PSU question

December 3, 2006 10:34 PM

Electronics filter: What can I do to make a 24v PSU turn into a 18-19v PSU? It's 0.5A power supply which as far as I can tell is important to know when using, say a resistor to lower voltage. But how resistant should the resistor be? I've found information on how to tell a resistor's value, but not how it will actually effect a line's voltage or how to calculate it. Are there any good sites with this information? Also what will happen to the amount of amps the PSU supplies when I do change it to 18/19v?? Also, what can be done to increase the amperage of a power supply? What inside the circutry is actually dictating the amperage?
posted by Napierzaza to computers & internet (25 comments total)

A resistor is a really bad way to lower voltage for many/most applications. What is the application? (Ie how much currant does it draw from the PSU, and is it always the same, or does it draw more when doing certain kinds of operations?

An easy way would be a string of diodes in series - each diode drops the voltage by about half a volt (0.2 - 0.7, depending on kind of diode vs kind of power running through it). So keep adding diodes to the string until the voltage reads 19V. Then check that it is still 19V with the load attached.

Also - the amount a diode drops the voltage will vary based on the ampage running through them, so if the application does change its load, then you may wish to minimise the voltage fluctuation this may cause if the application is voltage sensitive. To do this, you can use diodes that are way over-spec on current (ie designed for much higher current - 20A, for example), this way, even large fluctuations in the power going through the diode are still within 5% of what the diode is capible of, so the voltage fluctuations will be similarly minimal.
posted by -harlequin- at 10:41 PM on December 3, 2006

you'll want to make a voltage divider to knock the voltage down. note that using a voltage divider means the actual voltage will depend to a large extent on what you connect later.

the current (not "amperage") rating of the power supply shouldn't make a difference in the behavior of a voltage divider, unless your circuit draws enough current that you're approaching that limit.

you can't do much, really, to increase the current rating - most normal power supplies (i.e. those little black things you plug into the wall and then plug something into) are given a rating based on how much power they can supply without burning themselves up. this is generally a function of the size and quality of the components inside the supply.

you can probably try to exceed this rating by connecting your supply across, but i wouldn't be surprised if it:

a) has a fuse inside that burns out when you exceed half an amp, for safety reasons,
or b) will actually output that much current for a while, and then either stop working or catch on fire (see a).
posted by sergeant sandwich at 10:43 PM on December 3, 2006

sorry, i dropped some words: by connecting your supply across a low-resistance load
posted by sergeant sandwich at 10:45 PM on December 3, 2006

Easier and more effective than harlequin's suggestion, if the load you are driving won't need the full current of 0.5 A, is to set up an external precision regulation bridge, using an appropriate zener diode in series with a properly sized current limiting resistor. You'd need a 19 volt zener, rated for something up to 0.5 A (depending on how much current you want to pull out of the supply constantly through the zener regulator when your load isn't operating). According to Ohm's law, a 24 volt DC supply outputting 0.5 A would be fully loaded by a resistance of 48 ohms, producing 12 watts of total power to be dissappated in the load. Putting a 10 ohm resistor of 5 watt rating in series with your 19 volt zener as a current limiting resistor would be approximately correct. Connecting your load across the zener, as more current is diverted by your load around the zener, the zener acts to internally vary its own resistance, trying to maintain it's 19V zener potential, so long as it can do so while remaining on it's operating curve. If the resistance or impedance of your load drops way below 38 ohms, it will suck all the current through itself, leaving nothing for the zener to operate on, but this is probably the least of your worries, at that instant, as your 10 ohm current limiting resistor will be getting pretty warm, and any internal overcurrent protection in the PSU (fuse, circuit breaker, crowbar element, if so equipped) will be tripping.

As for your other questions, the internal components controlling the voltage available from the supply are typically the line transformer and rectifier bridge of a standard DC supply. In switching type DC supplies used in most electronic gear these days, the duty cycle of the series switching element is varied by the control electronics to create a near constant output voltage across the output filtering capacitors. Sometimes, these kinds of supplies can be reworked for different output voltages with minor changes of component values in the control circuit, albeit with usually somewhat detrimental effects on the degree of regulation attainable, if the filter capacitors and inductor are not changed.

