[HN Gopher] Tiny Ten DSP-Based HF Transceiver
       ___________________________________________________________________
        
       Tiny Ten DSP-Based HF Transceiver
        
       Author : wglb
       Score  : 60 points
       Date   : 2025-03-02 14:34 UTC (4 days ago)
        
 (HTM) web link (www.janbob.com)
 (TXT) w3m dump (www.janbob.com)
        
       | avidiax wrote:
       | I have been wondering if/when we could get a direct synthesis SDR
       | transmitter. Obviously the amplifier and filter stages will have
       | to be discrete & analog, but to me it seems more elegant that you
       | can software define any radio signal below 1Ghz or so and pass it
       | to an amplifier.
       | 
       | That doesn't detract from this project, which is quite an
       | achievement for a single hobbier.
        
         | 01100011 wrote:
         | After working at a company using SDRs, I realized that a well
         | performing SDR is a lot more than an antenna, a DAC/ADC,
         | amplifiers and a processor. You still have analog paths you
         | need to worry about and for many applications those aren't
         | going to cover your 1GHz range. You need to worry about noise,
         | amplification, signal fidelity, etc. You still need a good
         | portion of what makes up a traditional radio. Sure, there are
         | narrowband, low power SDR xcvrs and xmitters but that's not the
         | same thing.
         | 
         | That said, a long time ago I made a test fixture for a 6-bit,
         | 12GHz DAC and it was cool to run it as an arbitrary waveform
         | generator.
        
         | RF_Savage wrote:
         | The popular IC-7300 transceiver from Icom does that. Receiver
         | samples the whole band and does downconversion in the digital
         | domain (DDC) and same for the transmitter (DUC).
         | 
         | On the hobbyist side there is the TRX Wolf by UA3REO.
        
           | th0ma5 wrote:
           | Sure, just do the Zeta SDR schematic in reverse, but like
           | others have said you'd have to figure out the rest of the
           | chain as well as a tone of filtering. I've transmitted with
           | direct digital synth chips as well as rpitx which is
           | literally bit-banging RF off of a pin on a raspberry pi with
           | NTP sync even. Again, still had to do a lot of filtering, but
           | with wsprrypi I was able to be heard one time in New Zealand
           | from Ohio, USA without any amplification.
        
             | ale42 wrote:
             | How much RF power is actually coming out from the RPi
             | transmitter? A few mW?
        
               | th0ma5 wrote:
               | Yes maybe a couple of hundred milliwatts at most.
        
             | RF_Savage wrote:
             | Naturally there needs to be filtering and some sort of
             | analog frontend. My main point was that mainstream rigs now
             | do direct sampling, instead of zero-if or superhet with dsp
             | IF.
             | 
             | And WSPR sure is magic, people do thousands of km with it
             | off an rp2040 gpio pin.
        
               | th0ma5 wrote:
               | Yes sorry for the double reply I didn't know they were
               | using the rp2040! Thanks for mentioning that !!
        
               | topspin wrote:
               | I'm not sure this is what you have in mind as far as
               | "mainstream" rigs or architecture, but the FTDX-101D main
               | receiver is a "narrowband SDR" system, as Yaesu describes
               | it. Signal is downconverted through an analog chain
               | including a narrow crystal filter before it reaches the
               | 18 bit A/D+FPGA. And yes, it does perform very well,
               | currently at the top of the "Sherwood" Third-Order
               | Dynamic Range ranking, where it's been for about 5 years
               | now.
               | 
               | If I'm interpreting your view correctly, then I share it:
               | there is a lot of performance to be had with a well
               | designed analog front end, and the specs of the digital
               | parts can be quite modest and still perform extremely
               | well.
        
               | nullc wrote:
               | Quite a few radios are direct-- no intermediate
               | frequency, just a set of band selection filters. The
               | direct route does not quite hit the absolute state of the
               | art in dynamic range, but it's close. IC-7300, IC-7610,
               | Elecraft K4, etc. all have this design. I wouldn't be too
               | surprised if one more generation of ADC improvements take
               | it over the line. Certainly there is a lot more
               | engineering might being put into better ADC designs than
               | there is for superhet radios.
               | 
               | Beyond being simpler on the analog side the direct
               | conversion route makes spectrum/waterfall views of the
               | whole band trivial. In theory radios could use this wide
               | bandwidth for improved impulse blanker performance, but
               | I'm not sure which if any do.
               | 
               | Some (like the flex radio, and Elecraft) will do multiple
               | in-one-band simultaneous receive with on each ADC using
               | it... but (other than flex perhaps) the advantages of
               | direct converting a whole band are currently under-
               | utilized. Once they are I doubt there will be much
               | interest in superhet even if the dynamic range isn't
               | quite matched.
               | 
               | I think the future of radio designs though will be
               | finding ways to move the RF portion closer to the
               | antenna(s), away from RF-noisy buildings, and avoiding
               | expensive, lossy, and annoying coax runs.
        
