I'm going to build a power supply for my projects and I was wondering a few things. One thing is simply does it matter if the filter cap is to large? Most power supply circuits I've seen use around 2200 to 4700uF's and I was wonder that it couldn't hurt if it was larger as it should "filter" it even better? Obviously if its to large it would take longer to charge but after that it should work fine? I'm not talking a drastic increase in the capacitance but maybe 2-10 times as much?
Another thing is about regulation and overvoltage protection. Can anyone recoment some sites that discuss then for use in power supplies? I've seen several schematics of basic power supplies and it seems pretty simple but I'd like to know a little more about it just to be sure. (i.e., is simply a zener diode good enough for the overvoltage protection?) what about other cheap ways to make the power supply more safe?
If you are looking for a lab-power supply, try doing a search on "723 power supply schematic" on Google. The 723 has been around for decades and is still one of the best choices around for 0-36 volts 0 to 10 amps or more.
for a complete project, PCB and explanation. 3-30V 0-3A
The point of the filter cap is to prevent the supply from drooping below the point where the regulator can no longer regulate. Now, with a typical diode bridge scheme after a transformer, what happens is that the cap gets charged up to the peak voltage - 2x the diode drop twice a cycle. So, if you want a maximum of I amps, and the peak is Vp, and the minimum voltage you can have before your regulator drops out is Vmin, then an approximation is given by
I = C * (Vp - Vmin - 2*Vf) * 2 * AC_frequency.
C = I / (Vp - Vmin - 1.4) / 2 / AC_frequency
As an example, if you want a 10 amp max supply, and your transformer gives you 21.2V peak voltage, and you want a regulated 12V, using a regulator with a 3V dropout, then the smallest C you can use is 10 / 4.8 /
2 / 60 = 17.36mF
You can do the math for your situation.
Using a cap that is bigger than you need is not really an issue, except that it'll store more energy (make sure you use a bleed resistor of like
100k to keep it from storing power over long periods). Also, the cost of caps goes up quicker than the rating.
A zener isn't really adequate unless you have a tiny supply, because if your supply can provide a large amount of current, then the power dissipated by the zener is Vz * I. If I is 10, for example, and Vz is 12, then you need a zener that can dissipate 120W until the fuse blows. If it fails before your fuse blows, your circuit is toast.
A better way is to use what is called a 'crowbar' circuit, which consists of a device called an SCR, and a zener reference. When the zener conducts, the SCR turns on, and it CAN handle the current until the fuse blows. There are other schemes as well, search for "crowbar" and "SCR".
If your regulator is meant to be adjustable, then you need to figure out another way to limit current. Often, regulators do this for you, using a scheme called 'foldback' current limiting, which just means that if you short the output, it won't give you the full current you've designed, but will 'fold back' the amount of current to a far smaller value.
As far as regulators go, you can get one that can handle up to an amp at radio shack. That isn't too bad, actually. It is named an LM317, and can regulate voltages from about 1.5V up to 34V, using a few resistors. There are also fixed regulators, like the 7805, which regulates 5V with no external components (well, maybe some capacitors to prevent oscillations; look at the package).
If you are looking for a general hobby supply, however, I'd buy one. you can get them that have adjustable voltage, and also adjustable current, which is often useful. They often have a display of the voltage and current, and various other features. I have one of these:
I like it, but are various issues with it; the pots they use are fairly cheesy, and tend to be hard to adjust. Also, they don't have 'kelvin' leads, meaning that the regulation happens at the supply, and not at your circuit. Better supplies will allow you to have 4 wires, two for the current, and two for the regulation. Not doing this causes issues with higher current applications, because resistance in your wires drop some voltage.
You can also occasionally pick up nice bench supplies on ebay.
Everyone else has already said "it doesn't matter as long as it's big enough" but it certainly can matter. As you make the capacitor bigger, the conduction time of the rectifier diodes gets smaller and smaller, and while average current stays the same, the peak current goes up.
As the peak current goes up, not only do you have to worry about the peak current rating of the rectifier diodes, but also the I2R losses in the transformer. What this means is that if you use a 500mA transformer but you're only conducting a few percent of the time, you have to greatly derate the current capacity of the transformer and you might only be able to get 100mA DC current (or less) out of your supply.
