Many years ago, I had only a Garmin GPS-38 and wanted to try using an external antenna with it, so I built a re-radiating antenna setup. (This was before inexpensive commercial re-radiating sets were available). A re-radiating setup needs a source of power for the active receiving antenna, and I designed and built a power supply for this. Then I built another better one, having learned some things along the way. I wrote a couple of Usenet articles about their construction; a summary of those early articles can be found here. In particular, this contains a circuit diagram for the power supply.
I didn't have a digital camera at the time, and never got around to shooting film and scanning it, so the text referred to above has remained the only information available. But I recently bought a Canon G2, which is pretty good for technical photography, and someone who is planning to build one of these power supplies asked me for some photos. They are below.
Note that if I was starting to build one of these now, I would take a slightly different approach. If your active antenna is designed to operate over a wide voltage range (e.g. 2.5-5 V), the power supply does not need a voltage regulator. That eliminates most of the electronic parts. I would likely use a CR123 3 V lithium camera battery as the power source because of its low weight. A Usenet article that describes this simpler option is available here.
This photo shows both power supplies assembled. The first version is on the left. It is powered by an external 9 V battery connected to the battery clip shown. The second version of the power supply is on the right. It uses a 9 V battery too, but the battery is mounted internally. Both power supplies are in die-cast metal boxes sold by Hammond. Metal boxes were used for RF shielding as well as mechanical strength.
Here the Version 1 supply has been turned 180 degrees and the cover removed. The voltage regulator circuitry is on a PC board mounted on standoffs. The 78L05 regulator is nearest to the camera, with a metal heat sink clipped onto it.
Here is another view that shows the RF path better. The active antenna is attached to the BNC jack on the left while the re-radiating antenna connects to the jack at the top. The RF signal travels between the centre pins of the two jacks via the 100 pF ceramic capacitor shown sitting there on a diagonal. The case alone should connect the grounds of the two jacks, but I added a couple of grounding lugs and a wire "just in case". The 5 VDC power for the antenna is provided by the choke visible just to the left of the RF-passing capacitor.
This version is an example of poor RF design. The 1.6 GHz GPS signal does not have a 50 ohm signal path through the box, and there will be reflections where the RF capacitor connects to the two BNC connectors. However, this wrong-impedance section is so short, far less than a quarter wavelength, that the two reflections are nearly 180 degrees out of phase and almost cancel. So it works anyway. But if you want to be sloppy like this, keep the distance between the two BNC connectors small.
Here is the Version 2 supply with its cover removed. About 2/3 of the interior space is the battery compartment on the right. A metal partition was glued into the case to keep the battery in its place, away from the electronics. A couple of pieces of weatherstripping foam keep the 9 V battery from rattling around.
All the interesting stuff is in the left end. A power switch is mounted on the end of the box. The two BNC connectors are mounted on opposite sides of the box. The RF path between the connectors is now provided by a printed circuit board soldered to the two BNC connectors; you can see only the edge of it in this photo.
Here is a closer view of the RF path, shot from an angle. Now you can see the whole RF circuit board. The coil of green wire is the hand-made RF choke that feeds DC current to the active antenna; it is connected to the power supply (red wire) on the left and via a circuit board trace on the right.
Here's an even closer view of the RF circuit board. Note that the green coil is only about 6 mm long. This circuit board provides a constant-impedance 50 ohm transmission line for the RF signal using a technique called "microstrip" construction. It's actually very simple: one side of the circuit board remains covered by copper, which acts as a ground for the circuit, while all the signal traces are on the other side of the board. Each signal trace on the other side of the circuit board acts as a transmission line whose impedance depends on the width of the trace and the type and thickness of the PC board. Here, I used a line width that should give 50 ohm impedance. The line goes straight between the centre pins of the two BNC jacks. You can see the 47 pF surface-mount capacitor soldered over a gap in the RF signal trace; this is the DC blocking capacitor.
(Yes, that's a fibre of something in the picture. It's probably a strand of cotton from the Q-tip used to clean the solder flux off the circuit.)
This photo shows the rear of the RF board. None of the copper has been removed from this side of the board. At each end, an L-shaped sheet brass bracket is soldered to the BNC connector body and to the circuit board to provide an excellent ground path for the RF signal. The negative side of the voltage regulator circuit is also connected here via the black wire. You can also see a small white chip capacitor sitting in a notch in the edge of the board, soldered to both top and bottom sides. This is an RF bypass capacitor, and probably isn't needed.
Here is the voltage regulator board. It's mounted in the bottom of the case; you can see the 78L05 in the first photo in this section. To give you a sense of scale, the large black 78LO5 package on the right in the picture is about 5 mm in length.
This time, I used the surface-mount construction method rather than through-hole. The middle horizontal foil is ground and the lower foil is the 5 V output. Input is on the upper left, and the two wires coming in from the upper edge of the photo connect to the power switch. You can see the large input capacitor at the upper right, and three smaller capacitors in the lower row.
The completed circuit board is glued to the bottom of the case. The case itself acts as a heat sink for the 78L05 regulator, which is mounted with its flat face against the aluminum case. Some white heat sink compound is applied between regulator and case.
(I apologize for the uneven lighting. The board is effectively buried at the bottom of a steep-sided box and it's difficult to get light in there.)