Horus architecture questions

[scd31]

Hello,

I'm currently working on porting CATS to the DFM-17, and later the RS-41. I thought it would be pretty straightforward to add - just copy the Horus code for modulation - but I noticed something a bit odd. It seems that rather than using the FSK modulator built into the si4063, you're re-implementing it on the microcontroller and just manually toggling the frequency of the si4063. Is there a reason for this? Any reason you don't just send the raw bytes to the si4063 and have it handle the rest?

Thanks!

[scd31]

As a follow-up question:

• DFM-17 radiosondes require a GPS lock (and clear visibility to the sky) to calibrate its internal oscillator. This is necessary, because the internal oscillator is not particularly accurate. DFM-17 transmissions, especially APRS, may not decode correctly because of incorrect timing before the internal oscillator has been calibrated.
  • The RS41 radiosonde hardware uses an external oscillator, which is more stable, so RS41 does not suffer from the same issue.

Since the si4063 has an external oscillator, if we use that to do all the symbol timing (the way I'm used to doing) wouldn't this no longer be an issue? The precise timing of the microcontroller wouldn't matter, so long as it keeps the si4063's buffer full.

I think this would also solve the issue of the APRS code having very specific timing requirements (such as magic delays). Could maybe also put the mcu to sleep when not busy refilling the TX buffer, lowering power consumption.

Anything I'm missing? If not, I'd love start converting things over in a PR once I'm done my CATS stuff. Thanks!

[scd31]

Just as an update - got FIFO mode working. It seems to be quite accurate