Actually the conduction angle doesn't drop as fast as you might think because the resistance of the transformer windings (and also the ESR of the capacitor comes in) limits peak current for you. And as the transformer heats up, the winding resistance goes higher and prevents thermal runaway.
But for a typical hobbyist lash-up where you are not running at even
20% of the rating of the transformer and everything is in a office or bench-top environment you probably will never notice.
After you get to currents above a few hundred mA the dynamic resistance of the zener makes it ineffective, and you have to go to a SCR type of crowbar.
Safe as in protecting the load, or safe as in not catching fire? A properly sized AC fuse is always the first link in stopping it from catching fire! Make the capacitor too big and you won't be able to put in a properly sized AC fuse (getting back to your first question!)
I think I see this but I'm not completely sure. It would seem that the current initally when the cap is discharged completely would peak very high(as the cap would be short) but once the cap is charged it won't draw any more current than a smaller cap that is not to low to discharge completely, right? So basicaly one would be storing excess charge and it would be wasted in leakage but it should act very similar to a smaller capacitor that has enough charge on it to do the same thing? If there is a problem with the peak current at "power on" then one could surely fix that with some other device? (maybe resistor is not a good idea but some current limiting device might work(a zener in parallel with the cap?)?)
Would be nice to get some good indepth theoretical information on this. I have some radio shack book I brought a long time ago on power supply design but I can't seem to find it(which seems to happen when every I need something ;/)
ok, so I would need to figure out what the regulator can do then get a cap that can handle it and maybe just double it just to be sure?
yeah but wouldn't this bleed resistor be wasting power? Maybe just have a method that when the power switch is shut down it would discharge the cap to ground through a resistor? or is there some other saftey reason not to have such a large cap... say if my fuse is before the cap there is a short then that fuse won't do much good?
I found a schematic that talks about this... he uses a thyresistor or something to prevent this sorta thing. I'll have to look into it more.
yeah, I think that was the circuit he used for this.
yeah, that would be nice too. I'll have to look into it.
I've seen this in a lot of circuits I've seen. I'll probably end up using it once I figure out everything I will end up doing.
yeah, I was going to buy when then they seemed sorta expensive. I'm not sure why they are so expensive though and thought that I might as well just build one as I've always kinda wanted to do this anyways and it would be a good learning experience(and then I could customize it even more like put a signal generator on it too and have some other stuff).
The main issue I kinda have is that I've seen a computer power supply that has all kinds of crap in it and it kinda scares me cause I figure that all that stuff is needed to make it safe... I don't want to make a power supply thats going to kill me or ruin my circuits. But I'd rather build one if I can make it 90% of one that I could buy for 50$.
Basicaly I want something like +15 -15 variable down to +1 and - 1 (maybe in steps of like 15, 12, 9, 5, 3, 2, 1 or something like that).
And its not so much that I can't build one its just I don't know if I can build a safe one and I haven't found anything about making a good and safe power supply(all the technical details).
The larger capacitors will discharge less over each 50 Hz cycle and thus shorten the conduction time of the rectifier diodes. This causes the peak current in the rectifiers/transformer to rise as the average current remains constant. The rectifier- and conduction- losses in the transformer are proportional to the current *squared* .... so the bigger the cap, the worse the load on those components. Larger ripple will of course cause losses in the regulator but here the peak current is much lower so it does not matter so much.
So, the O.P. will want the *smallest* cap he/she/it can get away with; your formulae answer the "how small is that?" question.
I once had some old papers, i.e. 50'ies, on the lost design equations for linear power supples. Maybe they are on the net by the mercy of some misconfiguration of a $$$PAY$$$ per view server where all research litterature are now put to eternal sleep?
The radio amateurs handbook (ARRL??) - I Think it is - is strong on practical stuff too.
I was thinking about messing around with SMPS's. I have an old computer PSU that was damaged by something(not sure but it basicaly seems to "give out" after some use if I remember correctly(about 2 years old)). I've ripped it apart and found the TL494CN and an LM339N along with a crap load of capacitors, inductors, and resistors. There are some other ic's that are hooked up to some heat sinks(it looks symmetric so I guess one handles the + and the other the -). I'm still in the process of tearing it apart. At the very least I can use the parts to make an unregulated psu(I think... Most stuff seems to be there and I can always order the rest).
