Tuesday, May 7, 2013

Power Supply Digital WW

I needed a "digital" power supply (+5V and +3.3V) for a number of my small projects. I also wanted a low noise supply that didn't require building a power line interface. I did build one of those for projects requiring more power. You can read about that project at Power Up. The Power Up supply can deliver 800 mA per supply. The supply described here is only good for 150 mA per voltage and it runs a little hot at that current. So think of the 150 mA as a design limit with a more practical limit (longer life) of 100 mA or less. That is still plenty of current for today's low power designs. In addition the supply uses a Wall Wart (the WW in the name) for its AC power connection. No power entry construction required.

One more important point before you get started: the TI TPS7A4901 (data sheet) chip used for both regulators is a low noise regulator. This can come in very handy when powering precision crystal oscillators where you want to keep the noise down to maintain short term accuracy. It is not the lowest possible noise. But it is quite good for a one chip solution.

The design files (schematic, board layout, and parts list) for the board can be found at Power Supply Digital WW - design files. If you want to build one OSH Park has the boards for $11.00 each.

So lets start the build. First put the 22uF 6.3 V capacitors (C9 and C10) on the board. Then solder the U1 and U2 pins to the top side of the board. Next turn the board over and squirt some Chip Quick SMD291NL flux into the U1 and U2 grounding/heat sink holes on the back side of the board and apply your iron and solder to those holes (I use 63/37 tin/lead) until they are filled. That grounds the chips and also acts as a heat sink for them.

Now solder the rest of the surface mount components to the board. At this point I like to test the board because trying to work on the regulator chips with all the other components mounted is very difficult (read impossible) without removing parts. To do that testing you have to short out RV1 and RV2 and apply power to the chips. About 6 volts is OK for both chips because no significant power will be drawn - this is only a voltage test. The voltages should read about 2.9 volts for U1 and 4.9 volts for U2. If you don't get those voltages check U1 and U2 (the most likely culprits) for shorted or open pins. Since I plan to make a lot of these boards I built a pogo pin rig for this test using a bare board. The holes for the pins that go into RV1 and RV2 positions need to be drilled out (.040" - #60 drill) if you are using the Mill Max 0856-0-15-20-82-14-11-0 pogo pins to make your temporary connections. It will look like this with the board mounted on the test jig:
This is the test jig with the pogo pins:
I used brass washers between the nylon nuts to adjust the height of the jig.

Next mount all the 470uF 10 volt capacitors (C3 through C8). Now add in the transistors with thermal pads mounted loosely to the heat sinks. I used 6-32 by 3/8" screws and split lock washers to insure contact of the transistor with the heat sink is solid. Mount and solder Q1 first and then tighten the screw. Then Q2 followed by Q3.

Mount all the rest of the components and test the board under power. I used a 22 ohm 1% 3 watt resistor mounted on a power plug for the 3.3V section and a 33 ohm 1% 3 watt resistor mounted on the same power plug for the 5 V section. Anytime you want to test a supply under load just plug in the load. And that is all there is to it.

I wrote some more about low noise power at ECN Magazine.

Engineering is the art of making what you want from what you can get at a profit.

Sunday, May 5, 2013

Space-Time Crystals

This is a little far afield from what I usually blog about. But I couldn't resist.

Physics team proposes a way to create an actual space-time crystal
Earlier this year, theoretical physicists Frank Wilczek, of MIT put forth an idea that intrigued the research community. He suggested that it should be possible to construct a so called space-time crystal by adding a fourth dimension, movement in time, to the structure of a crystal, causing it to become an infinitely running clock of sorts. At the time, Wilczek acknowledged that his ideas on how to do so were inelegant, to say the least. Now another international team led by Tongcang Li has proposed a way to achieve what Wilczek proposed using a far more elegant process. They have posted a paper on the preprint server arXiv describing what they believe is a real-world process for creating an actual space-time crystal that could conceivably be carried out in just the next few years.

Wilczek thought that it should be possible to construct a space-time crystal because crystals naturally align themselves at low temperatures and because superconductors also operate at very low temperatures; it seemed reasonable to assume that the atoms in such a crystal could conceivably move or rotate and then return to their natural state naturally, continually, as crystals are wont to do as they seek a lowest energy state. He envisioned a rotation with a ring of ions that flowed separately rather than as a stream, likening it to a mouse running around inside of a snake laying as a circle. The bulge would flow, rather than the snake itself spinning and would just keep on going, potentially forever. The problem was, he couldn’t figure out how such a crystal structure could be created in the real world.

arxiv paper

H/T DeltaV

Engineering is the art of making what you want from what you can get at a profit.