[00940]

Everything in the title. I've a 2X15V transformer and need a very simple regulated 30VDC power supply.

Simply putting the two secondaries together would get me 30VAC, which means at least 40V, even after the diodes' losses. And the lm317 only accept 37VDC in.

Any suggestion ?

 

[Glassman]

doesn't it just the difference between LM317's input and output? it doesn't 'know' where is ground, it sees just the voltage difference across the regulator.. I think it's fine..

 

[Dreamslacker]

Quote:

Originally Posted by 00940 Everything in the title. I've a 2X15V transformer and need a very simple regulated 30VDC power supply. Simply putting the two secondaries together would get me 30VAC, which means at least 40V, even after the diodes' losses. And the lm317 only accept 37VDC in. Any suggestion ?

Just use the LM317 regulator. The 37v max is the voltage difference between the input and the output. Couple with tantalums caps as recommended in the datasheet and you'll be fine.
Cheers.

 

[tangent]

Quote:

Simply putting the two secondaries together would get me 30VAC, which means at least 40V

Only under no load. As load increases, the output voltage of the transformer will drop until it hits 30VAC at its rated output current. To get the 33VAC you need to keep the regulator happy, your transformer will have to be somewhat oversized for the application.

Or, you can just adjust the regulator to put out 27V or so instead. Very few audio circuits are so picky that they absolutely require 30V.

Quote:

The 37v max is the voltage difference between the input and the output.

Yep. The greatest amount of voltage drop in a typical LM317 circuit is across the adjustment resistor, not across the regulator itself.

 

[00940]

Quote:

Originally Posted by tangent Only under no load. As load increases, the output voltage of the transformer will drop until it hits 30VAC at its rated output current. To get the 33VAC you need to keep the regulator happy, your transformer will have to be somewhat oversized for the application.

I don't need 33VAC, I need 33VDC, right ?

30VAC = roughly 42VDC (30 * 1.414), no ?

I completly forgot about the lm317 being a floating regulator, makes me feel stupid

I'll just get the small velleman kit and i'll change the caps.

 

[tangent]

Quote:

30VAC = roughly 42VDC (30 * 1.414), no ?

No.

The 1.414x factor is the peak value of the unsmoothed "camel humps" waveform you get out of a bridge rectifier when you feed it a sine wave. You care about the peak voltage because the filter caps have to be able to withstand the peaks.

The unregulated RMS voltage cannot be higher than the input voltage to the bridge. There's no such thing as a free lunch.

All is not lost, however. Because a transformer is not 100% efficient, the voltage output from a transformer drops as the load increases. Therefore, the voltage spec for a transformer is given at maximum load. For a 15VA +/-15V transformer, for instance, you get that +/-15VAC at 0.5A. Below that current level, the voltage rises, up to as much as about 40% at no load, depending on the transformer design.

If you only need 0.1A out of our example transformer, it may not be unreasonable to expect to get 35VAC or more out of it, which would be enough to keep a regulator set for 30V happy after diode losses and such.

 

[00940]

Getting more and more confused

All designers assume an unloaded xformer then ?

Looking at headwize's library, KG gets +/- 48VDC from a +/-36VAC xformer, Jan Meier +/-15VDC (after a pair of 7815/17915) from a +/-15VAC xformer, Jorgensen +/-24VDC from a +/-18VAC xformer and so on...

 

[guzzler]

the voltage specified is at maximum load

to get the voltage after the rectifiers, times by 1.414 (root 2) - diode drop:

KG: 36 x 1.414 - 1.2 = 49.8V minumum, so fine for the 48V from the LM317/337
Jan Meier: 15 x 1.414 - 1.2 = 20V min, again no probs from 15V out
Jorgenson: 18 x 1.414 - 1.2 = 24.4 min, quite close but didn't check his regulation scheme

g

 

[tangent]

Quote:

All designers assume an unloaded xformer then ?

How about you just hook the transformer up and start playing with it, measuring the voltages as you vary the load on the circuit. You'll see what I'm talking about when you start trying things.

 

[00940]

Guzzler : I know that, see my previous posts, the calculations you give are in line with what I thought. The formulation "to get the voltage after the rectifiers, times by 1.414 (root 2) - diode drop:" is exactly what I used earlier (and actually what the dmm showed me when building a few amps).

I'm confused by Tangent saying I need more than a 15V xformer to get 15V regulated out of it. If Jan Meier gets 15v regulated from a 15v xformer, I surely don't need a 33V xformer to get 30V. Looking at some manuals for Velleman power supply, they ask a 28VAC xformer to get 35V from a LM317.

 

[tangent]

Quote:

I'm confused by Tangent saying I need more than a 15V xformer to get 15V regulated out of it

I'm going to repeat what I've already written, one more time. Read carefully.

