Loading ...
Sorry, an error occurred while loading the content.

Re: Switches vs Mixers: Confusion

Expand Messages
  • jabauzit
    After reading the original post and the different answers it seems that there is some confusion and I want to contribute to this thread. I do not pretend to be
    Message 1 of 19 , Jun 2, 2007
    • 0 Attachment
      After reading the original post and the different answers it seems
      that there is some confusion and I want to contribute to this thread.
      I do not pretend to be 100% right and any criticism will be welcome:
      one learn by his mistakes.

      --- In softrock40@yahoogroups.com, "Loren Moline" <DSNman@...> wrote:
      >
      > Hello...Can anyone tell me what the advantages of using switches or
      > conventional mixers for obtaining the I & Q signals for SDR's....

      The title of the thread should have been Tayloe Mixer-Detector/QSD vs
      conventional mixers. The QSD, which uses a bus switch chip, is a
      mixer. Among the conventional mixers the H-mode mixer or the DBM are
      commutation mixers: commutation -> commutator -> switch. For example
      in a DBM the diodes are used as switches and the harder you drive them
      the more they behave like switches, the best is to drive them with a
      square signal.

      Using a bus switch chip as a mixer does not imply using a QSD: the
      Pic-A-Star originally used a SBL-1 as a mixer which has been replaced
      by a commutation mixer using a bus switch chip.

      --- In softrock40@yahoogroups.com, "Kees & Sandy" <windy10605@...> wrote:
      >
      > I believe the main difference is that the switches allow
      implementation of a "discrete time sampling detector" .......which is
      not really a "mixer". "Mixer" is a "dirty word" in many SDR oriented
      groups....

      It is a mixer, not a conventional one but nevertheless it is a mixer.
      I do not think that "mixer" is a "dirty word" in many SDR oriented
      groups, the "dirty word" would be "preamplifier". Personally I do not
      think that there is any "dirty word": the original SDR-1000 did not
      have any preamplifier but Flex had to introduce the RFE board for
      performance reasons. It is true that a preamp should be avoided for
      best dynamic range and IMD characteristics but with the components
      available today compromises are unavoidable.

      > It does have it's limitations due to 4x, or even 2x, Fo requirements ...

      The 4x or 2x Fo requirement is not a requirement of the QSD, it is
      just a convenient way to generate 2 signals in quadrature over a wide
      frequency range: the Time Machine which uses 2 DBMs uses a LO at 4x Fo
      which is fed to a PLD and the PLD outputs 2 signals in quadrature. The
      SoftRock V5 which uses a QSD uses a LO at Fo which is fed to a RC
      phase shift network producing the 2 signals in quadrature. A
      quadrature hybrid could also be used. The negative side of the RC
      phase shift network and quadrature hybrid is the limited range of
      frequencies for which the 2 output signals are in quadrature.

      --- In softrock40@yahoogroups.com, Chris Albertson
      <chrisalbertson90278@...> wrote:
      >
      >
      > It's a "mixer" but really we are getting into semantics. ....
      > The switch does this nicely,
      > so do diodes. So it's not "switch vs. mixer" it's "what kind
      > of mixer"
      >
      > Switches just happen to make good mixers but only in the low
      > bands. You can't get them to work up in UHF and maybe not
      > even in 10 meters
      >
      This confirms my thinking.

      Besides the confusion between switches/mixers and LO 4x requirement
      the answers regarding the differences between conventional mixer/QSD
      are right on the mark. The QSD is superior for HF using available
      components but to make a VHF SDR today you must use conventional
      mixers: see the R2 series of receivers or the Time Machine for example.