The internal components limiting the ameperage available from the supply depend on the type of supply. In a standard simple DC supply, it is usually the rectifier bridge. In an overcurrent protected supply, it is the series protection crowbar element. In a switching supply, it is the series switching element, or if overcurrent/overvoltage protected, the crowbar protection element.
posted by paulsc at 11:16 PM on December 3, 2006

You don't want to make a voltage divider if you're drawing enough power that you're worried about needing more than 0.5 A. You'd need to track down power resistors for one thing (not rare, but not 1/10 cent items you see everywhere like 1/4 watt resistors), they'll get hot, the voltage is very dependent on the load, and it's wasteful. The diodes idea is better, although just like you can't use just any resistor you can't use just any diode. Another option is to get a voltage regulator like this one, which is just a three-lead input (24V), output (18V), and ground deal.

What's the application? 18-19V is a sort of weird voltage, I don't know that I've ever seen it used before.
posted by TheOnlyCoolTim at 11:23 PM on December 3, 2006

You can use something like an LM317 voltage regulator to regulate the power supply's voltage down to whatever voltage you're looking for without burning up quite as much of it as heat as the zener solution would. LM317s (and clones) are cheap, reliable, available everywhere (even radio shack sometimes), and can be set to an arbitrary output voltage depending on the values of the two resistors you attach.

If the load is going to draw a known, constant current, then yes, you can just put a resistor between the load and the PSU (the voltage drop across the resistor is equal to the current times the resistance). But probably the current drawn by the load isn't constant, so you'll want an "automatically self-adjusting resistor", aka a voltage regulator.

What are you intending to use the 18-19 volts for? There might be an easier solution.
posted by hattifattener at 11:52 PM on December 3, 2006

Easier yet, use a LM317 voltage regulator which you can get for less than a dollar at Digikey.com. Datasheets are also available there.

You can easily set the output voltage to whatever value you want with a couple of resistors, one of which can be adjustable. This supply has much better load regulation than a zener regulator and you don't need any big honking zeners or power resistors. The maximum output current will be the same as your input PSU, 0.5A. If you want more current, the easiest solution is to get a bigger PSU.

(By the way, Paul, I think you intend to refer to a current limiting circuit. A crowbar circuit is designed to handle over-voltage, not over-current.)
posted by JackFlash at 12:05 AM on December 4, 2006

Sorry, hattifattener. I guess I'm a little slow on the draw.
posted by JackFlash at 12:06 AM on December 4, 2006

Count me in to the "use a voltage regulator" camp. A voltage divider isn't a good idea when you're talking about substantial amounts of current, and I think that wiring up a LM317 from its datasheet will end up being a lot less complex than going the Zener route. For a dollar or so an adjustable regulator and a few filtering caps and a heat sink will get the job done and save you a lot of messing around.

Last time I checked, "good" RadioShacks (ones that still have a rack of electronic parts, as opposed to the crummy ones in malls that only seem to stock cheap PCs and AA batteries) used to have LM317-type regulators. However a search of the RS web site doesn't turn anything up, so you may be looking at mail-order from Digikey/Mouser/Newark if you don't have another source locally.
posted by Kadin2048 at 12:47 AM on December 4, 2006

I think you are better off just getting an 18v psu.
posted by caddis at 4:41 AM on December 4, 2006

I've made my own power supplies, and limiting one is essentially like making the "back half" of one. Check out this plan: Small Wart 200 power supply. It makes a .2A 9V (adjustable from about 12 to about 2V I guess). You'd only need everything after the rectifier and I think you could probably skip some or all of the filtering caps.

Note that it also uses a resetable fuse - mouser probably has these in .5A versions also. They're about 50 cents and they basically conduct fine until their amperage is exceeded and then they switch to super high resistance, essentially shutting the device down. When you turn everything off and back on they reset to low resistance. You could also use a fuse holder and a traditional .5A fuse.
posted by RustyBrooks at 6:28 AM on December 4, 2006

Okay, thanks for all the answers everyone, still learning a lot about electronics!

I found a HP p3 1Ghz pc (one of the really small kinds) and I've been trying to get it working. It's missing a lot but the most important missing part is of course the PSU which is rated at 18v. So I'm idly looking for a PSU and I have a 24v one from a printer and figure I might be able to mod it so I can set it up for cheap.

I can't recall how much draw the computer takes, but it sounds as if I will have to find out.
posted by Napierzaza at 8:15 AM on December 4, 2006

As long as we're being thorough.. :)

The LM7818 is a one part solution. At full load, you will be dropping 6V at .5A = 3W, which means you will need a heatsink. Just bolt any old hunk of aluminum - 1" x 1" x 0.1" should be enough - to the little hole in the TO-220 package.
posted by Chuckles at 8:26 AM on December 4, 2006

Ah, 24V at .5A will not be enough for what you are trying to do. It will require between 2 and 5 amps, I'd think (possibly a little more even). I think we should be able to dig a number up for you..
posted by Chuckles at 8:29 AM on December 4, 2006

On second thought, before I go searching, you better post the model number of that computer. Don't worry if it seems convoluted. Also, if you are lucky, you are just not noticing where it says the current requirement. Try looking again in the area where you read the 18V requirement from. Or, post pictures of any writing on the case..
posted by Chuckles at 8:32 AM on December 4, 2006

Here you go..