A transformer only puts out its rated voltage at full load. At lower loads, the output voltage rises. This is entirely separate from the peak vs. RMS voltage issue. If you use a 30V transformer and you want to get 33V or more out of it, you simply have to get a transformer rated for significantly more output current than you will actually use. This will ensure that it puts out more than its rated 30V because it is not under full load.

All I'm trying to tell you is that you cannot just use any arbitrary 30V supply and expect to get 40V or whatever out of it. It only happens under certain conditions.

As to what is in the Headwize Library, I wouldn't base your design work on that. The transformer loaded vs. unloaded voltage rise is specific to each transformer design. You can't say that all transformers have a 40% rise in voltage when unloaded, for example. A very efficient transformer might have a very small voltage rise. If you had such a transformer and had counted on getting 20% more than the rated voltage, you'd be screwed.

 

[00940]

Tangent :

I fully understand that, based on VA rating and design, a xformer could output more than it's 'official' output if I don't draw all current it could provide. That's not the point.

Let's take a 30VAC xformer at full load of 100mA. It gives me a 42.42V peak current or 27V average. But if I've enough caps in the supply to sustain the load (and for a load of probably around 100mA, I don't need that much), my resulting current should be nearer 42 than 27. Setting diodes losses aside of course.

If the xformer was rated for more than 100mA, I could get actually more like 45 or 46V peak. But it doesn't really matter here, since I'm gonna feed a regulator. For practical matter, it's more realistic to base myself on a voltage of at least roughly 40V.

PS : I suppose all the confusion came from the fact I always assumed a capacitor bank and you didn't.

 

[sanaka]

What I understand:

When we say we're getting "30 volts AC" out of a transfomer, we mean "RMS." This is what a multimeter will read/display. This is exactly the same thing as getting 30*1.414 = 42.42 volts "peak to peak," meaning the AC waveform peaks at 42.42V, 60 times per second (if your mains is at 60Hz.) Just like a normal 110V wall socket is at 155.54V peak to peak.

How much more voltage a transformer gives over it's rated voltage when it's not loaded has nothing to do with the above. It might give about 40% more, or 29% more, or only a few percent more. It might even by freaky coincidence give right about 41.4% more, making it appear like it's some RMS vs. peak-to-peak issue, but that would be a freaky coincidence and not a RMS vs. peak-to-peak issue.

Furthermore, a capacitor bank receives the "pulsing DC" waveform from the rectifier bridge and squashes it down to its RMS value. This means that the current flowing out of your capacitor bank ideally no longer has a peak-to-peak value. The 1.414 factor is then irrelevant. That's Direct Current. This action by the capacitors is called "smoothing" or "filtering" and is why it's called a "filter bank." The secondary function of the filter bank is to act as a resevoir of electrons so that current demands from the load can be supplied quickly. This is because capacitors can give up their charge, hence supplying current, faster than a transformer/rectifier can supply current. This does not affect the voltage at which this is all occuring.

Again, this is merely what I understand. As I have no formal electronics training and little experience, I defer to most anyone who does, provided the constraints of logic appear to be in play

Peace,
Sanaka

 

[aos]

Unless you're really going to use the full rated power of the transformer (and I never had to so far), you shouldn't have problem getting 30V DC regulated from a 30V transformer. As long as you regulator doesn't require too much of the dropout voltage - if so, use an LDO regulator, LT1086 (though I don't recall its voltage limits, check datasheets). If the load is much lighter than the rated transformer current then the voltage on the filter caps will tend to be closer to the peak voltage (1.41 of RMS) allowing plenty of room for regulation. However, since RMS is defined as the voltage that would case the same power dissipation in a resistive load as the same DC voltage (it's simply the integral of the power during a period, divided by the period), if you really load the transformer to its limit then potentially it will drop to its rated 30V, and if your regulator needs a few V then it won't work.

 

[00940]

Quote:

Originally Posted by sanaka Furthermore, a capacitor bank receives the "pulsing DC" waveform from the rectifier bridge and squashes it down to its RMS value.

Hum, no.

Looking back to my old electronic course (what I should have done since the beginning), the caps of the filter are loaded at peak value. When the current coming from the rectifier disappear, the caps discharge slowly, near peak value. As switching is fast, the caps don't have the time to unload a lot and thus the resulting voltage should be higher than rms.

What I completly forgot is that as the load is nearing the rating of the xformer, this one is unable to load enough the caps and thus the voltage goes down, as it would goes down with undersized caps. In fact, I practically never noticed that since I always tend to oversize the xformer by a factor of 3. Actually, since almost all diy projects I've seen so far do so, using the 1.414 factor to get an idea of the resulting voltage is practical if not exactly true.

And it has indeed nothing to do with the fact that an unloaded xformer could give more than it's official rating.