      To use the bus switch chip in a QSD up to 10m and may be 6m one should
      look at the excellent work done by Giancarlo- i7swx:

      --- In softrock40@yahoogroups.com, "Giancarlo" <i7swx@...> wrote:
      >
      > Hi all,
      >
      > Kees is right.
      > I have a SoftRock V6 driven by my 2xLO digital quadrature generator
      > with manual adjust (see Files). It uses a 74ACT74 and the SR is OK up
      > to nearly 30MHz. The squarer uses a 74AC86 which limit is just over
      > 60MHz.
      >
      > There are some new fast ICs to be tried and certainly will work OK
      > and permit the FST3253 to reach 50MHz. I expect to try them as soon
      > as possible (soon ???).
      >
      > For squarer SN74LVC2G86 or 74LVC2G86 (2x XOR);
      > For FF SN74LVC2G74 or 74LVC2G74 Fc = 280MHz typ (2x FF)
      > timings seem to be very balanced (datasheet).
      > Another possible improvement should be the replacement of FST3253
      > with a FST3125 (best Pericom PI5C3125 Ton=Toff= 0.5nS) driven by
      > 74ACT00 to get 25%T on each phase.

      Jean-Claude PJ2BVU
    • Kees & Sandy
      Good discussion on mixers there are still some disagreements, but we re all learning. Since some of the analog switches have t-plh and t-phl, break before
      Message 2 of 19 , Jun 2, 2007
      • 0 Attachment

        Good discussion on "mixers" there are still some disagreements, but we're all learning.

        Since some of the analog switches have t-plh and t-phl, break before make, etc problems and those which do have break before make have other time constraints ....has anyone looked into (and tested) gating the switches by driving the OE lines ? For example, set up the select on S0/S1 at 90 degree intervals and gate the select using OE. Since you would like this to be "fixed", maybe potentiomenter adjustable, make it 5-10ns? "gate off" during select transitioning. At lower frequencies the analog sample would still be close to 90 degrees and drop off as the frequency increases.....but hopefully you get away from the switch overlap, at least to some extent.

        73 Kees K5BCQ 

      • N0UU
        A little different view on mixing here based upon reading early writing on mixers. With traditional diode mixers the signal and switching waveform shared the
        Message 3 of 19 , Jun 2, 2007
        • 0 Attachment
          A little different view on mixing here based upon reading early
          writing on mixers. With traditional diode mixers the signal and
          switching waveform shared the same path. You wanted the rise time of
          the switching (LO) signal to be as rapid as possible because the RF
          signal rode up and down on the LO signal and undesired things happened
          there. So the 'stronger' mixers used higher power to get to the more
          vertical part of the LO signal quicker and avoid that area as much as
          possible. When the diode was off or saturated not much happened
          unless the RF was very strong in comparison...rare.

          With the FET mixers or switches, the LO is not in the signal path
          directly. Also because these devices are designed for fast switching,
          for relatively low frequencies the rise and fall times are pretty
          vertical with low switching drive (LO). I SUSPECT that one of the
          reasons that they start to fail at 10 and 50 MHz is that the switching
          turn on/turn off starts to look more like a sine wave and the nice
          sharp edges are gone.

          Hopefully some new switches (for the Gigahertz LAN for example) will
          be faster, but I suspect that like other new products, they will be
          buried in some complex chip and not useful to us.

          Enjoy Tony's products on the low bands and not too bad on the higher
          ones with some reduced performance.

          Can't wait to see what front end the new SDR5000 uses.

          Flames?

          N0UU
        • Kees & Sandy
          That s my understanding. The actual LO signal is never combined with or part of the RF signal with the FET switches. The end result, however, is basically the
          Message 4 of 19 , Jun 2, 2007
          • 0 Attachment

            That's my understanding. The actual LO signal is never combined with or part of the RF signal with the FET switches. The end result, however, is basically the same but with some pros and cons. It's all a matter of what goals you are after with the design. 

            I would also be very interested in what the SDR5000 uses, especially since Gerald continues to do a fantastic job and those guys are right down the street.

            73 Kees K5BCQ

            -- "N0UU" <stoskopf@...> wrote:

            A little different view on mixing here based upon reading early
            writing on mixers. With traditional diode mixers the signal and
            switching waveform shared the same path. You wanted the rise time of
            the switching (LO) signal to be as rapid as possible because the RF
            signal rode up and down on the LO signal and undesired things happened
            there. So the 'stronger' mixers used higher power to get to the more
            vertical part of the LO signal quicker and avoid that area as much as
            possible. When the diode was off or saturated not much happened
            unless the RF was very strong in comparison.. .rare.

