Sunday, April 6, 2014

Si5351A Circuit Build - Failure

As I have mentioned in previous posts, I want to build an Si5351A RF VFO, which is similar to the Minimal Si570 VFO.

The Si5351A has three RF Outputs that I would like to have available on my LAB Test Bench, and I will also try it as replacement for both the VFO and BFO for my experimental Farhan Minima Transceiver. The Si5351A is programmed via I2C, which is similar to the Si570.

Si5351A Three Port RF VFO
Today, I created the Si5351A PCB via the Toner Transfer Method, and I installed all of the necessary "vias", and 95% of the SMD parts. All that was left to install were the SMA RF Connectors, Power Header, and the Crystal.

I planned to install the Crystal next.

But then, . . . . Failure !

Small Crystal - Big PCB Pattern
(As Viewed Through the Microscope)
Dang, . . . I used the wrong Pad Pattern for the Crystal on the PCB, it is much too big for the Crystal that was purchased for this purpose. The Crystal is shown up-side-down in the photo (pads up) for pad comparison.

According to the Spec Sheet, the size of the CX3225SB Crystal only is 3.2 x 2.5 mm (0.125 x 0.098 inch).  The Crystal is only about as big as one of the Pads of the four Pad Pattern that I used.

I will correct the Crystal Pad Pattern in the layout, and try AGAIN!

This is one of the down-sides of not using "Ugly Style" construction, many time there is NOT an acceptable recovery.

I am getting better-and-faster at creating Double-Sided Homebrew Toner Transfer PCBs, and better-and-faster at connecting the two sides with very small vias. So, . . . I guess when necessary, I can now, Fail Faster :-)


UPDATE: Apr 7, 2014 10:45
For future reference, the Crystal Pad Pattern that I fond that appears to fit is labelled "CFPX-5". It is interesting to note that "Pin 1" is not marked, "Pin 2" is the indexed pin (see the inside filleted corner).

--

1 comment:

jbm said...

It looks like you can probably get away with floating the crystal on some "balls" of solder on the corners of the pads. At that point in the assembly, it's definitely worth giving it a shot. I don't think it will have a huge impact on the quality of your signals, unless those pads are a bit more recessed than they appear. In that case, you might want to hot-glue around the package to reduce microphonics, if you care about such things.

(Also: I always print out the top copper, put it on top of some foam, and dummy up my boards. This is the big downside to using fabbed PCBs: it can take two weeks to find out you've got a bad footprint!)