my dulal layer psu board.. needs crit!

Hey guys

wondering if the experts could look over this and go nuts :)

ive had to go double sided to fit within my 100x60 size limit. so ive had to make some compromises.

some notes:

no labels as yet sorry, but it should be pretyy obvious , bar the following:

- the 7805 regulator is for a 50mA aux supply required by the class D IC im using it with. is no way related to the +- rails.

the third ground pin on the baby aux supply' s 3pin header is intended to be a ground return for the amps input signal, going to the 'psuedo' star earth at the bottom right of the board,

my main concern , tho i dont know if its justified is the dual ended ground plane, with the left end going to the transformers centretap. does anyone see a problem here?

also the -ve line crossing over the AC line on oposite side of board.



any help form the experts appreciated

Cheers!

Jimmy

psu2.jpg
 
Hi Pinkmouse,
Try to reduce loop length grnd/output cap on reg. Rotate c10 up beside c9 or put it on the back on the output trace side of the reg.
Your pseudo star earth is very wise giving you 4 earth connections. Can you provide some more? This then allows you to make this your real star earth.
I think there may be a slight advantage in moving the clean earth tie in points slightly farther away from caps c4 & c8, maybe even around the corner towards the output terminal. This is just to ensure that charging pulses in the caps across the main return and back to the transformer/rectifier without coming near the star earth and generating a small volts difference around the reference area.
Keep the transformer connections away from the star earth. Your layout meets my ideal criteria perfectly.
re the AC crossing the DC at the rectifier; there is a board thickness between them. There should never be a risk of shorting unless the board were very badly charred between these two points and for this senario to occur then your problems are much worse.
How about adding some small hi freq caps bypassing the main smoothing?
And maybe some fusing either on the AC or the DC before the caps or both?
 
thanks for the reply guys,

pinkmouse, sorry bout the lack of extents, im pretty new to eagle and have kind of rushed into it without labeling things properly :p , ill fix it up tho.

With the ground plane , i was refering to the fact I have the "star earth" on the right of the caps grund plane , yet the centre tap connection for the transformer is joining on the left.

AndrewT: , is that ^ what youre refering to by making it a "true star earth" ? , do you think i should bring transformers ground down under the caps and join the star earth on the right instead? (ill move the caps to the left a bit )


Re the cap on Regulator. I see what you mean now, ill swing it round like C9 , and add bypass caps to the rails .

Fuses i almost forgot about , to save space i think ill have to add some minature pico fuse style in before the rectifier, moving it along a bit.



cheers
Jimmy
 
Having the centre tap connection at the other end is no problem; as has been mentioned above, there are large charging current spikes flowing through there which it is good to keep separate.

Overlapping -ve and ground is a good thing as it adds a little extra capacitance between them. It is a good idea to deliberately run them on top of each other as much as possible (and +ve too of course). The extra capacitance keeps a low impedance up to very high frequencies too, which is a bonus.

Do you really need four pairs of capacitors in parallel? There was a thread a while back about paralleling reservoir caps, and it became clear that going past two in parallel gave diminishing returns. Thus using two (of twice the capacitance for the same total) in parallel, possibly also with smaller bypass capacitors, might be better, as it would allow for a tighter PCB layout.
 

costiss

Member
2002-12-26 1:27 pm
greece
1: use the top layer as a full ground plane, along with a pad for the capacitors only at the bottom layer. use thicker tracks.. if you dont really need those parallel caps, dont use them.. you are better off with some compact high quality audio oriented capacitors (panasonic tsup, aerovox, to name a few).. this way you can have a nice layout.. i also think you dont need two layers.. why dont use sth like this? its only 60x70mm.. what capacity are you planning on using?
 

Attachments

  • ccc.jpg
    ccc.jpg
    25.4 KB · Views: 273
sigh :dead: , totally unhappy with my common earth point now, i want to move it right up to the output header but its getting late and ive had enough for now :p , moving it up there is going to take some major rearrangment i think.


things ive changed:

added Fuses

Added hi freq bypass caps 11 and 12

moved common earth point

flipped bypass cap on reg around as suggested, close to output of reg.

I think thats it.


MR evil, i only used the 4 caps (theyre 2200uF Low ESR psu caps) due to reading that whilst using more in parralel has an ever decreasing advantage over one, its allways better . Ill have to read the thread you talk of tho if it goes against this theory. thanks for the tipoff :) , using only 4 would probably allow me to route the ground trace to the Left side of the output connector, and make my common earth point there. hmmm.

andrewT whats your thoughts on using higher uF caps if u dont mind?


cheers!

Jimmy




psu4.jpg
 

jcx

Member
2003-02-17 7:38 pm
..
why not a real ground plane - "fill" the entire ground side with copper (observing safety clearance requirements of course)

I would also route V + and - as close togeter as possible to reduce loop area, both well overlapped by bottom ground plane

routing seperate reg gnd ref from output gnd terminal is overkill but costs little

4 term conn on output could be good too, twist each pwr with a gnd then twist both pairs together - reduces gnd wire resistance and cuts loop area

fuses on secondary seems unusual, pri side fuse protect xmfr, fuses on sec beteen bridge and resevior caps are equivalent to pri fuse in function, fuses after resevior caps can give better protection to equipment as they don't have to be oversized to take turn on current spike
 
Hi Jimmy,
I am a brute force advocate for smoothing caps.
My current amps use 21mF + 21mF (3//6.8mF) for each 100W8R channel with extensive plastic bypass at rectifiers, at cap bank and again on board.
I have used // combinations since my earliest experiment.
I have an AARON 100W8R stereo which uses 32// 2.2mF for a cap bank but feeding both channels. Sounds nice but does not meet the manufacturer's exaggerated spec.
My future plans will use more cap power probably 3//15mF for 250w4r
Glad to hear you're not changing the transformer return connection and that no one has tried to explain why changing it would (never) be better.
 
OK well ive taken on some of the things suggested.

moved the output conneciton completely, making star earth point for input signal and aux power virtually the same point as power ground connection

thickend the tracks .. i couldn't really make the whole bottom layer a ground plane due to the other stuff I have running through it, i have to keep the board the same size, and want to keep the same caps as they are.

thanks for all the suggestions, hopefully this si pretty close to how ill export it ;p






psu5.jpg




cheers
Jimmy
 
Hi,
you have lots of empty board space. Leave space for a heatsink on the rectifier, or is it under the board to bolt onto the chassis? If so then you need a hole to access the screw.
Can you squeeze/optimise your layout to make the board smaller?
I still recommend a big increase in the number of ground connections off the star ground for all future requirements. You don't have to stuff all the terminals positions now.
 
Hi,
are you two channel?
Then be very carefull to keep all the 2 channel grounds going to the correct star for each channel.

The PSU common will need to go to both star grounds AND you must ensure that there is not a ground loop around the double PSU ground.

It took me many years to (nearly) solve my stereo amp grounding before I found out how to interpret the requirements correctly.
An internal loop inside the amp case was my initial problem and it was wired exactly as the manufacturer's schematic/hard wired layout.