[Premium-Rx] New diode mixer

[Michael O'Beirne]

Hi everyone,

    This posting is aimed at non-RSGB members who do not have access to RadCom.  
    Pat Hawker's Technical Topics column in the current January 2005 issue describes a new diode mixer that is the subject of US Patent 6,111,452 entitled "Wide Dynamic Range RF Mixers Using Bandgap Semiconductors".  The patent is in the names of Fazi and Neudeck and is assigned to the US Army.
    The inventive step is to use bandgap material such as monocrystalline silicon carbide (SiC) for the diodes in a double balanced mixer rather than conventional narrow bandgap semiconductors.
    Another wide bandgap material, GaN (gallium nitride), is currently being used to make blue LEDs and the suggestion is made in Pat's piece that it should prove equally effective for the same reasons.  [One could I suppose discuss the pros and cons of LED mixers by reference to the more traditional form of glowing mixer!!!]
    Pat says that the patent provides an excellent discussion on the operation of solid state diode switching mixers and the various methods of increasing their dynamic range, and compares a conventional silicon ring mixer with a mixer using SiC diodes in the same environment.  The results are impressive.   But there is a cost - the LO power for a SiC mixer needs to be 20dB more to provide the same 10dB loss because of the higher turn-on voltage required.  As they say in Yorkshire : "You don't get naught for naugt".
    The source reference to this patent arises from an article by David White (either WN5Y or KN5Y) in the March 2004 newsletter of the "Flying Pigs QRP Club".  
    The patent seems well worth reading, though whether one would be able to deliver the required LO power is another matter.  We already need about +17dBm to drive a high level diode mixer, so finding another 20dB will not be easy.  I could just about do it with my ancient R&S power signal generator, type SMLR.  It produced so much umph that it burnt out one pi section of an expensive external Marconi UHF attenuator, which did not please me one jot.
    I cannot but feel that a better approach to mixer technology is to use the H-Mode mixer  developed by Colin Horrabin, G3SBI, which uses a high-speed bus switch.  The receive section of the CDG 2000 transceiver (described in RadCom June - August 2002) incorporating such a mixer has an IP3 of +40dBm with a noise figure of 10dB.  It also uses a very low phase noise LO which achieved -140dBc / Hz at 9kHz offset from the carrier, and -150dBc / Hz at just over 20kHz offset on the 20 metre band.  The limitation in dynamic range was, in part, not the mixer but the coils in the bandpass filters, being 13dB better for hand-wound coils than for commercial Toko inductors.  This is rather like the old RA1772 where the limiting factor in the prototype was the IP3 of the roofing filter and not the, then, novel switching mixer using a ring of four FETs.  All the original ferrite material in the filter had to be removed in order to achieve the IP3 spec of +27dBm.  
    In the CDG2000, with an IP3 of +40dBm and a noise floor of -130dB, an IP dynamic range of 113dB was achieved.  That's a lot better than many commercial premier receivers.
    I hope this of some interest.
    73s
    Michael O'Beirne
    G8MOB
    

     
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.qth.net/pipermail/premium-rx/attachments/20041219/4356701a/attachment.htm