What scares me is the number of capacitors(about 20), resistors(about 100), and inductors(atleast 3 but I think there are some "chokes"(I think thats what they are called... more cylindrical type of inductors but they could be resistors or something)). It also has 3 transformers and about 10 transistor like things. While I'm sure its because its a SMPS I just get confused when I see all that stuff going on. I'm going to try to figure out what its doing though to some degree but it looks much more complex than any circuit I've seen for SMPS's on the internet.
There certainly is a lot to learn about switching supplies, but starting with an offline switcher that has 300VDC in it (like a PC-clone supply) is not the best idea.
Instead make some linear supplies, then go to offline switchers based on TL497's or the NatSemi "Simple Switchers" like the LM2574. And when you do go to switchers please probe around with a scope, and if at all possible do it with a current probe to learn the most. Then you can start tinkering with offline switchers.
Horowitz and Hill (2nd edition) does a lightweight summary of a real offline switcher, it's recommended but it's far from enough to start building one from scratch.
Assuming you are designing a simple full wave rectifier-filter type supply, here are some simple rules of thumb that serve to provide a starting point for filter capacitor selection. Let C = filter capacitance (Farads) Let R = load resistance (Ohms) Let F = ripple frequency = 2 X the power line frequency.
2*PI*F*R*C = 10 provides approximately 10% Ripple and 10% no Load to Full Load DC regulation. Scaling C up by X% scales the ripple and regulation down by X%. Remember, this is just a rough rule of thumb.
Assuming you're making a linear power supply, bigger cap's won't hurt. Lots of caveats, but it won't hurt. For the overvoltage protection, a zener (or zener used to turn on an SCR or transistor) across the output is probably okay (assuming the tolerances in the zener turn-on won't stop the thing conducting before the load is destroyed), but make sure you have a fuse in the output which blows in time to protect the load. Or use the SCR to turn on a relay which disconnects the load. Depends on what complexity you want. Keep Googling and get a feel for what complexity you want/need.
It is ok to raise the capacitance, up to a point. However, you can have too much capacitance, which will stress the recetifiers and transformer. Setting those issues aside, more capacitance won't hurt the ability of the supply to regulate, and will reduce ripple at the input to the regulator.
"simply a zener diode" is not good enough for overvoltage protection for a typical project supply. The "standard" OV protection circuit is an SCR, fired when the zener voltage is exceeded, that shorts the supply output and causes the fuse to blow.
An adjustible current limiting circuit is a nice addition.
It is indeed true that inductor-input power supply filters are much kinder to the rectifiers, with the inductor doing a great job of smoothing out those current spikes.
But for 60Hz power supplies and currents in the hundreds of mA, the inductors required are several Henrys to tens of Henrys. They're big and heavy and expensive. They have appreciable DC resistance which leads to appreciable voltage drop across them. They don't do very well if the load current varies widely (after all, the inductor's job is to keep current constant). In extreme cases of load current changes (for example going from 200mA to 0mA) the inductor will try to keep current constant by raising the voltage across its terminals - if you aren't careful you can build up a few hundred to a few thousand volts, which will arc over somewhere.
Look in any radio book from before the 1970's and you will find several pages devoted to inductor-input (probably the term used is "choke-input") power supply filters.
For most (but not all!) applications today, it is far cheaper to beef up the rectifier diodes to handle the peak current rather than try to smooth 60Hz current spikes with a big massive and expensive inductor.
But go to 60kHz instead of 60Hz and you're in the realm of switching power supplies - the principles are similar, and you will find lots of inductors and magnetics in any switcher.
Well, I was thinking about the problem and I understand it when it comes to capacitors(except I'm not sure how much it is an issue with my case or not(since its only 60hz and < 50Vs).
But then I thought about using an inductor. Do you know if there are similar problems with it(besides the size of the inductor needed)? It would seem to me that the inductor would cause another problem in an "opposite" way that the capacitor does(but really its not the capacitors fault but the diodes and hence maybe the inductor works better with the diodes?)