variables) charge, we need the battery to drain off some power before giving it some juice to keep it happy.

Here is what I suggest, put a diode (1N914B/1N4148) in series with the output of the regulator, anode towards the regulator, cathode to battery (don't want too much of a drop here). The diode will be reversed biased until the battery voltage drops enough for the diode to conduct. We select an output voltage of the regulator that allows the battery to drop off a couple tenths of a volt.

If you decide to use the LDS431, for example, or which ever circuit you end up with, I would set the output voltage of the regulator to 3.25-3.35 volts, perhaps even 3.2 volts. You may need to play around with the voltage to find the 'sweet' spot but I think 3.25-3.35 volts should work with the the external variables. For the LDS431, R1=158K and R2=100K.

This will allow the battery to discharge some before a trickle current resumes. The diode's forward voltage will slowly increase from about 275uV at 1uA to 400uV at 15uA charge current, it is a slow and linear increase in voltage.

You will likely have to play around with the regulator's output voltage a little to find a reasonable trade off. You don't have to maintain the battery's terminal voltage at nominal, just keep it from discharging too much.

Hey Wiz. So the "trickle" charge in this situation will be intermittent because the photo cell will not always be at full blast. Bird movements and nighttime will see to that. As for the diode, that was my plan exactly. The diode will have to be part of the design to prevent discharging the battery at night when the photo cell will just act as a big resistor. As you say, I don't need to maintain a full charge. I just need to keep the battery alive. I think the first thing I will try is a 3V fixed regulator and a schottky. That failing I'll tinker with other options. I also like your original solution with the dual JFETs. I'm just worried that my limited knowledge and experience will keep me from fine tuning that circuit correctly. Thanks for all the help.

Hello Brid0030:

You are quite welcome. First, don't get bogged down in minutiae, you have a fairly wide range of operation here and have room to play. Use the common and cheap 1N914 diode instead of the Schottky, at these currents, there is no advantage. Unless you are under an approaching deadline, take your time and experiment, best way to learn. If you end up needing a little more 'swing' in the lithium's discharge voltage, we can change the diode for one with a little more voltage drop. Your battery has plenty of capacity at even 2.5 volts terminal (if your 'load' can tolerate that) voltage.

It would appear that with this arrangement, charging won't take place for seveal days under your estimated current drain. Should keep the battery happy for a long time.

Good luck and 'call' if you need to.

Everything I work on is an industrial higher voltage and current level than what you are working with.

But I must commend your efforts and documentation.

Keep the CR4 community appraised of your progress.

We may than then benefit from your research.

Good luck.

Just wanted to extend my thanks again and mention that I am waiting for some parts to arrive. I will sign on again after I have had a chance to tinker a bit and see what works. Tune in next week.

I tried several voltage regulators today with a bit of success. The first couple of linear regulators did not work--not nearly enough current I would guess--and couple only output a volt or two even with maximum output from the solar cell. I did have some success with the LM4040. It yielded exactly 3V but there had to be very high output from the solar cell. If the cell experienced cloudy conditions or any amount of shade, the voltage would drop rapidly (even though the output from the solar cell would be well in excess of 4V). Nevertheless, this arrangement might work as I expect the birds to spend a fair amount of time in full sunlight--they like open areas as opposed to forests. But I would like to get some battery charging done whenever the cell outputs 4V.

Perhaps a more promising device is a CMOS regulator, the S-812C by seiko. Only problem is I used a 5V regulator instead of a 3V. So this is the wrong part, but I was able to 5V regulation when I set up 2 small photo cells (4V & 50uA) in series. Together they output 8V and 50uA. The S-812C gave 5V over a wide range of light levels--almost all the way down to full shade. So my next move will be to get the 3V version of the S-812C and see if it performs in a similar manner. If necessary I can get an 8V 50uA version of the photo cell, so my hopes are high that some arrangement based on a S-812C will work. So once again I will wait for some parts and chime in later.

Hello,

I suggest using the 3.3 Volt version of the S-812C and installing a diode between the output and the battery. The data sheet warns that there is a parasitic diode in the regulator pass transistor, and if this diode is forward biased (charged battery in the dark), the regulator could be damaged. Otherwise, it looks very promising.

I looked at a few Linear Technology parts and found an interesting possibility if the Seiko regulator doesn't work out - the LTC1540. Page 8 of the data sheet shows how to use this voltage reference/comparator to make a low dropout regulator with only 5.5 µA quiescent current. The data sheet is at http://www.linear.com/pc/downloadDocument.do?navId=H0,C1,C1154,C1002,C1463,P1593,D1777.

This design may not need an anti-reverse diode (depends on the MOSFET), and the hysteresis pin can be used to control charge/discharge behavior. I would make the following changes: adjust the feedback voltage divider R₂-R₃ to get 3.1 V, try a smaller value for C₁ if that will save weight, and perhaps increase R₄ to set the maximum current below the solar cell short-circuit value. Note that if you use hysteresis, the quiescent current will increase by the amount used in the hysteresis voltage divider.

