[Garbz]

Ok I designed it a while ago. I got around to etching the boards today, and I took a look at the ground trace. Something about the design does not smell good, just a hunch really.

The basic idea originally was to make a TL431 based shunt supply. Then the idea was made to split it so I get +/- from 2 separate xformers, and make a ground at the regulator output kind of like the pre-regulator on the gilmore Dynahi PSU.

Anyway can someone here confirm my hunch that this may not work? Specifically I do not like how the LM317 (configured as CSS) is connected to GND on the -ve side. I get the feeling it will let the magic smoke out.

http://www.garbz.com/sch.pdf

 

[DaKi][er]

I cant see anything wrong with it

 

[Garbz]

Just goes to show go with your first instincts and do not overthink things

 

[JahJahBinks]

You need a negative linear regulator like the LM337 like shown in the Gimore PSU:

 

[Rescue Toaster]

Quote:

Originally Posted by Garbz Anyway can someone here confirm my hunch that this may not work? Specifically I do not like how the LM317 (configured as CSS) is connected to GND on the -ve side. I get the feeling it will let the magic smoke out. http://www.garbz.com/sch.pdf

First, huzzah for CCS-fed shunt.

Second, assuming the power supply transformer secondaries float (not centertap) and have low interwinding capacitance, you should be fine. It certainly won't hurt to switch to the 'correct' LM337, though.

But if you do switch, resist the urge to make a ground plane or anything silly. Ground is only the trace between the two shunts, output caps, and output connector.

 

[Garbz]

Quote:

Originally Posted by JahJahBinks You need a negative linear regulator like the LM337 like shown in the Gimore PSU:

No I don't.

Check the Dynahi PSU. That is the conseptual idea I was talking about. You will notice it uses 2 positive regulators.

Quote:

Second, assuming the power supply transformer secondaries float (not centertap) and have low interwinding capacitance, you should be fine. It certainly won't hurt to switch to the 'correct' LM337, though.

What the. Did you look at schematics (no offence intended ofcourse) but the LM317 is configured as a CCS. The amp would behave identically if I used the LM337 and I would have still asked this question. Also I am using 2 separate transformers, not 1 with 2 windings or a centre tap.

Anyway the design looks alright. Worst case I fry a TL431 and I'm a few hours and $3 down.

Quote:

But if you do switch, resist the urge to make a ground plane or anything silly. Ground is only the trace between the two shunts, output caps, and output connector.

Lol no kidding. Even if ground were insanely complicated I would not use a groundplane. Most of my projects where you can see a ground plane it is just a fill plane not connected so I do not use as much ammonium when etching