Re: Are the uBITX receivers ripe for improvement? -- And some other miscellaneous thoughts.

Tom, wb6b

I don't see a way to re-edit a post, so here it is again, I hope, in an easier to read font. It looks like the issue is with writing in another editor and pasting that into the on-line editor here.


I see a lot of good conversations here about improving the uBITX transmitter output. However, I see very little about how good the receiver is. My question is, have many people done comparisons of the uBITX receiver vs their other radios; particularly on SSB?

In my case I have noticed that my SDR using a cheap dongle seems to outperform the uBITX receiver substantially. Even though I don’t consider the SDR dongles to be the pinnacle of RF design, I guess throwing brute force real time math at the receiver side makes up for some of it.

Are the uBITX receivers ripe for improvement?

Additional miscellaneous thoughts:

Rather than cutting up the uBITX or trying to cram more software into the one Arduino Nano on the uBITX I was pondering running multiple Nano’s communicating on the I2C bus. They are only about $3.50 apiece. (The 3.3v bus [albeit barely] is in spec for the Nano.) 

But, rather than cutting up the radio I think I’d rather simply treat the uBITX as a peripheral to a Raspberry Pi. I’d place the Pi (and USB audio adaptors) in a metal box packaged in the same case as the uBITX.

What I’d do with the Pi is: 

  1. Process and level the microphone audio for transmitting.

  2. Process, level and noise filter the receive audio, as well use a DSP library to add bandpass and notch filters.

  3. Use one of the digital mode libraries for digital mode communication. 

  4. Maybe get rid of the T/R audio pop while processing the receive audio. 

  5. Control other gizmos I may add to the transceiver.

  6. Put a fancy touch screen display on the whole thing.

Additionally, I’m doing a literature search to see what work has been done on training Neural Nets to decode Morse Code. It would seem that a technology that can identify people, dogs and cats from low resolution photos (although there was a study that found AI could mistake pictures of muffins for chihuahuas), could finally realize an adaptable, reliable CW decoder. (This would run on the Pi, also).

I’m sure others must be working on the same things. Anyone know about such projects. That would save time over starting from scratch.

Tom, wb6b

Join to automatically receive all group messages.