In the last project log, I gave a bunch of options for improving the design if it wasn't good enough after the most recent set of changes. Taking all of those changes at once yields a design that has the controller and fan-out buffer running at 5v, the GPS module being fed from a 3.3v LDO, and the oscillator and analog section being fed from a second 3.3v LDO (potentially a high performance, low noise one, like the MIC5323, for instance). Since the controller is running with 5v, it can be clocked at 20 MHz, which offers the opportunity to double the error granularity measurement without increasing the measurement time.

But 20 MHz isn't as useful as a frequency standard. Most lab instruments want 10 Mhz.

Fortunately, the controller offers not only a solution to that, but a nice extra feature.

The ATTiny4313 has two timers in it. We're using the 16 bit timer as the accuracy measurement tool, and in the current design, we're just turning the 8 bit timer off. But it can also be clocked from the system clock, and has the ability to output a square wave that represents a divide-by-2n of the system clock, where n is any value between 1 and 256.

So I sat down and fleshed out a potential future design that has not only the 3 LDOs, but also moves the fan-out buffer chip from taking the oscillator output to taking the output from the controller instead. Two of the unused pins of the controller are connected to DIP headers that can be used with jumper blocks to select one of four output frequencies. In my default design, the choices are 10 MHz, 5 MHz, 1 MHz or 100 kHz. Not only that, but the controller has the ability to turn the output on and off. The plan is to turn the output off when the lock is nonexistent or unacceptable. That sort of obviates the need to worry about the accuracy during holdover periods. Of course, since the firmware is open, this can be enabled or disabled as desired.

But all of this a look into the future. I'm still waiting for boards to arrive that will hopefully bring the accuracy and the stability closer to what the parts should be able to achieve.