[firefox360]

Not sure if it's been mentioned or not (if it has been, then sorry for the repeat), but why use the LM317 when you can use the LM7805? It's a tad bit simpler to implement in my opinion, because it's set to output 5vdc at 1A without the need for adjusting resistors.

If you do have specified a reason for using the LM317 then I'm sorry if I've missed it.

EDIT: also on another note, you might want to have a small plane underneath the regulator that doesn't have a solder mask to sink away the heat dissipated from the heatsink pad, by the down-volting process.

 

[balou]

firefox360: yes, a 7805 might actually be an idea. the advantage of the lm317 would be the fine tuning of the exact output voltage with the trimpot. I do not exactly know how much the 7805 output will vary - I'll check it, maybe it really is a good alternative.

about the heatsink: yes, I will do that, but I intended to use an external heatsink, I do not know if the board alone will be able to dissipate so much heat (12v to 5v, 0.5A are a maximum of 3.5W heat). and as of now, the ground plane is on the backside of the board - I will either make a separate heatsink plane, or a second ground plane on the top (with ground plane on the top, I will also have to make sure that it can be double-used as a plate capacitor)


here's the current revision:

I totally forgot the current source from input V+ to GND. a resistor would just introduce the noisy usb power to the groundplane, so I added a simple CCS consisting of one JFET and a resistor. any votes for a full JFET cascode?



edit: firefox360: regulation accuracy is indeed a problem. the 7805 has variance of 4mV, the TREAD-design 0.060mV.
but the lm340at might be a candidate.. I have to check the datasheet

edit2: the lm340 is just a newer version of the 7805. The LM317 also has less-than-stellar performance on the datasheet... so the 7805 might indeed work. Could anybody with more experience building power supplies comment on the question whether the 7805 will work satisfactorily?

 

[firefox360]

Here is a site that list certain specifications for the USB. Might be an interesting and useful resource: http://pinouts.ru/Slots/USB_pinout.shtml

EDIT: According to the site above, I believe the host detect the presence of a device via the Data signals. I think the computer just puts out 5v constantly, ofcourse at a different level of current.

 

[firefox360]

I've attempted to make a circuit myself, and it is definitely far more simple than the ones you've posted. Whether the board I've posted works or not is something else though...

:

I've designed it so that it would fit in the small Hammond 1455C802/1455C801 enclosure, which should match some of the Alien DAC's out there perfectly.

External powersupply is via two wirepads (PS+ & PS- respectively), which would allow for different types of connectors, and even the option to hardwire if so desired. The two USB ports are nearby, each a different connector. "USB-IN" would be the type you would see on the Alien DAC, printers, and etc. "USB-OUT" are the standard connectors, such as those on your computer.

"C1" would be the input filterning capacitor, while "C2" would be the output capacitor, which should minimize some fluctuations. Whether C2 is needed or not, I'm not quite sure. "D1" is the diode which should prevent accidental reverse voltage. I took it from the PIMETA circuit, which means it should short power supply if installed in reverse. The regulator would be the LM7805. I've attempted to leave enough space for the regulator to be bent flat down onto the board, along with a simple heatsink.

Hopefully that describes it enough. I could also post the schematics if anyone is interested.

 

[fordgtlover]

I am building one of these on perfboard (just about to go and solder it up).

In both of the layouts here electrolytic caps have been used. What is the benefit of the electrolytics over the tatalums as suggested in the LM317 datasheet and as offered as an alternative by Tangent in his TREAD?

 

[firefox360]

well electrolytics are avaliable in larger capacities and at a lower price. Also if something happens to go wrong, electrolytics are a lot more forgiving than tantalums.