Stable power supply
Catweasel
Posts: 34
Hi,
I am currently working on a little project that involves a Stamp2 and a Propeller microcontroller.
At the moment, they draw their power from their respective evaluation boards. What I now want to have is a combined power supply for the processors and other hardware.
The power shall come from an array of five solar panels which deliver a combined output of 30V and a rechargable battery pack that provides 6V.
I have attached a circuit draft that would feed the parallel power sources through a capacitor. Then there are three loops (3.3V;5V and 12-18V). I just placed two LEDs and a motor as dummy elements in the drawing. In the real application, the Stamp would be conected on the 5V rail, the Propeller on the 3.3V and any other high power hardware would be fed by the 12-18V connection.
I just don't have enough knowledge/expierience to complete the circuit and divide the voltage in a way that would accomplish the following:
1. Power the "low voltage ports" primarily via the solar panels. Only if power is insuficient, supplemet from battery.
2. If solar energy production exceeds power consumption on the "low volatge ports", the battery is charged.
3. If the previous applies but the battery is fully charged, keep it full, don't overcharge.
4. If insufficient power is available (empty battery and shaded collectors), the 12-18V should be disconnected, thereby postponing power starvation on the 3.3 and 5V ports.
My ideas so far are:
1. If I run the solar power through a transistor that block the battery current, the transistor would open up when voltage drops on solar array, thus the battery supllies continuous power. But how to set this up?
2. How would I step down the solar power to a steady 6V charging current for the battery if comsumption is variable on the "low voltage ports". I guess a simple voltage divider won't do. Also the solar array could produce anything between 0V and 30V. Any idea?
3. Would the battery actually overcharge at all? When fully charged it there should be no voltage difference in the charging circuit, thus no overcharge. Correct, or am I too naive here?
4. A transistor that only opens up above a certain voltage should do it. But I guess I am, again, too naive here.
If aynone could help me here, step by step, understanding how to do this, that would be very nice.
Sorry if some things might be explained a bit quirky, but English is not my native language and I try my best :-)
Cheers,
Catweasel
I am currently working on a little project that involves a Stamp2 and a Propeller microcontroller.
At the moment, they draw their power from their respective evaluation boards. What I now want to have is a combined power supply for the processors and other hardware.
The power shall come from an array of five solar panels which deliver a combined output of 30V and a rechargable battery pack that provides 6V.
I have attached a circuit draft that would feed the parallel power sources through a capacitor. Then there are three loops (3.3V;5V and 12-18V). I just placed two LEDs and a motor as dummy elements in the drawing. In the real application, the Stamp would be conected on the 5V rail, the Propeller on the 3.3V and any other high power hardware would be fed by the 12-18V connection.
I just don't have enough knowledge/expierience to complete the circuit and divide the voltage in a way that would accomplish the following:
1. Power the "low voltage ports" primarily via the solar panels. Only if power is insuficient, supplemet from battery.
2. If solar energy production exceeds power consumption on the "low volatge ports", the battery is charged.
3. If the previous applies but the battery is fully charged, keep it full, don't overcharge.
4. If insufficient power is available (empty battery and shaded collectors), the 12-18V should be disconnected, thereby postponing power starvation on the 3.3 and 5V ports.
My ideas so far are:
1. If I run the solar power through a transistor that block the battery current, the transistor would open up when voltage drops on solar array, thus the battery supllies continuous power. But how to set this up?
2. How would I step down the solar power to a steady 6V charging current for the battery if comsumption is variable on the "low voltage ports". I guess a simple voltage divider won't do. Also the solar array could produce anything between 0V and 30V. Any idea?
3. Would the battery actually overcharge at all? When fully charged it there should be no voltage difference in the charging circuit, thus no overcharge. Correct, or am I too naive here?
4. A transistor that only opens up above a certain voltage should do it. But I guess I am, again, too naive here.
If aynone could help me here, step by step, understanding how to do this, that would be very nice.
Sorry if some things might be explained a bit quirky, but English is not my native language and I try my best :-)
Cheers,
Catweasel
1024 x 751 - 45K
Comments
www.dimensionengineering.com/products/de-sw033
Here's one link I found on a simple Google search.
There are also these, which take in voltage from 2.6v to 14v, and can output 3.3v or 5v: http://www.dimensionengineering.com/products/anyvolt-micro
Also, you haven't mentioned what battery chemistry you are using? If you're using rechargeable lead acid then you should be able to just trickle charge them with a diode.
