Thomas’ talk about the Minima at Pacificon has got me thinking about building one of my own. I actually have quite a few of the necessary parts in my junk box, but lacked a few things, so I made a quick list and sent a quick order off to Tayda. Tayda has really good prices on many of the discretes, connectors, sockets, etc… and even carries the ATMega 328 and LCD panels. Oh, and I got 25 20 Mhz crystals at $.10 each.

But they don’t carry the Si570 (not surprising). I found I could order those from Digikey, but they have an unusual footprint, and didn’t seem the most easy thing to breadboard. Most of the homebrewers seem to build the Minima on perfboard, but they seem to sketch over the details as to how they mount/solder this part down. I’ve been considering making my own version of the digital board and getting a pcb manufactured from OSHpark, and still might do that: I am reasonably capable at surface mount soldering. It might also allow me to make a couple of other changes to the digital circuitry:

• I think relying on the disabling of pull ups in the ATMega to keep from frying the Si570 is just a bit sloppy. Two resistors and a 2N7000 will implement a simple level converter, and cost about $.50 for both lines.
• I’ve actually thought it might be good to make the entire digital board run on 3.3v.
• Considering a change in the LCD as well. Might be good to have a graphical LCD, or one that runs on lower power.
• I like the idea of using rotary encoders instead of a pot.
• A basic keypad might also prove useful.

Anyway, I’m getting a bit astray. To experiment, I thought it might be good to see if I could get an Si570 on a tiny breakout board so I could breadboard this up before building it for real. While surfing around and also reading today’s mail from the Minima mailing list, I learned about the Si5351. There is some speculation on the mailing list as to whether it is a reasonable chip to use in place of the Si570. In particular, people are concerned about its noise performance. But it does at least have some intriguing features:

• It’s cheaper.
• It is a more conventional package (10-MSOP, still surface mount, but more conventional)
• It can generate 3 different clocks simultaneously (it could potentially replace both the BFO and LO)
• NT7S has already written the code to interface to it.
• Adafruit already has a nice little breakout board for it that only costs 8 bucks. It’s also already got pads to add some SMA connectors, which could make it a nice little bit of test equipment.

I’m not sure how to evaluate its performance, but it is at least possible to use it as the building block for an SSB transciever, as demonstrated by Jason NT7S using it in this reciever based upon the SA602:

And another early test in a QSO with AA7EE:

Jason has clearly been on top of this chip for a while, surfing his blog for the si5351 tag gives a ton of cool information. He even has his own board, quite similar to the Adafruit board, but it includes isolation transformers for each channel. Very nifty, and probably better suited for RF work than Adafruit’s board. Jason says he might e kitting a batch of the boards for sale: I’m gonna stay tuned.

Nifty stuff.