HP e-PC PIII 933MHz @ 1015MHz/256MB/6.4GB. It uses a laptop style power supply whose max rating is 19V x 3.16A = 60 watts.

and the search I used.
posted by Chuckles at 8:40 AM on December 4, 2006

Okay, I'll have to go home and check later. I'll post the name of the computer then (hopefully thread won't be dead).
posted by Napierzaza at 8:41 AM on December 4, 2006

Wait, you have a printer power supply rated at 0.5 amps and you expect to run a computer with that, and you will be wasting some of those precious amps regulating voltage? From what you have told us this power supply is undoubtedly too small to power a computer. I could be wrong and HP may have designed an entire computer capable of running on a 12 watt power supply, but I doubt it.
posted by caddis at 10:51 AM on December 4, 2006

Better question: should I get a cold-heat solder iron? Seems like it might be expensive, not sure if it'd be better than my regular iron. Will it prevent the stripping the lamination from the PCB?
posted by Napierzaza at 3:35 PM on December 4, 2006

The tip of a cold heat iron is very brittle, and you have to hold it very very carefully to get it to heat at all. On top of that, the solder doesn't flow around tip/lead/PCB/whatever the way it should. It's possible I didn't give it enough of a chance, after I broke the tip, I took it back, but I think they are probably crap.

If you are damaging PCBs with a regular iron, something is wrong.

1. It isn't hot enough, which causes you to hold the iron in place (and often press hard on it),
2. it is too hot, which burns the PCB,
3. the tip is worn out, which has a similar effect to having too little heat,
4. the iron isn't powerful enough to adequately heat what you are soldering - for example, multi-layer boards, where you have to heat up a lot of board area, and often hidden ground planes- the effect is similar to not enough heat,
5. you just need more practice.

Anybody have experience with lighter fluid (butane?) powered irons?
posted by Chuckles at 4:20 PM on December 4, 2006

With all due respect Napierzaza, your comments here show that you are very much a beginner here. There is nothing wrong with that but I think you are biting off more than you can chew and chasing the wrong stuff with exotica like cold heat irons. I suggest you build yourself a kit or two which can quickly teach you an awful lot about electronics. This amplifier (there is a nude photo which doesn't show that much anyway of the designer's wife on the page) from Bottlehead is easy to build and comes with a guarantee to make it work even if you screw up building it.
posted by caddis at 5:15 PM on December 4, 2006

Hey man, I think you're over exaggerating there caddis. First of all, I'm not burning any PCBs, that's an assumption. Also the only reason I'm asking about Cold Heat is that I am presently making a thinkGeek run with my brother and was considering adding that on IF it is an actually useful device. I am certainly novice, but not in the mechanics of soldering, I just need to know more on how the components work.
posted by Napierzaza at 7:58 PM on December 4, 2006

Okay I read the S.E.X page and got very interested, until I read it was 369$!!! Is there a parts list anywhere?
posted by Napierzaza at 8:09 PM on December 4, 2006

With Bottlehead you have to buy the kit to get the schematic, but then you also get all the parts and the detailed instruction manual and the guarantee. I pointed out Bottlehead only because they are the absolute best place I know for people who have more desire than knowledge. They teach.

You can find some schematics online and build your own stuff, like this Chu Moy headphone amp or a Darling amp (I hesitate to link to any particular one as there are so many variations, just search for "Darling Amp*"). The Chu Moy will take your iPod to the next level with the right pair of headphones. The SEX will literally blow you away.

A good background in electronics helps. If you aren't going to take college courses, and you do not need too as the basics are not that complicated, then I recommend getting the Navy NEETS electronics books. (online version here) They are written for the person who has a technical mind, but lacks a technical education. Study these and you will know your stuff.

By the way, that last link is valuable. These things have not been easily available online lately so you might want to snap them up while they are. Being government documents they are not copyrighted and available to the public, but there has typically been some charge just to cover the copying or hosting fees. Modules 6 and 7 will answer all your power supply questions.

I never thought you were burning pcbs, but guess what, it is a lot easier than you think.
posted by caddis at 9:22 PM on December 4, 2006

OK, here is one of many Darling links.
posted by caddis at 9:26 PM on December 4, 2006