You could have the power supply for the electronics automatically switch over by using diodes: simple connect a diode pointing from the battery to the electronics regulator, and a diode pointing from the solar cell to the electronics regulator. This way, whichever has the higher voltage will power the electronics.
Your schematic has the solar cells wired in parallel. In order to get 30V out, you'll need to wire them in series.
Thank you for your quick reply.
In the final application, the Stamp and the Propeller will sit on their own PCBs, so I have to leave the safety of the evaluation boards and the processors will be at the mercy of my circuitry.
Thats why I want to get it right :-)
Ok. When you know what to look for, google makes life easy :-)
But this link has really given my some insight and ideas...
Especially the LM317, previously unknown to me. As far as I understand it, i could use three of them to control the voltage on my output ports, or not?
About the circuit in general. If I see it correctly, once the battery is full, it drives the Transistor that controls the charge circuit. Since that would discharge the battery, the transistor would switch after a while.
Would such a system not tend to oscillate. Like perpetous charging/discharging at the brink of a full battery? I once heard of a thing called Schottky-diode. Could this be used to give the charging some leeway?
Cheers,
Catweasel
I was playing with several options to avoid series cuircuits on the cells. If I connect the five arrays in series and one array is shaded, I would loose more that 1/5 of the voltage due to the panel's inner resistance.
On the other hand, I could wire them parallel and could regulae it more easyly to 5V and 3,3V. But then the 12V would be unsupplied.
I just had the ad-hoc idea of splitting the array.
Parallel connecting 3 panels for 3.3V-5V supply and wiring 2 panels in series to get 12V (I don't really need 18V. That would be a "nice" have).
The only drawback is that power might be not used efficently.
Thanks for the correction.
Cheers,
Catweasel
http://www.amazon.de/Rolllra-Akku-1550mAh-Ersatz-Akku-1500mAh/dp/B005CU4PH2
Most lead batteries are a bit too heavy for my use. My application has a very tight weight budget.
What kind of battery would be ideal in terms of weight/easy to implement ratio?
Cheers,
Catweasel
The Stamp module itself has a 5V regulator on it. You can use another LM317 to produce 3.3V for the Propeller or you can get a fixed voltage 3.3V regulator. The schematics for most of the Propeller boards are on Parallax's website and you can duplicate their circuit.
I use Propeller Educaton Kit. Iwired it just like in the diagram.
And I just had another idea. Maybe I am thinking too complicated all along.
Correct me if I am wrong, but:
The Propeller runs at 3.3V. However the BOE setup provides for 5V. As far as I understand it, that is to drive the EEPROM which runs at +5V.
Can't I connect the Stamp then directly to the 5V line?
Then I would only have to build the charging circuit and no more voltage dividing... or am I missing somethinge here?
Cheers,
Catweasel
Are you talking about the Propeller's EEPROM? It uses 3.3V like the Propeller.
I curious what you want to do with the Stamp. There's isn't much (or anything) the Stamp can do that the Propeller can't. The resources required to communicate with the Stamp would probably be more than what the Stamp could contribute to your system.
Edit: I possibly exagurated the Stamp's uslessness. I have a hard time imagining a project where a Stamp would add much usefulness when one is already using a Propeller.
Then I misinterpreted/misunderstood the theory of the board setup. I just winder why there is a 5V output implemented then. If its not driving the EEPROM, it seems only to be used to feed the 3.3V regulator. Why this two stage down-step. Why not using one regulator from 9V to 3.3V?
Sorry if this is a dumb question. :-)
Cheers,
Catweasel
A legitimate question. :-)
Basically, I want to have two different microprocessors cooperate/ work together. Creating the signalling protocols will be fun.
So it more of a challenge than a necessity.
In the end the Stamp will be driving some servos via PCM and read some (very few) sensors. The steering commands would come from the propeller and the stamp would do the rest. So communicaton would go mostly in the Propeller->Stamp direction, with the exception of an abort status line, so the Stamp can stop on defined sensor conditions and signal the Propeller an "error".
The Stamp is then a programmable servor driver / sensor sampler...
eg a standard solar charge controller http://www.futurlec.com/Solar_Charge_Controller.shtml
And grab a solar panel - eg 10W or 20W. These have 36 cells and an open circuit voltage of around 18-21V. Much easier than wiring up and waterproofing individual cells.
And a 12V SLA battery.
Then you can add your regulators etc and the 5V and 3V will be there all the time. Or use a board as mentioned by others that has the regulators on the board. If you are worried about efficiency then use a simple switcher eg LM2574 that steps the 12V down to 3V with around 90% efficiency so little energy is wasted as heat.
