Maybe We're Almost There

[NZ0I]

It occurred to me yesterday that there is another important discovery that Jerry made in his recent testing: there was no reported sign of instability, even with the gain-killing resistor set to zero. That indicates that we might have significantly more room to increase the IF stage gain before the SA605 starts to oscillate.

We are currently not using the IF gain stage (pins 16-19) at all, and have its output tied to a dummy load (R211). If we were to include the IF gain (~ 40 dB) it would, theoretically, increase the sensitivity well beyond our target goal of 0.2 uV.

Realistically, we are likely to see oscillations if we include the full amount of gain that the SA605 is capable of (~ 100dB), but we only need to add about 20dB more to reach our sensitivity goal. We just might be able to achieve that, and do it without any gain in the receiver front end. That's just what we were hoping for going into the project, and would greatly simplify and reduce the cost of the receiver.

Of course, by increasing the gain (and sensitivity) we will also lower the threshold at which we see limiting. But that's OK. We know there's at least one effective way for the processor to kill the gain (pin 4 bias) when the limiting level is approached.

Jerry, if you feel like doing another experiment, adding the IF gain stage might be a very interesting experiment to conduct. I will create a schematic showing how you can modify the board you are using. I will also update the Rev B PCB design to include the option of adding in the IF gain stage... so we will be able to experiment with it more easily using the new boards.

[NZ0I]

I'm not seeing an obvious way to modify the current IF Module PCB that Jerry has built, to include the gain from the IF gain stage. At least, no way to do it without removing parts, cutting traces, and adding wires. That kind of surgery would probably add a great deal of stray inductance and capacitance, and would not be representative of a properly laid out design. So I'm not recommending that we try to modify the existing board.

The application designs provided by NXP all use two crystal/ceramic filters: one between the first mixer output and the IF gain, and the second one between the IF gain and the limiter input. The SA605 is actually designed to work best when the loss provided by those two filters is provided. Also, doubling the bandpass filtering should increase selectivity... and increase the cost by $4.64. 

What I think I will do is add a footprint for a second 10.7 MHz xtal filter to the Rev B design. If we want, we can experiment with leaving it out, replacing it by a resistor and capacitor to kill enough gain to keep things stable. The receiver should benefit from the additional selectivity if we decide to keep it in the design.