Or an Si5351, $1 on Mouser, cheaper than a mixer.
Assuming 250mhz or so max is sufficient.
The Sig-Gen mode of the tinySA is interesting in that it might be able to emit a sine wave
capable of going from 0 to 250mhz.? ?(The 250mhz LPF removes any harmonics.)
But a square wave is fine for this tracking generator.
Is it possible to approach the 120dB dynamic range of fig 31 on p61 of the datasheet?
The RBW implemented in the Si4432 will not have sufficient stopband suppression to see this,
but the SAW filters should compensate if RBW is set equal to SAW filter bandwidth.
It will be interesting to build a tinySA with proper shielding, perhaps optoisolators on
the SPI bus into the RX Si4432, a separate SPI bus into the LO Si4432
to keep the RX quiet.? See how far we can move the noise floor down.
But the noise floor is flat, and doesn't vary much when we change the rx frequency?
within the SAW filter passband.? So it's not just that the noise floor is some
local oscillator harmonic.? Suggests to me that shielding may not be sufficient,?
From p61 of the datasheet, the RSSI register gets updated once every four data bit periods,
and each time it is updated there might be 10ns or so of invalid data if you happen to read it then.
You can emit a data clock out an Si4432 GPIO pin, so perhaps interrupt the SPI master
to do a read synchronous to that data clock?? No info on exactly where in the data bit period
the RSSI gets updated.? Setting the data bit rate is a tradeoff between fast RSSI updates and
having more frequent episodes of invalid RSSI data.? ?(We aren't actually using data bits here,
but we may have to pay attention to this data bit rate thing anyway.)
On p8 of the datasheet, Tlock of the Si4432 synthesizer is spec'd at around 200us
for large changes in frequency (actually, Tlock is time to lock from initialization).
I'd imagine it is much faster than 200us for the small changes typical of a scan.
But must give it sufficient time when crossing a 10mhz band boundary.
At line 675 of the Jan 17 tinySA.ino code, reg 0x69 is left with a value of 0x60.
According to page 50 of AN440, that should leave the AGC enabled.
Perhaps Eric determined that the documentation is incorrect?
With AGC off by clearing bits 5 and 6 of reg 0x69, we can set the LNA and PGA gain
with a register write, getting up to an extra 44dB of gain.? With this gain set high, keeping
noise low becomes extremely important.? If it works, it may pretty much remove the
need for the PE4302 attenuator.? And/or could conceivably adjust the gain for each sample.
If an RSSI sample is out of range then adjust the gain and take the sample again.
But as noted above, the noise floor is probably such that this is all moot.
On page 9 of the datasheet, the LNA input impedance varies considerably with frequency.
Not an issue for the current tinySA architecture since that input is always at 434mhz.
The LO output into the mixer is overpowering, and can vary somewhat without issue.
But something to keep in mind.
Jerry, KE7ER
