Rectifier bridge, high-end style

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

peranders

Well-known member
Joined
Oct 14, 2004
Messages
186
Location
Göteborg, Sweden
I have made a rectifier bridge for people that think plain ones won't do. The bridge is made of fast recovery diodes MUR3020. The pcb is a 4-layer board with gold pads and this prototype series became yellow.
 

Attachments

  • rfb03r0_proto.jpg
    rfb03r0_proto.jpg
    83.4 KB · Views: 142
Don't you want an isolation kit for mounting these MUR3020WTs (same as MUR3020PT pin 4) on heatsink because of common cathode?
 
Samuel Groner said:
What's the reasoning behind removing the ground (?) plane below HS3 and HS4? Are R1-R4 series resistors for the snubber caps?

Samuel
I want to be able to mount the heatsinks tight to the board without have to worry about short circuit.
 
peranders said:
I have made a rectifier bridge for people that think plain ones won't do.

I doubt you'll find many here. Not so many will fall for the audiophile belief systems.

We just had discussion on a somewhat similar subject here: http://www.groupdiy.com/index.php?topic=35154.0

So, why wouldn't schottky diodes work better for you as well? And are you sure your snubber caps aren't actually making things worse, ie. moving the switching crud to lower bands, say, audio you can hear. Snubbers are bandaid for slow diodes...
 
peranders said:
The pcb is a 4-layer board

...why?

I know this isn't your first design by a long stretch, and that there must be a technical reason to go with a multilayer PCB. It's just that for the life of me I can't figure out what it is.

JDB.
 
jdbakker said:
I know this isn't your first design by a long stretch, and that there must be a technical reason to go with a multilayer PCB.
http://www.groupdiy.com/index.php?topic=33963.msg415234#msg415234

Wikipedia:
High-end audio can refer simply to the price, to the build quality of the components, or to the subjective or objective quality of sound reproduction.

But I still don't understand...



 
jdbakker said:
peranders said:
The pcb is a 4-layer board

...why?

I know this isn't your first design by a long stretch, and that there must be a technical reason to go with a multilayer PCB. It's just that for the life of me I can't figure out what it is.

JDB.
I wanted to remove copper under two heatsinks because I wanted a simple mounting with any insulation AND I wanted much copper for carrying 30A. I'm aware of that you hardly will get 30 A for longer periods in most amps but at least it's hard to destroy the pcb. Besides, I felt for it.
 
audiox said:
http://www.groupdiy.com/index.php?topic=33963.msg415234#msg415234
Wikipedia:
High-end audio can refer simply to the price, to the build quality of the components, or to the subjective or objective quality of sound reproduction.

But I still don't understand...
... and you never will. Discussing with audiophiles is like discussing religion or politics. It's a discussion you can't win.
 
peranders said:
... and you never will.

Do you have any measurements, simulations, listening tests or anything proving that there is difference between 2-layer and 4-layer rectifier diode board? That would help.

 
peranders said:
jdbakker said:
peranders said:
The pcb is a 4-layer board

...why?

I wanted to remove copper under two heatsinks because I wanted a simple mounting with any insulation AND I wanted much copper for carrying 30A. I'm aware of that you hardly will get 30 A for longer periods in most amps but at least it's hard to destroy the pcb.

That won't be terribly effective. For the same amount of copper inner layers have rather less current-carrying capacity than outer layers, as it's easier to get rid of heat on the outside of a board than on the inside (see the relevant IPC charts for details). Besides, pretty much all the board houses I work with charge less for a 2L board with thicker copper foil than for a 4L board with a standard build.

(I personally prefer hi-current traces to be on outside layers as such traces run the highest risk of melting on major overload, and on outers it's easier to rework/reinforce them with copper wire. But that's me.)

peranders said:
Besides, I felt for it.

Ah. Well, that's really up to you, of course.

It would be decent of you if you would point this out in your sales thread, though. The technically not-so-wary might get the impression that four is better than two.

JDB.
 
Isn't that little masochistic to introduce your hifi/high-end products in a pro audio forum like this? It is 100 percent sure that you face criticism if you don't have arguments based on science. There are huge amount of forums where immediate reaction would be "Gold plated 4-layer board, wow! That must be good!"
 
audiox said:
Isn't that little masochistic to introduce your hifi/high-end products in a pro audio forum like this? It is 100 percent sure that you face criticism if you don't have arguments based on science. There are huge amount of forums where immediate reaction would be "Gold plated 4-layer board, wow! That must be good!"