I think the 5V regulator is there because so many sensors and devices (like servos) use 5V. Many of the Propeller boards have both 3.3V and 5V lines even though none of the parts on the board need 5V.
Absolutely true. But thats is what I am interested in to find out how to do. Just buying a "black box" that does all the magic would largely defeat the idea of learning by doing (it yourself). Making a working application is for me only 50% of the joy. I also want to know how and why it works what I have done.
When I use lavishly complex prefrabricated parts, I feel somehow like a cargo-cult engineer. :-)
I am aware that a solar charger might fall in the "if you have to ask, wou will not be able to do it" category, but thats just how all learning starts, by asking.
Maybe I start bit by bit, implementing the battery as a backup power source first and then thinking about a simple charging circuit and then adding voltage protection. The main thing is that I am learning something.
I also had the idea of using a microcontroller exclusively for power management. I just thought it would be "safer" (in terms of overall system reliability) if the powermanagment would be implmented as an analog circuit.
For example: What would be the best power management solution if the solar/battery assembly was to be installed on a satelite (heat and radiation issues set aside)?
I already have five of the following panels available. The protection of the cells is not that critical.
http://www.parallax.com/Store/Accessories/PowerSupplies/tabid/165/CategoryID/39/List/0/SortField/0/catpageindex/2/Level/a/ProductID/619/Default.aspx
But maybe its just a misunderstanding and you assumed tgat I fabricate the collectors from individual photovoltaic cells.
Nah, I am referring to the item 750-00030 as a "cell"....
The weight is a somewhat critical factor in my application. The overal weight of the apparatus should not exceed 10Kg.
With all the structural parts, the servos, sensor arrays and other supporting hardware, I cannot really afford a heavy lead battery.
The less weight, the better. The only reason I am not considering Lithium-polymer batteries is the cost factor. If NiCa is so "tricky" to handle, what is the next best alternative that would be in the same weight range?
Yes. stepping down the voltage (if stable 12V are present) should be rather "simple". The whole thing gets only complicated due to the unstable nature of solar energy production.
I also thought about maybe just using a processor, like a stamp, solely for power management. Of course that would introduce a single point of failure to the system.
Cheers,
Catweasel
Ok, those solar modules will be fine. I count 12 cells per module, so put 3 in series and you have a standard 36cell panel for charging a 12V battery. You need one reverse voltage diode as well, so effectively you lose one of the cells and it is 35 cells.
If you have a lower voltage panel then you still lose that one cell from the reverse diode, so as a percentage, it is better to charge a 12V battery than a 6V one.
You need more volts open circuit than the battery because you only get the full open circuit volts in direct sunlight at midday in summer with no clouds etc etc. As a rule of thumb, charge a 12V battery with 18V of panels. So a 6V battery with 9V of panels, which sort of does not work as that is one and a half of your modules.
Hmm - if you are using nicads then you can tailor the voltages to suit. How about two of your modules in series gives 12V open circuit and go for about 8V on the batteries? 1.2V per nicad - 6 nicads is 7.2V is ballpark about right. Don't forget the reverse diode.
Nicads and NiMH will dissipate overcharging current as heat, and as long as that is not a great % of the total capacity you could get away with that. So 2000mAH nicads, charging with 100mA solar cells and you probably don't even need an overcharge protection.
Then you just need to watch for undervoltage.
LM2574 switch mode is one 8 pin chip, a $2 inductor and a 10c diode and the chip does not even get warm and will give you a regulated 3.3V perfect for the propeller. The propeller has enough smarts to become its own voltage monitor. Lots of homebrew possibilities there.
Do you know battery cells that deliver more than 1.2V, so a series assembly would not need 10 cells?
That was my initial motivation for a rather low battery voltage as it would only require 6 cells and 6V is the "native" voltage of the solar cells.
But I just found out that I can modify my design to handle the weight of 10 cells.
I gave some more thought to the choice of battery. Thisone looks just suitable to me (in an array of 10 cells)
http://www.produktinfo.conrad.com/datenblaetter/250000-274999/251736-da-01-en-SANYO_NICD_AKKU_SUB_C_2500MAH_ZLF.pdf
However, there is another version:
http://www.produktinfo.conrad.com/datenblaetter/250000-274999/251735-da-01-en-SANYO_NICD_AKKU_C_HOCHTEMP_2500MAH_ZLF.pdf
Although it is slightly heavier than the firstone, it can handle higher temperatures, thus it should cope better with overcharge heat. And it says something about trickle charge. The otherone doesn't. Would that mean that one can't be trickle charged?