Hi MNIce. The series diode has always been part of the plan, and was in place in my testing circuit. I don't see any voltage drop--probably because of the low current. I already ordered a 3.0V regulator. If the final product gets me anywhere close to 3V then I will be satisfied. The device should work with as little as 2V, a bit less than 3V should be OK. I wish there were a real electronics store right around the corner where I could get one of everything....

I looked at the LTC1540. This will sound crazy, but I don't have enough room for an 8 pin package. I'm more or less restricted to an SOT-23 and a couple of resistors or caps. I might give the LTC a shot, but it would require a complete redesign. Thanks for the suggestion.

Hello Brid0030:

Sorry, couldn't get to you any sooner. That S-812C has a rather fat data sheet! Looks like it will work with a light load such as yours although there is a warning about output voltage rising with loads near yours. You may want to put an additional resistor on the output to pull 2 or 3 uA more, it shouldn't cause any problems with the battery charge. Input-output differential voltage appears to be nearly zero with microamp loads, I don't think you need to worry about temperature drifts, not that much here. Definitely use the diode as planned between the battery and the regulator.

I agree, it is tough to find parts, especially if you're hunting then through catalogs, no one has it all. Shipping and handling can become a real pain along with minimum order amounts.

As I said, very few regulators work at the current levels you're running at. Remember, minimum output current of the regulator is very important, quiescent current is only important as to the total current draw on the battery. If your load is too light for the regulator, it won't regulate!

Good luck,

Wiz

I think I have done it, thanks to all the help from this forum. The 812-C works really well. It puts out 3.02V no matter what I do. Vary the input current, vary the input voltage--nothing but 3.02V coming out so long as the input is over 3V. I do get a voltage drop of .25V with a diode and about .1V with a schotkey. In the end I will probably go with a 3.3V regulator (as per MNIce's advice), such that a regular diode will drop the voltage to an ideal charge level (between 3 and 3.1V). Regardless, I'm pretty sure that what I have on the breadboard right now will work.

The max current output is a bit lower than I thought--45uA. I have a 511K resistor between Vout and ground to ensure a 5 or 6uA load on the regulator (Note that this did not appear to be necessary, but I did not simulate all battery conditions). I guess the regulator is eating more current than I thought, but I am satisfied with 45uA.

I have the circuit charging a battery under a bright lamp. Battery voltage is up to 2.73V and rising. We'll see where it peaks out. For those of you looking for voltage regulation at very low currents, the 812C is a very easy, 50 cent solution.

Glad to hear the 812-C worked out for you. Your lamp may not be putting out that much 'light' at the wavelength your cell is most sensitive to. Many lamps do not put out the same spectrum as the sun even though they 'look' similar.

Good luck.

Yes, it looks like you have a winner. What happens when you turn out the lights on the circuit when the battery is fully charged and the load is connected? I would want to check for two things - no potentially damaging reverse current through the regulator, and battery voltage staying above 2 Volts after 24 hours with the solar cell covered. The circuit should pass on both counts, but it would be good to verify that.

For my part, I'm glad to know about the 812-C; I'll save the datasheet in case I run into a project with similar requirements. Let us know how the final build works. I'd also like to see a link to your research when you have something to publish; I'm interested in many scientific disciplines.

Thanks for keeping us up to date.

If you're really close to your power budget you're better sticking with the Schottky.

Another thing you might consider if you're not quite there is smoothing the output of the solar cell with say a 100 µF cap: this is a 1210 (0.12" X 0.10") device:-

http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail&name=587-1965-1-ND

This will only work if you think you're getting bursts of power less than a second long because of the birds flitting in and out of sunlight.

Randall:

While many diode data sheets don't show it, the forward voltage drop of even a run-of-the-mill diode such as a 1N4000 series will be in the 250mV-350mV range at very low currents and I have not seen very much data on very low microamp currents. The computer diode I mentioned does show very low current data, I do not think that there will be any advantage to using a Schottky diode in this case.

While your suggestion of capacitor storage is not a bad idea, at these microamp load currents, the leakage current of the capacitor can become a significant issue. Even a low leakage capacitor can be significant compared to the load current. Since this project is expected to see a fairly wide variation in temperature, the capacitor's leakage is going to be varying all over the place. It certainly wouldn't hurt to try a capacitor if there appears to be a need for it.

I'm not clear on how variation in the output from the solar cell will cause damage. When simulating different light levels, solar cell output below 3V caused a near identical drop in voltage output from the regulator. There was no apparent switching on and off. So, what does it matter if solar cell output fades in and out from time to time. I get >3V even in the shade, so there should be at least some charge current throughout most of the day regardless of what the bird does. I don't expect to be actively charging the battery all the time. Sorry if I'm missing something obvious.

As long as you're getting enough from the solar cell to keep the battery charged there are no problems. I was just trying to extract the last little bit of power out of the solar cell: probably not worth considering.