            With the FET mixers or switches, the LO is not in the signal path
            directly. Also because these devices are designed for fast switching,
            for relatively low frequencies the rise and fall times are pretty
            vertical with low switching drive (LO). I SUSPECT that one of the
            reasons that they start to fail at 10 and 50 MHz is that the switching
            turn on/turn off starts to look more like a sine wave and the nice
            sharp edges are gone.

            Hopefully some new switches (for the Gigahertz LAN for example) will
            be faster, but I suspect that like other new products, they will be
            buried in some complex chip and not useful to us.

            Enjoy Tony's products on the low bands and not too bad on the higher
            ones with some reduced performance.

            Can't wait to see what front end the new SDR5000 uses.

            Flames?

            N0UU

          • Bill Dumke
            A diode mixer wants a high level LO source to keep it out of the active region. It is desirable to have it be either on or off but not in the active region
            Message 5 of 19 , Jun 2, 2007
            • 0 Attachment
              A diode mixer wants a high level LO source to keep it out of the active
              region. It is desirable to have it be either on or off but not in the
              active region where there is loss. This permits in transmit mode, more
              output power, and in the receive mode less intermodulation, in other
              words it is more linear with high input level received signals in a
              mixer application.

              However, for best noise figure in a receiver, diode mixers want to be
              run at less LO power. High LO power will cause the junction temperature
              to rise, resulting in worse noise factor, since the noise factor (noise
              temperature) is proportional with temperature. So there is a compromise
              in a receiver that much be reached with respect to LO power.

              Diodes have an exponential VI curve. By playing with various
              coefficients in the fabrication of a diode for use in a low noise mixer,
              it is possible to get almost a square knee, and therefore much less
              loss, and a better noise figure. One of the professors at the
              University of Virginia in Charlottesville used to build these for use in
              low noise figure millimeter and sub millimeter wave radio telescope
              mixers for the National Radio Astronomy Observatory. in the 1970's. In
              this case they almost behave like a true switch, only in this case
              operating even into the submillimeter wave frequency range.

              But an active mixer like a JFET at HF or GaAsFET at the UHF and
              microwave frequencies is always better than a diode mixer, due to its
              gain which helps overcome the noise temperature of the IF amplifier
              following the mixer, and it's lower noise temperature. With a
              conventional diode mixer the loss adds directly to the noise figure of
              the IF amplifier. But an active mixer has to be biased correctly and
              have just the right LO power to perform well.

              Diodes are usually just poor switches, but they have the big advantage
              of being usable as mixers well into the UHF and microwave frequency ranges.

              The switches used in the SoftRock series mixers are designed for digital
              use, not for this application. They are used because they are very
              inexpensive. The sound card limits the dynamic range of the receiver.

              Bill wa9pwr

              Loren Moline wrote:
              > Hello...Can anyone tell me what the advantages of using switches or
              > conventional mixers for obtaining the I & Q signals for SDR's. I
              > believe that with mixers you feed the 2 with the LO 90 deg different
              > in phase. I believe that using mixers may overcome the frequency
              > limits of switches but must be reason why switches are used alot. Thanks!
              >
              >
              >
              >
              > Yahoo! Groups Links
              >
              >
              >
              >
              >
              >
              >
            • Marco IK1ODO
              ... Not only, Bill. They have an almost ideal switching characteristic, and the resultant dynamic range is excellent. The standard SoftRock is close to 100dB
              Message 6 of 19 , Jun 2, 2007
              • 0 Attachment
                At 20.46 02/06/2007, Bill wa9pwr wrote:

                >...
                >The switches used in the SoftRock series mixers are designed for digital
                >use, not for this application. They are used because they are very
                >inexpensive.

