Power Supplies again ;/

This might be a stupid question but I was wondering why one can't use transformers in series and/or parallel with different turn ratios to step down the mains to different levels... like 117VAC down to 50VAC, 36AC,

24VAC, 15VAC, 12VAC, 9VAC, 5VAC, 3VAC, 2VAC, 1VAC, or what ever one wishes. One can then rectify each one(ofcourse for n output levels there would n 2*n(for + and -) multiplicated circuits) to get DC approximately at those levels then use a regulator to be able to vary around those levels(so one could get 13VAC if they needed it).

Wouldn't this be more efficient(since transformers tend to have high efficientcy) than just using one regulator to vary over an extreme range? or is the cost and complexity in doing this just not worth it? (Although couldn't someone make a transformer with multiple taps that basicaly will convert 117VAC down to these multiple levels and one could just use one rectifying and regulation circuit that is "manually" switched onto the taps(by some rotary switch, say)?).

Just wondering,

Thanks, AD

(top posting fixed)

If you want to waste that much weight in iron just get a variac.

You seem to misunderstand the use of a regulator -- it is quite rare that you'll see one in an AC circuit. Any time you see "regulator" in an electronics context you can be 99.44% sure that it's a "DC regulator".

There are linear regulators which work by burning power -- ideally a linear regulator has as much current going into it as coming out, and a constant output voltage. Any difference between the output voltage and the input voltage goes toward heating the regulator's pass transistor, so you throw away an amount of power equal to (Vin - Vout) * I.

Harder to make, but more efficient, is a switching regulator. Ideally a switching regulator has as much power going in as coming out (i.e. it's

100% efficient). In the real world 90% should be quite achievable, 95% _can_ be done, and everyone claims to have built one that's 98% efficient.

Switchers have been used for years in computers -- that's what PC's have been using all along. They're getting smaller, more efficient, and (thanks to the semiconductor companies) easier to design.

If efficiency is what matters build a switcher.

--

Tim Wescott
Wescott Design Services
http://www.wescottdesign.com

Well, one certainly can, after all there's a physical *law* in place there. ;-)

If you want a continuously variable range, you'd use an isolation transformer followed by a variac.

Absolutely, if you mean linear regulator. Switching regulators, however, are essentially as efficient as power transformers, but you do have added complexity. But this keeps EE's in business, rather than the transformer winders.

A variac is essentially a transformer with one tap on every turn of the winding. ;-)

Tim

-- Deep Fryer: a very philosophical monk. Website:

what about of power? I didn't specify but I'm talking about 1 to 2 amps. Like what I was thinking was having one "main" transfomer that was a one to one and center tapped that could handle an amp or two... then off the secondaries have 2 more that are center tapped(for polarity) say with another 1 to 1(which since each step cuts the Vin by 1/2 I'd get one with +-

1/4Vin) and maybe one with 1 to 2 step up.

hence I could get +-1/4Vin and +- 1/8Vin. for 117VAC that is about 30VAC and

15VAC... then I could use a regulator to vary those two and get more efficienty(if I understand that regulators have good efficientcy when they don't have to attenuate the signal much). I could add another transformer to get down to 5VAC.

I suppose though it doesn't matter to have them like that since its better just to find a transformer that will step up with the right ratio that one wants..

i.e., I'm doing 1/4 by two transformers when its just more efficienty to find a transformer that has 1 to 4 turns ratio instead of "cascading" them to get it(which I guess one would have to do if they wanted a very specific ratio for some reason and couldn't make transformer for it).

Maybe my real question should be "How efficient are regulators"? I read that its like 30-40% but is that the worse case? Why would it be so inefficient when regulating a voltage without attenuating it any(i.e, you put 15V in and get 15V)?

I suppose its easier though just not to worry about that now and build me a damn regulated PS so I have something to use ;) I was just thinking that I could increase the efficientcy if I could get the input to the regulators near what I would actually use them at instead of having such a wide band of attentuation... i.e., if I wanted +-50V down to +-1V the regulator would have to attentuate a huge amount to get down to the 1V when if I could split it up into two regions with two regulators each regulator wouldn't have to attenuate that much. Maybe this is not cost effective though(although I am using it for personal use so the extra money for the components might be worth it compared to the wasted power?)