         | th0ma5 wrote:
         | Sure, just do the Zeta SDR schematic in reverse, but like
         | others have said you'd have to figure out the rest of the chain
         | as well as a ton of filtering. I've transmitted with direct
         | digital synth chips as well as rpitx which is literally bit-
         | banging RF off of a pin on a raspberry pi with NTP sync even.
         | Again, still had to do a lot of filtering, but with wsprrypi I
         | was able to be heard one time in New Zealand from Ohio, USA
         | without any amplification.
        
           | mschuster91 wrote:
           | > Again, still had to do a lot of filtering, but with
           | wsprrypi I was able to be heard one time in New Zealand from
           | Ohio, USA without any amplification.
           | 
           | Hope you got a QSL card out of that kind of DX effort. What
           | antenna did you use?
        
             | boxerbomb wrote:
             | I would imagine he was talking about running WSPR.
        
             | th0ma5 wrote:
             | Yes WSPR ... The acknowledgement of contact was online. I
             | had an MFJ antenna tuner but it was just a long strand of
             | wire lying over the roof maybe 14 gauge? Slim. Something
             | like 100mw output power? 200mw Maybe? It ran for a couple
             | years unattended and was regularly heard east of the
             | Mississippi and sometimes with some regularity in Europe.
        
         | picture wrote:
         | We've been in the age of direct RF for a while. The new
         | cellular base stations that Nokia makes are literally eight
         | RFSoC chips connected to 64 power amplifiers. You can get one
         | for yourself for pretty cheap:
         | https://www.realdigital.org/hardware/rfsoc-4x2
        
       | _benj wrote:
       | I loved the article! Even though I only understood about 1/3 of
       | the technical details, I was honestly quite amazed at how
       | "simple" was producing RF! The guy is using "sound" at 48kHz to
       | produce the signals needed and then a Si5351A (signal generator
       | for up to 200MHz or so controlled via I2C) to produce RF. I'm not
       | sure, or don't understand yet how the RF is being modulated, but
       | conceptually is understandable! DSP is a field I want to dig
       | deeper into, specifically DSP as it pertains RF.
        
         | picture wrote:
         | The reference signal generated by Si5351A is used to clock two
         | flip flops which I believe are used to generate a 90deg
         | quadrature local oscillator (LO) which enters a dual 1-4
         | multiplexer acting as a Tayloe quadrature mixer. And as you
         | know, the nonlinear switching action introduces frequency
         | products and allows frequency conversion.
         | 
         | This is a pretty clever system. The RF is low enough frequency
         | that faster logic chips can be used effectively. I recommend
         | getting yourself a cheap receive-only SDR for around 15 bucks
         | and play with GNURadio, if you want to get more familiar with
         | DSP for telecommunications/radio detection
        
           | _benj wrote:
           | Thanks! I think the limitation is in my knowledge and not in
           | the article or schematics!
           | 
           | I have a rtl-sdr and I've played with gnu radio, but the
           | signals that I had to explore were very limited! Apart from
           | few remotes for fans and stuff around the house I only had fm
           | radio and little more.
           | 
           | What I find very interesting from this is the possibility of
           | taking a Si5351A breakout board, configuring it with I2C (I
           | can do that), modulate it with something like audio and a
           | mixer (I can code that... maybe) and plug that directly,
           | likely through and attenuator, to gnu radio!
           | 
           | It would certainly be a lot easier to get a sdr transmitter,
           | but I wasn't aware of the possibility of modulating a Si5351A
           | with a "simple" signal like 48kHz audio!
        
       | gherlein wrote:
       | Here's another: https://www.4sqrp.com/T41main.php - I just bought
       | one that someone else had assembled. It's 100% open so you can
       | buy boards and make your own, but the kits themselves are no
       | longer available. I am thinking of ordering a set to take the
       | design further, as a learning exercise.
        
       ___________________________________________________________________
       (page generated 2025-03-06 23:02 UTC)