Will this PSU work?

[syn]

Hi

I need +/-12 VDC & +/-15VDC from a dual secondary transformer. Would this concept work?
Any pitfalls?

Thank you

 

[moamps]

Yes, there are few problems with that schematic. I would rather use the LM317/337 regs with additional power transistors to achieve the required power/current.

 

[syn]

Thanks

.

 

[Bo Deadly]

You don't need separate bridge rectifiers or filter caps for each secondary. Just use one set to feed both regulators.

Correction: As ccaudle pointed out, you should put secondaries in series, use the intersection as 0V and then regulate down to +-15 and +-12.

Or, if the +-12V current is not too high, you can just chain the 15 into the 12 (and parallel the secondaries).

Note that most circuits designed for +-15V will work perfectly with +-12V. And most circuits designed for +-12V will probably work with +-15. Although regulating down from 18V to 12V will dissipate a lot of heat if the current is high.

Finally, a discussion of power supplies would not be complete without at least mentioning that SMPS, when implemented correctly, are superior to linear supplies in just about every possible way.

 

[ccaudle]

I would join the secondaries in series, and make a center-tapped full bridge rectifier to a single set of capacitors, giving a +/- 25V (approximately) supply.

From there it depends on the current requirements and noise requirements of your 12V and 15V supplies. You could regulate down to 18V, then generate 15V and 12V from the intermediate 18V supply, giving better regulation by using two stages.
You could possibly just run everything from 18V if the downstream circuitry can handle that.
Is the circuitry is not using much current, you could just go directly to 15V and 12V from the 25V supply, but be careful of power dissipation, that is a lot of voltage across the regulator (especially when you take into account incoming AC tolerance, you would generally want to design for at least 10% high, and give some margin above that for component reliability).

 

[syn]

This PSU is for a modular synth that requires 2-3A per a voltage tap. I need to use positive regulators only to achieve that, and why I need separate bridge rectifiers and filter caps for each secondary. I was hoping to save on not getting one more 2x12-15VA/4A transformer for +/-12V hence the schematic. Even though the schematic simulates well (LT Spice) I was worried about joining both "negative" regulator outputs at the positive regulator gnd in order to achieve negative voltages (-12 & -15).
Space is tight and I need low parts count, no way I can add 8 power transistors to a 317/337 combo.

Screen shot is somewhat blurry, ignore that and part choices and values, that's no problem, it is just the concept that I need to check.

For LTSpice users PSU.zip contains asc file

 

[Bo Deadly]

This PSU is for a modular synth that requires 2-3A per a voltage tap.

15V at 3A is 45W. Your circuit is not ideal for that kind of power. Aside from the AC trickery, trying to make 12VDC from 18VAC is going to dissipate a lot of power regardless of how you arrange the circuity in between. You really need a 15VAC transformer with dual secondaries or center tap and a 12VAC transformer with dual secondaries or center tap, use the center taps as 0V and make a conventional linear supply with proper heat sinking for regulators.

Although if I were doing this, I would use multiple sets of SMPS with shunt regulators and capacitance multipliers.

 

[ccaudle]

I need to use positive regulators only to achieve that

I don't really understand the meaning of "need" in this context. Why would you "need" to only use positive regulators? You have a full reel you need to use up? That is the first thing which comes to mind since LT1083 is not recommended for new design.
Are you really using LT1083, or using LT1084 and the symbol is just labeled '1083?

2-3A per

With an 18V transformer you get around 24V DC at the smoothing cap after the diode bridge (at nominal input voltage and loading). That means the 12V regulators have to drop 12V while passing 3A, so the regulators will be dissipating 36W as heat.

Assuming you are using LT1084 in a TO-220 package, which has a thermal resistance of 2.7 degC/W, the package would be 97 degrees C above ambient with a "perfect" (i.e. infinitely large) heat sink.
The power section is rated for up to 150 degrees max, so in a cool enough room with a big enough heat sink it might survive, but it is a really bad idea to try.

 

[syn]

Guys,

I said above:

"Screen shot is somewhat blurry, ignore that and part choices and values, that's no problem, it is just the concept that I need to check.

Haetsinking is not a problem, I have a huge metal surface + huge(ish) heatsinks for that purpose, and the most loaded rail +12 will at max reach
around 1.7~1.8 A. Others even less.

I don't really understand the meaning of "need" in this context. Why would you "need" to only use positive regulators?

I'm not aware of 5A negative regulator part number, hence I "need" to use positive regulators for negative rail as well

I will use LM338k for both, positive and negative rails


Today I've played with simulator some more and definitely the circuit above is a no go.
As soon as I add second "negative" regulator and load it, everything falls apart.

 

[ccaudle]

not aware of 5A negative regulator

I have been looking at the common regulator suppliers off and on over the last few days, and I think you are right, I did not find any available devices for 5A negative regulator.
5A, especially at the input-output voltage differential you have, is definitely into the area where a switching regulator would probably be better. I don't know of a part number for a pre-built module that can handle negative supply regulation at that current, but might be worth spending an hour at Digikey, Mouser, etc. seeing what is available.

I have a commercial linear power supply which uses uA723 regulator controllers with external transistors.
uA723 from TI

That would be slightly more complex, but very flexible, you could use the same basic design for multiple voltages with a few component changes.