# +/-15 linear PSU - 2x15V or 2x18V transformer

A

#### Andre Majorel

Jan 1, 1970
0
I think that you could use a 2 x 15V transformer for a +/-15V
linear PSU. The LM317 and LM337 need Vin to be 3V above Vout.
Assuming the smoothing capacitors are big enough, a 15V
secondary would provide 15 x sqrt(2) - 0.7 - 15 = 5.5V.

Yet some people recommend a 2 x 18V transformer in this
application. Why ? It makes the regulators work harder. Is it
to avoid the bigger inrush current associated with larger
capacitors ?

J

#### john jardine

Jan 1, 1970
0
Andre Majorel said:
I think that you could use a 2 x 15V transformer for a +/-15V
linear PSU. The LM317 and LM337 need Vin to be 3V above Vout.
Assuming the smoothing capacitors are big enough, a 15V
secondary would provide 15 x sqrt(2) - 0.7 - 15 = 5.5V.

Yet some people recommend a 2 x 18V transformer in this
application. Why ? It makes the regulators work harder. Is it
to avoid the bigger inrush current associated with larger
capacitors ?

--
André Majorel <URL:http://www.teaser.fr/~amajorel/>
(Counterfeit: [email protected] [email protected])
"La presse doit diffuser des idées saines." -- Serge Dassault,
propriétaire de la Socpresse.

(usually factor for 2 diode drops at about 0.8V-1V)

No straight answer, as it hinges on the maximum current drawn, the
transformer quality and how much your're prepared to cough up for smoothing
caps.

Cheap transformer, cheap caps, 1A load then go for 18Vac.
Cheap transformer, cheap caps, 100ma load then go for 15Vac.
Good transformer, good caps, 1A load then 15Vac.

From a sim' I already have ("good" transformer and caps) ...
18Vac, 2200u, 1Amp. Gives 21.5V to 25.1V ripple pp. (pass)
18Vac, 10000u, 1Amp. Gives 24V to 25V ripple pp. (pass)

15Vac, 2200u, 1Amp. Gives 17.2 to 20.5V ripple pp. (fail)
15Vac, 10000u, 1Amp. Gives 19.5 to 20,5 ripple pp. (pass)

Also bear in mind the capacitor ripple currents run at about 2.8Amps rms for
a 1Amp load. The 10000uF's can usually handle this in a 35V rating but (say)
2200uF's may only be able to handle this up at about 100V rating. Cost of
each capacitor is markedly different.
Minimising the smoothing capacitor costs is a classic juggling act.

regards
john

J

#### John Popelish

Jan 1, 1970
0
Andre said:
I think that you could use a 2 x 15V transformer for a +/-15V
linear PSU. The LM317 and LM337 need Vin to be 3V above Vout.
Assuming the smoothing capacitors are big enough, a 15V
secondary would provide 15 x sqrt(2) - 0.7 - 15 = 5.5V.

Yet some people recommend a 2 x 18V transformer in this
application. Why ? It makes the regulators work harder. Is it
to avoid the bigger inrush current associated with larger
capacitors ?

I was one who recommended considering the 15 volt winding to provide a
(low current) 15 volt regulated output. This works pretty well if you
have a bit of extra transformer (for low current supplies, you often
cannot get a small enough transformer to be fully loaded), because a
lightly loaded transformer puts out a bit of extra voltage.

But remember to take the low line voltage case into account when you

D

#### David L. Jones

Jan 1, 1970
0
john said:
(usually factor for 2 diode drops at about 0.8V-1V)

No straight answer, as it hinges on the maximum current drawn, the
transformer quality and how much your're prepared to cough up for smoothing
caps.

Cheap transformer, cheap caps, 1A load then go for 18Vac.
Cheap transformer, cheap caps, 100ma load then go for 15Vac.
Good transformer, good caps, 1A load then 15Vac.

From a sim' I already have ("good" transformer and caps) ...
18Vac, 2200u, 1Amp. Gives 21.5V to 25.1V ripple pp. (pass)
18Vac, 10000u, 1Amp. Gives 24V to 25V ripple pp. (pass)

15Vac, 2200u, 1Amp. Gives 17.2 to 20.5V ripple pp. (fail)
15Vac, 10000u, 1Amp. Gives 19.5 to 20,5 ripple pp. (pass)

Also bear in mind the capacitor ripple currents run at about 2.8Amps rms for
a 1Amp load. The 10000uF's can usually handle this in a 35V rating but (say)
2200uF's may only be able to handle this up at about 100V rating. Cost of
each capacitor is markedly different.
Minimising the smoothing capacitor costs is a classic juggling act.

regards
john

On top of that the mains voltage will vary too. It might be lower (or
higher) than the "nominal" 110/220/240V

You have to design for the worst case scenario (and then some), so with
Linear regs you will always be throwing away a fair amount of power if
you properly design for those worst cases. If this keeps you awake at
night, then you should be thinking about a DC-DC converter.

Dave

Replies
10
Views
2K
C
Replies
4
Views
2K
Greetz
G
Replies
13
Views
730
Replies
2
Views
751
Replies
2
Views
510