PSU quality - topologies, significant factors?

[volki]

hi all,

i'm racking some console modules and need to build a multi-supply psu. i would like to take this opportunity to collect some opinions. there's been recent discussion about stuff like V.regs and other things... maybe we could get some condensed, straight forward answers here once more? :razz:
like, are LM317/337 really a significant improvement over L78xx/79xx regarding noise actually imposed on the audio path? or only when it comes to certain loads/voltages? how do emitter followers compare to both, and TL783c?
as for racking console modules, they each have a dedicated (more or less good) filtering stage for the supply anyway, most mic's as well. so is there a point in caring about a little more or less noise in the supply line at all?
also: is a voltage doubler inherently inferior to using a xformer which puts out the desired voltage in the first place? meaning: are 2-way rectifiers much inferior to bridge rectifiers, provided you use a large charging cap and decent regulator ?


anyway, my psu should provide +/-20v (close to 1A), +24v, +44v (a few 100 mA), +48v (phantom (=not much)).
i have a 28v - 0v - 28v xformer, the +/-20v regs are of course supposed to sit at either secondary tap, and the input of the +24v reg parallelled with that of the +20v reg.

here are my questions:

1. for the +/-20v, i figured to use an LM317 / lm337 combo, to be able to precisely adjust the symmetry of the voltages. many circuits (e.g. jakob's ssl-comp, etc.) use L78xx/79xx for that - don't these V.regs have a unit-to unit variation which might be critical, or am i being too picky? e.g. i have an L7824 in another circuit which only puts out 23,5v, with load attached...

2. for providing ac for the +44v / +48v regs, should i best use
a) a voltage doubler fed from the "positive" 28v ac tap
b) another 24v xformer in series (so that i get 52 - 28 - 0 - 28v ac) with another bridge rectifier & charging cap or
c) use the xformer with an entirely separated, second 20/24v supply and relate the output potentials of both psu's accordingly?

for a)&b), should i use
i. L7820/24 or LM317 and simply relate their bases to the 24v potential of the first psu (so that i get 24+20 = 44 and 24+24 = 48v)
ii. L7820/24 or LM317 with 24v zeners at the bases, and relate those to 0v of the first psu?
iii. LM317HV and directly relate those to 0v?
for c), i would simply relate 0v of the second psu to 24v of the first?

3. what is the max. input voltage of LM317 and -HV? couldn't find it in the datasheets. the max. out voltages are 37 / 57v respectively, the max. in-out differential 40 / 60v respsctively. so does that mean you can actually feed them with max. 77 / 117v ??


thanks a bunch for your suggestions, and enjoy the midsummernight :wink:
volker

 

[sismofyt]

Hello volki,

I don't have the time right now, but I'll be happy to discuss this in a fe days :grin:

 

[Gus]

One thing to think about is that almost all 3 trem regs are series regs. They can source current but not sink current. Zeners are shunt regs.

A simple "fix" is to use a resistor load to allow current to be shunted to ground.

Most 3 terms don't get to the best regulation and noise without about 150ma drain.

 

[Learner]

Hi volki,
You can try this design tool on National's web site

http://www.national.com/appinfo/power/webench/

please keep us informed on how you went, as I'll be doing the same thing shortly. As soon as I finish my mid year exam ......

Have fun!! :thumb:


Oh, there is more info on PS here

http://www.web-ee.com/primers/Tutorials.htm

 

[volki]

sismofyt,
you ready now? :green:

[quote author="Gus"]One thing to think about is that almost all 3 trem regs are series regs. They can source current but not sink current. Zeners are shunt regs.[/quote] excuse my dumbness... (or maybe it's the language barrier after all? ) what exactly do you mean by sourceing vs sinking current?
of course, when using regs with a zener at the base, you have to use a bypass R in order to feed current to the zener. does that relate to your statement?

A simple "fix" is to use a resistor load to allow current to be shunted to ground.

that sounds like something different, like a minimum load?

Most 3 terms don't get to the best regulation and noise without about 150ma drain.

interesting. that's prolly why there are dedicated low-curent v.regs? anyway, my app. needs much more current than that :green:

learner,
thanks for the link, but somehow this web bench thing gave me options of switching regulators only. not really suitable, i guess... :sad:

 

[sismofyt]

Still thinking Volki, be patient please :?

Yes, it's a good thing to load Vregs..