Modern 317/337 alternatives?

[maxw]

The LM317 and LM337 commonly used to make a +/- power supply are pretty old now. I see lot of the more modern DACs are using newer low noise regulators like the LT1963 etc.

Does anyone know of a newer set of positive/negative regulators that are superior for audio use?

[jcx]

use dual sec, dual bridges - then you can use the best performing reg for both polarities

[jackinnj]

I won't throw any rocks at the LM317/337 -- but we found that the Linear Tech LT1963A and LT3015 provided the most pleasing sound of any OEM 3-legged regulator when compared with 12 of their peers.

Noise is the least important criteria -- output impedance and PSRR seem to be more important factors in empirical evaluation of regulators. Nevertheless, listening always wins. There were regualtors which tested superbly but really screwed with the sound.

[banana]

tps7a3001, tps7a4901 ;-)

[maxw]

Originally Posted by jackinnj

I won't throw any rocks at the LM317/337 -- but we found that the Linear Tech LT1963A and LT3015 provided the most pleasing sound of any OEM 3-legged regulator when compared with 12 of their peers.

Noise is the least important criteria -- output impedance and PSRR seem to be more important factors in empirical evaluation of regulators. Nevertheless, listening always wins. There were regualtors which tested superbly but really screwed with the sound.

Ah that's what I wanted! I couldn't find the negative counterpart for the LT1963A.

Originally Posted by banana

tps7a3001, tps7a4901 ;-)

Yes, these sound nice:
"One of the primary TPS7A30xx applications is to provide ultralow noise voltage rails to high-performance analog circuitry in order to maximize system accuracy and precision. In conjunction with its positive counterpart, the TPS7A49xx family of positive high-voltage linear regulators, the TPS7A30xx family of negative high voltage linear regulators provides ultralow noise positive and negative voltage rails to high-performance analog circuitry, such as operational amplifiers, ADCs, DACs, and audio amplifiers."

They only come in MSOP-8 package though, at 3x3mm, it would be tricky to solder.

Thanks for the suggestions!

[jcx]

as I said - you don't need "the negative counterpart" you can use the same type for both polarity supplies by floating both secondary windings with respect to each other

which allows you to use the best reg you can find for both supplies - and use the same part # in both

[banana]

Originally Posted by maxw

They only come in MSOP-8 package though, at 3x3mm, it would be tricky to solder.

It's not only MSOP-8, but also the exposed PowerPAD that makes it even trickier to solder.

[canvas]

Originally Posted by jackinnj

Noise is the least important criteria

On my CD transporter, I use an OP-based pre-reg and compare the LT1085-CT5 with LT1763-5. The result from LT1763-5 is quite impressive, much quieter than LT1085. I guess noise is important for digital part.

[maxw]

Originally Posted by jcx

as I said - you don't need "the negative counterpart" you can use the same type for both polarity supplies by floating both secondary windings with respect to each other

which allows you to use the best reg you can find for both supplies - and use the same part # in both

I've seen this used in a Gainclone before but why isn't this approach taken very often? Transformers with dual secondaries are pretty standard these days but you never see this used in datasheets or elsewhere?

[agdr]

Originally Posted by maxw

Ah that's what I wanted! I couldn't find the negative counterpart for the LT1963A.

There is one small hiccup to watch out for if you wind up using the LT1963A and LT3015. The maximum input voltage on the LT3015 is -30Vdc while the LT1963A is only +20Vdc.

I'm doing a design and layout right now with both parts being fed by a 16Vac half wave and I want +/-16Vdc out. That results in the incoming ripple going from 20.5Vdc to 21.5Vdc, too much for the LT1963A but OK for the LT3015 on the negative rail (-20.5Vdc to -21.5Vdc). So I borrowed the great idea that pete_shumacher recently posted here for his power filter of a diode (LED) drop from the rail feeding a series pass to drop 3V of incoming voltage for the LT1963A. So now it gets ripple that goes from 17.0Vdc to 18.0Vdc. Problem solved.