Diodes in regulated PSU...
Feb 25, 2007 at 12:51 PM Thread Starter Post #1 of 12

Mayday

New Head-Fier
Joined
Jan 14, 2007
Posts
43
Likes
10
I'm going to build a 15V stabilised PSU for my B-Tech BT928.
I've got the trafo, caps and regulator sorted, but I can't decide on what type of diodes to use as rectifiers...I've got:
BYV-27 soft recovery diode
1N4007 rectifier diode
1N4148 fast recovery diode
1N5819 Schottky diode

What are your thoughts on this?

/mayday
 
Feb 25, 2007 at 4:31 PM Post #2 of 12

Nerull

100+ Head-Fier
Joined
Jun 14, 2005
Posts
344
Likes
10
Use rectifiers, and snubber caps to stop those nasty high frequencies from going anywhere. The reason I say rectifiers is because they were made for rectifying (duh, hehe) the very power you get from the wall, and those other diodes weren't really made with that in mind (although they will do the job). I also suggest that you use a bridge, because all the diodes are built on one substrate and probably have better matching than those picked out individually. These things are probably better than any "ultrafast" diode in a PSU.

~Tom
 
Feb 25, 2007 at 7:05 PM Post #4 of 12

Pars

Can Jam '10 Organizer
Joined
Oct 28, 2003
Posts
4,123
Likes
51
I'd use Schottky's, something like 11DQ10 or 31DQ10 depending upon voltage/current requirements. The 11DQ10s work quite nicely in stuff like CDPs, etc. and the 31DQ10s for preamps, etc. (I used these in my Dynalo). Digikey carries these. With Schottky's, you shouldn't need to bother with snubbers as they don't ring like soft/fast recovery types do. If you want a bridge (Hexfred), IXYS makes some (Michael Percy carries them).
 
Feb 25, 2007 at 7:10 PM Post #5 of 12

Mayday

New Head-Fier
Joined
Jan 14, 2007
Posts
43
Likes
10
Quote:

Originally Posted by Pars /img/forum/go_quote.gif
I'd use Schottky's, something like 11DQ10 or 31DQ10 depending upon voltage/current requirements. The 11DQ10s work quite nicely in stuff like CDPs, etc. and the 31DQ10s for preamps, etc. (I used these in my Dynalo). Digikey carries these. With Schottky's, you shouldn't need to bother with snubbers as they don't ring like soft/fast recovery types do. If you want a bridge (Hexfred), IXYS makes some (Michael Percy carries them).



What's your thoughts on the schottky's I have allready?

Will be used After a 2 x 15V~ 20VA trafo. Electrolytes will be 4400uF or 8800uF depending on if the trafo can hande the higher capacitans. Regulator will be a LM317T.
 
Feb 25, 2007 at 7:58 PM Post #6 of 12

Pars

Can Jam '10 Organizer
Joined
Oct 28, 2003
Posts
4,123
Likes
51
Quote:

Originally Posted by Mayday /img/forum/go_quote.gif
What's your thoughts on the schottky's I have allready?

Will be used After a 2 x 15V~ 20VA trafo. Electrolytes will be 4400uF or 8800uF depending on if the trafo can hande the higher capacitans. Regulator will be a LM317T.



You mean the 1N5819? Its rating is 40Vrr/1A. It should work OK, but I personally would want something with more margin in the reverse voltage spec. I would defer someone with more PSU design experience than me however. What voltages are you going for... +/- 15V? Your 15V secondaries are perhaps as high as 18V unloaded (maybe more), which would give 18V * 1.414 = 25.5V across the rectifiers.

Also, the 1N4148 in your list is a small signal diode (generic) and not suited for PSU use.
 
Mar 16, 2007 at 12:50 AM Post #8 of 12

mono

1000+ Head-Fier
Joined
Jan 10, 2005
Posts
1,412
Likes
11
How much current does a BT928 use? How much rail capacitance-induced inrush current (or will the PSU have current limiting)?

1N4148 should be eliminated right away as it's a low current generic switching diode.

BYV-27 will give you the least diode noise, but I'd stil use a snubber.


1N4007, well it's cheap. High 1000 voltage but I doubt you need that.


1N5819 is fast, and will probably have least dissipation from lower forward drop, but it's not so outstanding either.

Unless your current is high enough that none of these are suitable, BYV-27 wins for being soft recovery. Will you actually hear a difference if you had a snubber on any/all of them? Would you hear a difference even if you didn't? Hmmm. You tell us.

OK, I just saw a page with BT928, just a chip/op/amp circuit so there shouldn't be a current problem.
 
Mar 16, 2007 at 1:50 AM Post #10 of 12

mono

1000+ Head-Fier
Joined
Jan 10, 2005
Posts
1,412
Likes
11
You could just use some 10nF caps, around that range is a good start as many people don't use the resistors.
 
Mar 16, 2007 at 2:57 PM Post #11 of 12

Garbz

Headphoneus Supremus
Joined
May 19, 2004
Posts
3,573
Likes
12
Quote:

Originally Posted by Mayday /img/forum/go_quote.gif
I've got a few UF4007 coming, these are a ultra fast version of 1N4007 I was told. Are they pherhaps the best choice?


Only if you listen to Ultra Fast Music.
rolleyes.gif
Compare the price of 4 UF4007 to a rectifier bridge which as Nerull already said comes closely matched, and you will see that any percieved sound difference, even the one in your mind is hardly worth the price.
 
Mar 16, 2007 at 4:18 PM Post #12 of 12

Pars

Can Jam '10 Organizer
Joined
Oct 28, 2003
Posts
4,123
Likes
51
Having worked in the semiconductor industry 20 years ago, I would seriously doubt that bridges, even if they are built on the same substrate, are matched very well if at all. It is a commodity part, and at the selling price, they are not going to spend the money on this. The Toshiba 2SJ109 and 2SK389 FETs are touted as being built on the same substrate and matched... some of these that I have run across had an rather interesting concept of matching
biggrin.gif


UF4007s are ~$0.20 apiece. The soft recovery characteristics are typically audible in my experience (maybe in my mind, but whatever). Schottkys are even better in terms of their noise performance at turn-off (and don't need snubbers), and are well worth using if they meet the electrical parameters required. Yes, the Cree diodes touted at Partsconnexion, etc. are pretty (maybe stupid) expensive, but I won't go there.
 

Users who are viewing this thread

Top