5V Step Down SMPSU Using XL2596S
Table of Contents
For running RPi-type boards and other stuff that required 5V at decent current, I had been using these very nice step-down modules:
Synchronous rectification with discrete MOSFETS, specified as 5A. I have been running a couple of these continiously for a few years now with no worries at all. They seem to be very good devices. Recently, looking to buy some more I couldn't find them for sale anywhere. I wonder if they're still being made.
Anyway, finding that I had considerable stock of XL2576-5 ICs on hand. Why not?
So I did, using junk-box electros and inductors.
Circuit
The circuit was taken directly from the datasheet.
The only addition being the output enable circuitry. R2 pulls XL2576S pin 5 high and so defaults to having the output
turned off. Putting a jumper on J3 enables to output permanently, alternatively 3V or more on pin 1 of J2 allows for
a microprocessor or whatever to turn output on and off.
PCB Artwork
PCB
As usual, the PCB came out looking absolutely beautiful in the 3D-render world:
But perhaps not so much in the home-made PCB world:
It works at least:
Although the second board was not too bad:
Homemade Boards.
To help with heat dissipation, I figured that a double-sided board would be best and added a few vias. Homemade vias are simply little bits of wire poked through the board and solder on both sides and so can't be put under
the XL2576 device, but I put a few near to it so we'll see how that goes.
Toner Transfer
Having had a fair bit of practice by now, I find that my toner-transfer boards generally get done on the first attempt but making them is rather time consuming. I use the yellow waxy paper that is available reasonably cheaply online. I have
had no success trying to use various types of glossy paper.
As can be seen in the second photo below, using even just a little bit too much heat causes a waxy, fiberous film to be left on the traces.
Using a lot of too much heat can cause the film to stick to the copper as well and act as unwanted resist during etching.
A bit of gentle scraping and peeling with some pointy plastic tweezers takes care of it though. The copper ground plane on the front side was masked with cellophane. Here's the board after etching in a mixture of equal parts 30% HCl and 3% H202:
The etched board was drilled using a drill-press. It's best to drill before removing the toner as it makes it easier to align the bit with position of the holes. Toner is then removed by wiping with methylated spirits soaked scourer pad, soaking the board for a few minutes can make it come off easier, otherwise using a little acetone will take it off immediately.
With the front 'silkscreen' ironed on and copper removed from the non-ground pads:
A disk shaped grinding attachment for the rotary tool makes removing the copper from the non-ground pads reasonably easy.
On the workbench completed and working:
Conclusion
These two boards are working fine running a couple of Banana Pi servers that I've been building. According to the datasheet XL2596 can supply 3A continiously with built in thermal protection.
The modest heatsinking provided by the homemade PCB being the limiting factor I guess, but the BPis only idle at about 3W, 5-6W with HDD active and briefly peak up to about 10W on startup, so there
should be plenty of margin, I hope.
