[splaz]

Okay well here's the pics.

http://img301.imageshack.us/img301/3...hematicjr5.png
http://img490.imageshack.us/img490/5...erboardsa5.png



Just wanted to know in general if there's anything wrong with either.

Just did a pretty much copy of the schematic to board, it's simple to do and makes it far easier to etch. Plus I don't feel all that much board space is wasted.

edit: Updated schematic based on suggestions.

http://img71.imageshack.us/img71/915...hematicmh3.png

Starting the board layout again as I stuffed it up a bit while re-arranging.

Does anyone know of a good way to do the four rectifier diodes ?

I'ts probably next to impossible but I'm trying to have the ground plane surrounding all traces were possible, yet also no jumpers and input pads should be neat. With the snubber caps it's really hard to arrange it to do that. Oh and single sided of course.

 

[Epicurean]

Umm, it looks like you have half the rectifier bridge after the regulators?? And just one AC-in. I'm confused

 

[tomb]

Nah, he's just calling the second AC-input "Gnd", which strictly speaking, could be confused. "Gnd" could be the third hole on a wall outlet, which is not the same thing. You are correct about there being only 1/2 of a rectifier bridge, though.

Splaz,

The circuit may cause big problems from the capacitance after the regulator: it's too much. Large capacitance after a regulator will cause ringing and instability. This issue is covered in many articles and in the regulator datasheets. You need -one- small and fast cap after the reg. Look at Tangent's STEPS and TREAD design more carefully - there is only a single, tiny cap after the regulator circuit itself.

 

[DaKi][er]

I'd like to see R1-4 closer to the regulators on the pcb layout

I've never had any problems with that amount of capacitance after the regulator and don't believe you will either (at least for lm3x7's)

Also check that the -ve regulator is actually a lm337 as it says lm317 on your schematic and they do have different pinouts (edit: though from memory it looks right anyway)

 

[Paragon]

The gilmore PS has 4700uF before and after the LMs.

I am confused here.. Why not build a full wave rectifier? I don't see how the power from a half wave would be good DC.. more like a square wave.

 

[AndrewFischer]

Along with all the other suggestions.

Add heatsinks. It will help with layout. You can always omit them when you build or even leave them off the silk screen.

Move C11 and C12 much closer to the regulators.

You've got 2 air wires on the layout. I'm not sure why.

Replace C13 and C14 with tantalum caps around 10µF to 25µF. Consider removing C15 and C16.

Move the resistors closer to the regulators. Read the NS datasheet for more information.

With good cap choice C7 and C8 should be redundant.

Add solder pads or connectors for input and output. Add a few extras for test points.

Consider adding trimmers to R1 and R2.

Run a 10mil design rule check.

 

[tomb]

Here's some interesting links about increasing ESR too much after the regulator:

http://www.calex.com/pdf/3power_impedance.pdf
also
http://www.national.com/an/AN/AN-1148.pdf
This is the application reference from National given for many regs, including the LM317. I think you will find that cap sizes much larger than 47uF in electrolytics and 4.7 in Tantalum will not be optimum for LDO's. The LM317 is actually a NPN regulator, so it's not as sensitive to this scenario, I believe. Whether you can get away it without introducing instability is a question. Regardless, increasing ESR after the regulator is not necessarily in the right direction. For other LDO regs such as LM7800's/7900's, it may not be a good idea at all.

 

[splaz]

Thanks for all that input.

Well I'm changing it around now, adding in spots for 2 diodes so it can be full wave doubler as intended or full wave bridge with a centre tap. Removed the 1uf capacitor on the input. Heatsinks, none in the eagle library like what I'd use and can get. Few test points are in now.

I'll reduce the size of the output cap to maybe space for a 100uF electrolytic max, tantalum should easily fit there too.Is the 100nF film cap needed or not ?

I included it as the datasheet said it is for improved transient response but would the electrolytic/tantalum perform that function on it's own, the lower ESR leaving it in is probably bad though right ?

Updated the first post.

 

[AndrewFischer]

I would leave in the space for the film caps. You can experiment with them in or out and use what sounds best. If you have access to an oscilloscope, you can check for noise and oscillation.

Aavid heatsink libraries are in the download area at cadsoft.de. Google turns up all sorts of freebie Eagle libraries too. Even if you don't use it you could drop in something for spacing and then take it back out before you make your Gerber files. You could also make your own or modify something from the Aavid library. If nothing else you can cut and paste the schematic symbols and then draw your own package.

 

[mono]

I don't see any place for mounting this board.

The traces seem unnecessarily thin.

 

[splaz]

^ True, will have a look then and see if the heat sinks are like those I get locally.

Mounting holes are sort of the last thing on my mind, traces are thin because I don't use these crazy imperial measurements. If it was mm I'd actually have an idea...

Anyway to change that btw ?

 

[SiBurning]

What use will you put this supply to? It looks to me like a variable +3-15vdc, -3-15vdc supply. What about current requirements? Voltage tracking?

Not that I could necessarily give you any good advice, but it's very hard to suggest anything without knowing the use it'll see.

For example, if you need excellent tracking of the +/- voltages, you need a redesign. If you need high current, there are also better ways. Is it a general purpose bench-type of supply or one dedicated to a particular audio circuit. And so on...

 

[threepointone]

aren't the VO and VI on the negative regulator reversed? Also, check teh protection diode polarity if it is backwards.

I know someone said that the negative regulator should be a 317 on one of your previous posts, but i think someone should probably confirm that--i would think you need a 337 negative regulator for the -15V side.

 

[splaz]

^

True,I think I know why though. Must've copied the 317 and renamed it, but of course, different pinouts. Need to change that I guess or double check what it is. I know it's a 337, don't know if what's in the schematic/layout is a 337 though.

This power supply is for a headphone amp, needs to supply maybe 200mA just to round it up a bit. Not variable in this config with the fixed resistors but swapping them for trimpot/s could do that, haven't decided if I want to be able to tweak the output voltage slightly.

 

[mono]

If it's essentially a fixed supply for headamp, you won't gain anything in particular from being able to hit some exact n.nnV, BUT you may want a pot on one of the two rails. Having it in series with a lower-ohm resistor than the corresponding resistor on the other rail adjustment loop, you can tweak that rail with the pot/trimmer to match the other (fixed) one, since random variations in resistors could result in the two not being exact opposites. Or get some high tolerance resistors and swap if needed.