That's not entirely fair. Per-Anders is quite active on a few other fora, where his posts definitely show clue rather than audiophoolery (although it's debatable how much of an effort it is to be the voice of reason on diyAudio). He has released some interesting designs with full schematics available even if you don't buy boards from him. And let's face it, which of us hasn't at one time or another chosen shiny! over technical reasons, especially in a private project?

I doubt this design needs a 4-layer board, but if that's his choice then so be it.

JDB.
[assuming, of course, that this p-a is the same p-a we've seen elsewhere. On the internet nobody knows you're impersonating a dog and all that]
 
My intention is not to be rude and I don't mean that you sholdn't introduce your designs here. Please do that.

But I think it is not fair to give an impression (especially to members with not so strong technical background) that you get better results just by using some exotic techiniques. (Well, I am not sure if 4-layer boards are so exotic anymore, but they are much more expensive and have some reliability issues if not designed carefully.)

My opinion is still that you should tolerate more criticism than average DIYer, since you are advertising your products here.
 
I can tolerate criticism, not worries there.

My intentions here was to possibly get a few more people on my list which is positive when I send my order.

I'm aware what happens with high currents and multilayer. A while back I did a 300 A (continuously) design with 6 x 140 um which got pretty hot despite the amount of copper. And the project before had 150 A, continuously and 475A peak in the pcb. Before those projects I was rather scare of currents more than 25 A but now I have worked with 55 kA.
 
peranders said:
I'm aware what happens with high currents and multilayer. A while back I did a 300 A (continuously) design

In multilayer boards you can't use the inner layers for currents larger than maybe hundred milliamps (or even less if reliability is important).

That is because the only connection between the inner layer and outer layer (or component) is the joint of inner layer copper and hole plating. And that is very very thin. The currents you are talking about will vaporise it in fraction of a second.
 
audiox said:
peranders said:
I'm aware what happens with high currents and multilayer. A while back I did a 300 A (continuously) design

In multilayer boards you can't use the inner layers for currents larger than maybe hundred milliamps (or even less if reliability is important).

That is because the only connection between the inner layer and outer layer (or component) is the joint of inner layer copper and hole plating. And that is very very thin. The currents you are talking about will vaporise it in fraction of a second.

There are ways around that. If you have money to burn, you can have the vias filled with copper (preferred) or tin (oldskool). A low-tech way is/used to be to simply solder wires (or component pins) in your vias, but it's hard to guarantee that there will be no voids. If you have area to burn and no traces in between, you can stitch the planes together with as many vias as you like (in reality: as many as your board house supports).

Having said that, pretty much all board/system designers I know would go through great lengths to avoid ending up with a 6L board with a total copper thickness approaching 1mm. It simply doesn't make sense, especially when you consider that the heat buildup is cumulative (so when traces at one inner layer are at max current for your max temperature rise, you can't also have the same current running through traces in other layers). Most hi-power stuff (motor drivers, large SMPSes, industrial battery chargers) either have all hi-current traces on outer layers with 4oz (140um) or even 6oz (210um) copper, with low-current signalling on inners, or they simply use separate components/subsystems with wires running between them. (Really high power RF systems are yet another story, and a case where in some settings tubes make more sense than semiconductors). A six layer board with 4oz foil on each layer may be a nice proof-of-capabilities for a board house (laminating such a stackup is hard when copper distribution is even the smallest bit nonuniform), but I cannot imagine a scenario where suggesting such a monster be put in production at an electronics design company wouldn't get you laughed out of the design review meeting or worse (if you or anyone else has a war story where a similar board was actually necessary I'd gladly listen and learn).

Point remains: you're building a shiny! four-layer rectifier board where a two-layer board would have worked at least as well for less money. If this was just for your own use then no one would mind. But you're also offering them for sale to less technically inclined people, and it would be decent of you to explicitly point out that the four layers (and the gold, and the yellow solder mask) serve the same purpose as racing stripes on a car: very pretty, but little to no impact on total performance.

JDB.
[full disclaimer: I've been known to do commercial designs in 4L which could have worked in 2L, but in those cases I knew that we were running a four-layer panel with multiple board designs anyway, and starting up a separate 2L run would have made things more expensive rather than cheaper]
 
Back
Top