What would be a good way to approach this?
I am also more and more thinking in favor of an "active" power control. I just thought by using non programmed analog circuitry, the system would be "more robust" against program errors, or glitches.
Cheers,
Catweasel
http://www.youtube.com/watch?v=xyyqwQLuNFw
Too bad they don't make shows like that anymore.
I came up with a new design for the power circuit.
It uses two LM317. I changed the battery to be about 9V. With a 12V charge from 2x2 parallel/serial solar cells that should sufficiently feed the battery. It should be possible to keep the thing alive with only a 4 cell array.
On the other side the 12V port is feed parallel to the battery, thus also 12V should be available.
I just don't know how to figure out the proper values for R1 and R2 as the input from the cells can vary.
I looked at the circuits in the LM317 datasheet and tried to understand Figures 8,9 and 10.
Unfortunately its not stated for what voltage and battery the example circuit is designed.
Coukd you guys please have another look at it and tell me:
- if it makes sense at all
- if the 1N4001 is sufficient, or a 1N4002 should be preferred
- could the following work:
+Replacing R1 and R2 with 2N3904 transistors.
+Drive the transistors with the fifth solar panel, thus regulating the LM317.
(I assume that all panels have simmilar lighting conditions).
Thanks for helping me :-)
Cheers,
Catweasel
No, avoid NiCd cells completely. They have a severe memory effect and unless you regularly discharge them fully and recharge they effectively lose all capacity (this can be recovered with care, but its a real pain). These days you use NiMH cells - hardly any memory effect, and no toxic cadmium to worry about. NiMH also handle mild overcharging very well. NiMH cells are 1.3V, slightly more than NiCd, and unlike NiCd the voltage varies with state of charge so you can estimate state of charge from the voltage. NiCd were notorious for discharging at a fixed voltage and then suddenly dropping to nothing without any warning in literally seconds.
You can get "hybrid" NiMH cells now too, which have a lot lower self-discharge. NiCd self-discharge was also very high, its basically a historical technology now.
The servos, for example would only be active for short periods of time, but in those periods surely drain more than the solar cells can provide.
Then, with the setvos at rest, the battery can recharge until the next servo activity. During night time, the battery should keep the Stamp and Propeller alive. (No servo operartions during the night). The aperatus will ooperate inaccessible to maintenance, thus changing the battery when its empty, is not an option.
It is then the operator's responsibility to ensure for enough battery power left before nightfall to sustain the "hibernation mode".
But maybe I am thinking wrong here. I don't have that much expierience with regulators. They are more than simple voltage dividers, right? So, do they need a constant input voltage, or just a sufficient one?
I could even use a 2.4V 6V or 12V cell.
Of course the prices are somewhat obscene (compared to the NiCa cells). Of course the weights are appealing. Also less volume. This would help a lot.
http://www.conrad.com/Rechargeable-Battery-Packs,-NiMH.htm?websale7=conrad-int&ci=SHOP_AREA_14717_0501043
Do you have a recommendation what cell(s) might be best?
Cheers,
Catweasel
Thanks for your schematic. Just two questions:
1.The whole thing seems a bit clearer to me now, but how did you calculate the resistor values?
2.Is it still possible to hook up a 12V rail fo the servos? Most servos use 12V, or more.
Although, I could use the continuous rotation servos from paralax, which run at about 5V. I'm just not sure if they are strong enough for my purposes.
The specs say: Torque: 38 oz-in @ 6 V. That means its a bit less at 5V. Since 1Oz is about 28g, the motor should be able to winch up a weight of about 1Kg.
That should actually be enough, except if I made a mistake in the Oz->g conversion.
If I can also successfully use the following heating elements, I wouldn't need any 12V supply.
http://www.electroplastics.com/OEM/EP5013r4%20-%20OEM%20Element%20Data%20Sheet.pdf
The expected ambient air temperature during operation could drop as low as -50°C, so I have to heat it.
The whole thing is going to be a "high altitude" balloon probe. So lots of sunlight (hopefully), but rather cold air.
The intended level will be about 8000 and 9000 meters and the duration of the flight about 3 weeks (hopefully that long).
So that is why weight counts and the power management is so important, so the probe survives the night.
The servos will actuate some sampling devices, so they are not in use all the time.
Any creative suggestions? :-)
I am also open to suggestions for experiments to add to the probe.
