With 8 employers through the years, there have always been dedicated PCB designers, so I've never had to learn a PCB tool properly. At one place I used a router to mill out prototype boards, but I just pushed polygons around in ADS to do the artwork (wasn't tied to a schematic except for some microstrip structures). I did use Eagle to do a couple boards many years ago, but didn't keep up with it. I do mostly RF, and I often just use x-acto and/or dremel to make simple stripline proto boards. A few weeks ago I set out to take the plunge and learn a tool to make a board. I don't like what's happened to Eagle, so I didn't go that route. I downloaded KiCAD, but it barfed on the Win7 computer I was using. Then I downloaded DIPtrace. I have to say that with only the video tutorial I was able to make a board fairly easily. Made footprints for two couplers, a SAW filter and an edge mount SMA. The only thing I found strange, and I don't know what it's like in other tools, but to place a bunch of ground vias when I was done routing, I had to connect them to ground for the copper flood to work.
After Googling I picked OSH Park for the board house. Used 4 layer to get reasonable 50 ohm microstrip width (19.3 mil, 31 mil gap to ground). Not much discontinuity with 0402 series part, and I can cross the gap with a 0402 if I need a shunt somewhere. Since 19.3 mil was narrow for my SMT center pins, I dropped the ground reference below the SMT center pin to layer 3 to make a wider 50 ohm pad.
Everything went well. Cost was $11.50 (free shipping, 3 boards). Bare board shown with a couple of the parts to be mounted (brd.jpg). I decided to reflow on a hot plate with some Zephpaste 63/37 that I bought many years ago. I tried first with a few of the 0 ohm resistors. After reflow, slip board over to metal plate. That went ok, so I did the other parts. Used a dental pick to put paste on pads. Not a great way, but for doing only one board, it worked. Hand soldered the SMAs.
The board is to test a GPS notch filter topology. It can be configured for four different arrangements (large or small coupler), one or two SAW filters. I only had a few SAW filter samples, so I wanted one board to test all arrangements. Simulations suggested that the large coupler with one SAW would be the winner (which it was), but I wanted to try out all possibilities. Simulated results with s-parameter component models (didn't model any board traces) matched measurements reasonably well. Tweaked inductor to get more rejection at 1575 MHz.
If I do another test board like this, I think I'm going to try JCLPCB. They have a 4 layer stackup where layer 3 isn't so far from the top, so if I want to drop a ground reference for an edge mount SMA I can make the pin trace more like 30 mils. This board was for 1.6 GHz, so I wasn't too worried about the microstrips, but I also do some things at 6 GHz.
From this experience I would recommend DIPtrace and OSHpark to anyone wanting to learn a PCB tool to make a simple inexpensive board. The free version is "two layers", but it is unlimited for power planes, so this I guess is considered a one layer board (even though there are cut outs on layer 2, and solder pads on the back). OSHpark had a large copper to outline space, so I did file the board edges so that my SMA launch wouldn't have a gap.