I’ve Just finished and tested the Master Oscillator! This board is responsible for providing the global reference frequency for the entire spectrum analyzer. It’s based on a monolithic oscillator with a quartz crystal. In addition to the basic oscillator, it includes several inverting buffers and power supply filtering. In the above photo, the metal-capped device is the oscillator, along the top margin is the main power supply and conditioning bits, including a 7805 5-volt fixed regulator, a 10 Ohm resistor (it helps isolate noise from elsewhere in the analyzer) and a collection of capacitors. Along with each inverting buffer (the inversion is immaterial, it’s just a cheap buffer) there is another 10 Ohm resistor and capacitor for local power conditioning. Finally, there is a 33 Ohm resistor between each buffer and its output. These help match the impedance between this board and whatever it’s connected to. There isn’t a lot to it, really – one of the simplest devices in the analyzer.
As it turns out, not only is this device simple to describe and construct, it’s one of the easiest to test. To begin, I hooked it up to power and measured it’s current. The way I did this was through the control board (so it’s tested in it’s natural environment) measuring the current draw with this connected and not. The results are that the control board draws about 40 mA w/o the oscillator and about 100 mA with. The net result is that the oscillator draws about 60 mA.
Next, I decided to check the frequency of the oscillator to see if it was in the ballpark. The tested value is shown in the image above (63.993 Mhz), and is within about 6 khz of “perfect”, though it remains to be seen whether the frequency counter or the oscillator are more “correct”. As it stands, I’m pretty happy with this result.
I also hooked an output to my oscilloscope. I’ve never tried taking pictures of the screen before, so hopefully my skills in this area improve. For now, thank god for digital photo editing software. Anyway, the waveform looks generally good, though I’m not completely sure what it’s supposed to look like. There is some strange shape on the rising and falling edges. I’ve decided to avoid reading too much into these shapes because there are many things that could cause them. I’m using some cheap 20Mhz probes, for instance, and the waveform is 64Mhz. For now, I’m assuming that the probes are greatest source of error. If I decide to put SMA connectors on this board (I probably won’t) I’ll be able to ensure a solid end-to-end 50 Ohm impedance path to the O’scope and get a better image. Like I said, I’m not worried about it, so I probably won’t.
In an attempt to reduce some of the cost of PCB, we’re looking to reproduce some of the most stable designs using a PCB program other than expressPCB. As I’m most familiar with CADsoft Eagle, I volunteered to design an implementation of the Master Oscillator schematic. This is by all means not a final design, but more of a starting off point.
The image above is a rendering of component side of the Eagle version of the master oscillator board. The rendering was made using Gerbv.
The “solder” side is pretty boring. The large orange pads are exposed metal for use while soldering the ground part of either the direct-soldered coax or connectors. This image is also generated using Gerbv, and the horizontal stripes won’t actually be seen, they’re an artifact of the polygon fill. Also, this image is as if you’re looking through the PCB, that’s why all the text is backward.
Here are the files, if you’re interested. Using the .zip file if would be possible to order this board from advanced circuits (4pcb.com) and maybe iTead studio. The .sch and .brd files will open in Eagle.