Currently the list of experiments include:
- wind pattern
- Humidity
- air particle samples
- UV exposure
- Electrostatic charge
Cheers,
Catweasel
Hobby servos are generally powered with 4.8V to 6V. There are some servos (hobby) that can be powered with 7.2V.
Hobby King sells a bunch of inexpensive servos which I think work very well.
Their HX12K servo claims 10 kg*cm torque. These are about the easiest servos I know of to convert to countinuous rotation. The mechanical stop on the final gear can be pulled out of the gear with a set of piers.
Another nice thing about the HX12K is the screw used to connect the servo horn has a 3M thread (into the metal gear). This makes it easier to install other sorts of servo horns which might require a longer screw.
If you need more torque, the VS-11 claims 19kg*cm torque. This is also an easy one to convert to CR. This is a larger than normal servo so it wont fix in normal servo mounts.
It looks like their out of stock on these servos right now. It usually doesn't take them long to get more.
For small loads, the HXT900 is a great little servo. I think these could be modified to CR, but I haven't tried it yet.
When I was talking about a "12V servo" I had something like this in mind:
http://www.parallax.com/StoreSearchResults/tabid/768/txtSearch/motor/List/0/SortField/4/ProductID/788/Default.aspx
I checked out the details of the HXT900. It looks very interesting. Thanks for the suggestion.
Cheers,
Catweasel
You should download the convert program from http://joshmadison.com/convert-for-windows/ for doing unit conversions. It is a small .exe program so no installation is required. Comes in very handy so I keep it on my desktop.
For calculating the resistor values:
The 317 adjustable regulators put out 1.2 volts between the OUT and ADJ pins.
For the one on the battery charger I selected a resistor value that would provide a constant current (assuming the solar cells can supply that much current) suitable for charging NiCd or NiMh AA cells. Typical AA cells are 2500 to 2800 mA hours so should be charged at 250 to 280 mA max. The resistor is calculated by R = V / I, in this case 1.2V / 250mA = 4.8 ohms. The nearest standard value is 4.7 ohms which means that circuit will limit the maximum charging current to about 255mA.
For the other two regulators we want a specific output voltage so we place a 120 ohm resistor between the OUT and ADJ pins to provide a constant current of 10mA. To get a specific voltage out we need to place a resistor between the ADJ pin and ground that will have a voltage equal to the desired voltage – 1.2V.
For the 5V supply that would be 5 – 1.2 = 3.8V. The desired resistor value would be R = 3.8 / 0.010 = 380 ohms. The nearest standard resistor value is 390 ohms which would give us 5.1V out.
For the 3.3V supply that would be 3.3 – 1.2 = 2.1V. The desired resistor value would be R = 2.1 / 0.010 = 210 ohms. The nearest standard resistor value is 220 ohms which would give us 3.4V out.
PS – Just noticed the decimal point is missing on the charger regulator. That should be 4.7 ohms, not 47 ohms.
My concerns with the unit conversion was not the math in itself as in 28*38, but more about a little ambiguity in the Oz definition.
Obviously its not the fluid ounce, but I was simply not sure which ounce was meant (~28g, or ~31g). Though I believe the latter is used only for gold and such.
So correctly stated, I had more a concern with the unit definition, rather its conersion.
And that is inependend of the solar voltage. So the 317 would step the voltage up and down as needed (for example if the solar array only delivers 2.5V), because its current that counts, not voltage?
Sorry if these questions may seem stupid, but that part of how electricity works was left out by our school teacher (re rather enjoey more to impress us with his magnets).
:-)
Cheers,
Catweasel
In redesigning that project to use a Propeller I had to ask questions on the forum. I saved a lot of money by replacing PCBs with code but I had a lot to learn.
I thought of using a Stamp to do prototypes but now it is just as easy to experiment in "Spin".
For the '317 questions, no, the output voltages are not independent of the solar cell voltages. The '317 is a simple linear regulator so the output voltage and current is dependent on the available input voltage and current. The solar cell can only deliver a small percentage of the light energy that falls on it so It acts like a power supply with a series resistor. As the load draws more current the solar cell output voltage drops.
The '317 has a dropout voltage of about 2V which means the input voltage must be at least 2V higher than the output voltage you want. For the charging circuit that means the solar cell must put out at least 9V (battery voltage) + 1.2V (voltage across current limiting resistor) + 2V (dropout voltage) or 12.2V before any current goes to charging the battery. A 9V battery would probably not receive any charge from a 12V solar cell using the '317. Better to charge the batteries directly from the solar cell. A small solar cell will not put out enough current to damage the batteries.