An FM Transceiver From An Unexpected Chip

The Si47xx series of integrated circuits from Silicon Labs is a fascinating series of consumer broadcast radio products, chips that apply SDR technologies to deliver a range of functions that were once significantly more complex, with minimal external components and RF design trickery.  [Kodera2t] was attracted to one of them, the Si4720, which boasts the unusual function of containing both a receiver and a transmitter for the FM broadcast band and is aimed at mobile phones and similar devices that send audio to an FM car radio. The result is a PCB with a complete transceiver controlled by an ATmega328 and sporting an OLED display, and an interesting introduction to these devices.

The Si4720 internal block diagram, from its data sheet.
The Si4720 internal block diagram, from its data sheet.

A look at the block diagram from the Si4720 reveals why it and its siblings are such intriguing devices. On-chip is an SDR complete in all respects including an antenna, which might set the radio enthusiasts among the Hackaday readership salivating were it not that the onboard DSP is not reprogrammable for any other purpose than the mode for which the chip is designed. The local oscillator also holds a disappointment, being limited only to the worldwide FM broadcast bands and not some of the more useful or interesting frequencies. There are however a host of other similar Silicon Labs receiver chips covering every conceivable broadcast band, so the experimenter at least has a good choice of receivers to work with.

If you need a small FM transmitter and have a cavalier attitude to spectral purity then it’s easy enough to use a Raspberry Pi or just build an FM bug. But this project opens up another option and gives a chance to experiment with a fascinating chip.

20 thoughts on “An FM Transceiver From An Unexpected Chip

  1. Years ago I did a commercial product using the Si4731. It was before the digital TV transition, so I snuck in the ability to receive audio from NTSC TV channels 3, 4, 5, and 6.

    I’m not clear exactly why Silicon Labs’s FM receivers go down to 64MHz; the OIRT bandplan’s lower band is 65.8MHz.

    1. I received yesterday one of the XHDATA D 808 RADIOS.
      As the British Army would say ‘It really is a nice bit of kit’
      I have held the call G3SBI since 1963 and a year or so afterwards I bought two AM handheld walkie talkies designed for the USA 27 MHz citizens band.They were not really legal over here but I was keen to see what range could be obtained.Although the receiver was a superhet the best I could do in a built up area was about 1/4 of a mile.
      I bought some crystals for our 4M band AM calling frequency 70,26 MHz.I built the mods dead bug style on the back of the pcb.My PA pushed out about 100mW.I used the telescopic antenna that came with the walkie talkie but attached it to one end of a parallel tuned circuit with a series tuned coupling loop to get a 50 ohm match.

      To set it all up i first tuned the PA into a 50 ohm load and then inserted an SWR bridge between that and the parallel tuned circuit/series tuned link and adjusted the two capacitors to get 1:1 SWR.
      Shortly afterwards G3RIK and I were operating /P in Scotland using an ex military B44 ( 5W out ) and a 4×4 yagi.
      The first time we drove up the road to the lighthouse on the Mull of Kintyre we stopped and I put out a call with the modified walkie talkie and got 57 from the west coast of Northern Ireland about 70 miles away.
      The point I am making is that barefoot you don’t need power to have long range contacts at VHF but that an SSB/CW facility would be highly desirable.
      Obviously if Silicon Labs do not allow user programmability of the chip that is off the agenda at least for now.
      In any SDR TX/RX suppression of the local oscillator frequency and opposite sideband could be an issue but not up to a few watts out ( using an external PA )

      It would be nice if the chip could do this on at least 6M 4M 2M and if possible the 70cm band.
      My XHDATA D 808 as received on FM covered 88 to 108 MHz.However I persuaded it to cover 64 to 108 MHz so that it covers our 4M band.Unfortunetely the firm ware doesn’t allow you multimode reception between 64 and 108 MHz and the tuning step options would be useless for receiving for receiving CW on 4M.
      However in the transceiver chip if it could generate multimode between 64 and 108 MHz it would be easy to have a HF add on with only a mixer and a 50 MHz LPF.

      regards colin

  2. Hmm. This must be the chip incorporated into the famous nokia n8 from a few years ago. It could retransmit internet radio via the fm band to be used in a car or very close to a domestic radio.

  3. I remember about 20 years ago, I was on the mailing list for one IC company and there was a string of intriguing cellphone ICs. Wait, it was Maxim, and they offered samples. I kept looking at those, knowing they had a lot of what was needed for “SDR’, but it was never clear from the bits they sent out whether they were too sculpted to cellphone applications or they could be shifted to “useful” frequencies.