Thanks, AD

You can connect transformers in series to do what you want. You can also connect secondaries in series/out of phase to yield the difference between the two secondary voltages. You must be sure that the load current is no greater than the current rating of the secondary with the lowest current rating. ~ You must be very careful when you connect transformer windings in parallel. Two transformers that are absolutely identical will share current. If they are not identical, they will not share equally. You must do a careful analysis taking into account voltage tolerances, winding resistance, inductance and leakage inductance. Usually the information you need is not available from the manufacturer (except as "maximum" values), so you would have to measure these values. Bottom line: Its usually not worth the trouble. Regards, Jon

no I'm not. I do mean for DC.. but after the fact. I'm talking about the outputs of the transformers for AC then I will convert that into DC with a regulator.

I've read that they are about 30-40% efficient compared to transformers being 95% and SMPS's being 80-85%.

From what I've read about regulators is that they are basicaly use variable resistance to "smooth" out the signal. The data sheets I've seen let for one regulator let let a 48V input vary by 1.25*(1 + R2/R1) + I*R2 and R1 was set at 240 ohms. Hence 48V's going in and 1V going out means that theres a lot of power wasted through its resistors(assuming that?

I want too, but it is to complicated at this point I think. I will eventually but I'd rather work on something that is easier like a regulator but I don't want it to be 30% efficient as I feel its a waste. Surely if I only needed +-5 VAC(and then convert that to DC by filtering and maybe a regulator) at about 1 or 2 amps then I can just get 117/10 ~= 11 to 1 step down transformer...

I could then build several power supplies giving me a several voltage outputs like +-5, +-9 that are all efficient(maybe even more efficient than SMPS since it's pretty much just a transformer doing all the work?)... I'd rather have all that in one PS unit though sorta like how a computer PSU have +-12 and +-5(although at a much greater power than I need).

I know I can use a regulator(LM337 I think or something like that) that will let me go from about 50V to 1V or so but at the cost of being extremly inefficient(from what I read... I could be wrong though.).

I'm thinking that maybe I shouldn't worry so much about the efficiency at this point though and just build a regulator PS for now and then move on to switching later

Thanks, AD

I've looked at variable transformers and from what I saw they are quite expensive and have a small "range"(except maybe the more expensive ones). Maybe I'll look some more though

yeah, ultimately thats what I would like to have... not necessarily every turn but just to give me a nice resolution and range. I was thinking of kinda approximating this by using several transformers(or one that is multi taped for specific values).

i.e., I would only be potentially using a "areas" of the output of a variac and don't need to have it continuous over its whole range.

so I have Vmain and I want to convert it to {+-V1out, +-V2out, ..., +- Vnout} for n about 3-5 and the current is at most 5 amps(but more likely just 1 or 2).

I figured once I could get those ranges out then I could use a regulator to get a continuous range around those outputs... basicaly

Vnout +- e where e is maybe at max 1 or 2 volts... the regulator should be much more efficient(I could be wrong here though) in just a small neighborhood around the each output instead of a much larger one(where it could attenuate up to, say, 30-40 volts).

Say on my PSU I could have a rotary switch that lets me select the Vnout then another "fine" tune knob that lets me select +- e from Vnout(it would be tuned to give me the "standard" values and a range of 1 or 2 volts(maybe even have an extra switch for this).

newark has the 3PN-MK for only 14$. Then they seem to jump up to about 50$. I seem to feel that sometimes they is some mistake on the listing cause such drastic price differences(I'm seen it with other things such a PSU for 10$ then it jumps up to 50$ or so). It seems the main difference is the amps they can do. Do you think getting that 14$ variac would be good enough for what I want(PSU that does about 1-50V's... I guess I would still have to convert it to DC though... Maybe I can tear it out of its case and put it in mine or something and add the rectification and regulator).

Thanks, AD

hmmm... seems that 14$ thing is such two pieces of metal(a bracket I guess)... sheesh.. glad I didn't order it ;/ On newark they just call it a freaken "variable transformer" ;/

?Ntt=variable+Transformers&searchtype=keyword&Nty=1&N=0&Ntk=gensearch&&&showImages=false&comSearch=null&paramSearch=false&viewType=&Nso=0&Ns=PRICE_PLS_001_PRICE1#results

AD

60 Hz transformers are *huge* to carry the amount of power at such a low frequency compared to switching transformer counterparts. The cost associated with 60Hz transformers is significantly greater than 100kHz tranformers without a significant effeciency difference: they're both transformers, afterall.

My commercially built bench power supply uses that approach - it switches taps on the power xformer in response to voltage adjustment in order that the voltage adj circuitry only has to cover a limited range.

DAvid

Abstract Diss> This might be a stupid question but I was wondering why one can't use

Cheers! Rich