    In the old days ICs were pretty general, but as they became more specific there were often limits. When AM stereo was trying to take off, I remember looking at an adapter in Radio Electronics, using a Motorola IC and realizing it was almost what was needed to make a sideband selective synchronous detector, but one stage might affect it’s use for that. It wasn’t clear, though much later I read that Sony used an AM stereo IC to bet the function in their famous 2010 shortwave receiver.

    A lot of ICs offer two mixers, and a quadrature oscillator of some sort, but can you easily shift the frequency down to more useful frequencies?

    The SI go in the other direction, certainly useful for shortwave, but can they be programmed for various modes, or are they “hardwired” for specific modes. I get the feeling modes and selectivity are adjustable, but from between some preprogrammed modes. So even if you attach your own controller to a Grundig G8 portable’s SI IC, you can get more bandwidths but no SSB.

    It also depends on what you want. If you do everything in a computer, you just need a quadrature oscillator and a pair of mixers for receive and another pair for transmit (and any receive and transmit amplifiers. That’s not as simple as a magic IC that does it all, but it’s simpler than building an SSB transceiver.


      1. Thanks Jenny,
        I’d be keen on a small as possible implementation around the 433Mhz region primarily with the same types of controls of the chip featured Or close to with discrete chips which I guess mostly unable to be done with any 74 series (which I cut my teeth on decades ago) but, with maybe a final stage up convertor. Ie. Wondering how to raise the frequency using the Si chip feeding into a small upconverter perhaps ? RF never been my strong point, though it occurs to get to approx 433Mhz with the Si chip at suitable frequency of around mid 80’s MHz with some chip offering integer multiplier to final output stage ? Eg for something like pocsag coding but, not quite ;-)

    1. The silabs chips can do SSB demodulation, but they seem to have contracts in place as they seem to exclusively supply those to tecsun and other portable radio manufacturers, but not the general public.
      Which is a damn shame.

      1. That’s the impression I got, suddenly there was an SI based SSB receiver and it sounded like. it was customized to add SSB. So all the hardware is there, but I guess firmware decides the capabilities.


      2. Back in the days when project ideas came in monthly magazines, not daily websites a common project was to add a BFO to an AM only shortwave receiver to allow it to demodulate SSB and CW signals. Receivers were almost always superhets with 455kHz IFs so the BFO was as simple as a one transistor oscillator using the if can from a sacrificial AM receiver as the tuned circuit.

        The only electrical connection to the radio was usually for power. The output of the BFO would be fed preferrably via a piece of tiny coax to the vicinity of the if can. The end of the coax would have it’s outer jacket and shield stripped off for a short length but not the insulation. The still insulated center conductor would then be wrapped around or maybe even glued to the radio’s if can. Enough bfo signal would then bleed into the if transformer to get the job done.

        So… for a while I was very interested in the Si chip that Sparkfun used to carry that had shortwave and LW along with with usual broadcast frequencies. I noticed from the datasheet that internally it was a superhet with a 455kHz if! I wanted to try the same old BFO trick with the wire glued to the top of the chip.

        I never got around to that for various reasons but what do people thing, might if have worked? Does anyone want to try it?

        I suspect that it would have worked but only with strong signals as the bfo signal would have bled into the agc too causing it to turn the gain way down.

      1. Deobfuscators or decompilers can still try to get into the firmware and/or software along with intercepting the traces or air gap (wireless) command/data streams if the NDA doesn’t want to share how to use the item invested in that is the owners tangible property and even intangible property if not used for profit to avoid arguments.

  4. Interesting, thanks for article.

    Wonder if its feasible to massage the chip so it can do digital modulation
    from a 0 to 1 binary source for +- 4.5KHz carrier modulation. Ah la local area simple
    paging ?

    Have a large of FM RF pagers with pocsag I could use in a country are for data logging
    to workers in the field as exceptions in farming like environment…

    ie. CPU spits out a binary pocsag into this chip or dev kit etc then spits out a +- delta
    on the carrier of 4.5KHz with the right polarity of course so
    the paging receiver on the plebs belt buckle gets message while they are roaming
    around the paddocks or back from the pub to manage the herd – as well as the cattle :-)

    So looking at Fig 20 of the data sheet it should be possible to fudge that maybe, ha !

    1. You will likely get significantly better range by using some ism band dedicated transceiver chip (like cc1101 from TI) than kludging a FM broadcast band transceiver to do it.
      You get more power and don’t have to share the band with stations running tens of kilowatts.

      1. Many ISM chips, including the CC1101 don’t go down into the VHF range. Check out Silabs Si446x, ST Spirit1 which go as low as 140-150 MHz, and the OnSemi AX5043 claims to go all the way down to 27 MHz.

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.