                Not only, Bill. They have an almost ideal switching characteristic,
                and the resultant dynamic range is excellent. The standard SoftRock
                is close to 100dB of 3rd order IMD dynamic range in 2.4kHz, if used
                with an high performance audio interface. With some modifications
                (substitution of the original opamps with one LT6231) the sensitivity
                is -125dBm (always in 2.4kHz). Similar results are difficult to reach
                with a traditional diode mixers, and are one of the reasons for the
                success of the SR series.

                The QSD mixer could be further improved; I'm working (currently very
                part-time) on this topic. See
                http://www.spin-it.com/sdr/ik1odo_sdr1.html and
                http://www.spin-it.com/sdr/ik1odo_sdr2.html

                73 - Marco IK1ODO / AI4YF
              • Bill Dumke
                Marco, Thanks for the info.. That is very good work you have done on the QSD. I have a couple of questions. Do you believe the Johnson counter is essential
                Message 7 of 19 , Jun 2, 2007
                • 0 Attachment
                  Marco,

                  Thanks for the info.. That is very good work you have done on the QSD.

                  I have a couple of questions.

                  Do you believe the Johnson counter is essential for proper operation of
                  the QSD? Kees brought this up in an earlier e-mail. The SoftRocks
                  don't use Johnson counters at present.

                  What ham band were you doing your tests on? I don't see it mentioned on
                  the web page, SDR-1. Also you didn't mention it in your reply to me
                  about dynamic range.

                  Have you seen your MDS degrade as the frequency is raised above 10 MHz?

                  Do you believe your QSD will ever work well on 10 meters? A lot of us
                  would like to use it on the higher frequency bands. I am particularly
                  interested in 15 meters myself, as well as 10 meters.

                  Bill wa9pwr
                • N0UU
                  ... driving the OE lines ? For example, set up the select on S0/S1 at 90 degree intervals and gate the select using OE. Since you would like this to be
                  Message 8 of 19 , Jun 2, 2007
                  • 0 Attachment
                    >....has anyone looked into (and tested) gating the switches by
                    driving the OE lines ? For example, set up the select on S0/S1 at 90
                    degree intervals and gate the select using OE. Since you would like
                    this to be "fixed", maybe potentiomenter adjustable, make it 5-10ns?
                    "gate off" during select transitioning. At lower frequencies the
                    analog sample would still be close to 90 degrees and drop off as the
                    frequency increases.....but hopefully you get away from the switch
                    overlap, at least to some extent.
                    > 73 Kees K5BCQ


                    Actually, one of the early authors suggested gating off the transition
                    time to get rid of the potential IM there. At LF shouldn't be much of
                    a logic problem, but higher????????

                    The other unmentioned so far this time is the bit about mixer
                    termination: You don't want the unwanted signals and products hitting
                    an imbalance on any of the ports and crashing back to mix again to
                    make more..... 'unless you can use them to boost the signal'. So for
                    linear mixers they use something like diplexers to provide a very
                    broadband load for all signals possible. The other way is to provide
                    an 'open' so that the signal has no current and thus no power, or a
                    short at all non-IF frequencies, thus no voltage and no power. I
                    rather suspect that the QSD with the low pass RC output filter acts as
                    the latter and again, a reason for good performance.

                    Great discussion and review for us all.

                    Since the noise level of good post-QSD amps is near practical noise
                    levels, at (at least at LF) transitions and mixers are very good, that
                    leaves us with matching the input filters inpedance to the QSD and
                    working on better quadrature LO noise, rise times, etc. It's fun to
                    look at the effects of phase and voltage imbalance on sideband
                    supression also.

                    Lots of circuit possibilities here.

                    N0UU
                  • aaintil
                    ... very ... Hi Marco et al, I decided to add some modifications of that development work. See in the Files section the simulation and schematic diagrams
                    Message 9 of 19 , Jun 3, 2007
                    • 0 Attachment
                      --- In softrock40@yahoogroups.com, Marco IK1ODO <ik1odo@...> wrote:
                      >
                      > The QSD mixer could be further improved; I'm working (currently
                      very
                      > part-time) on this topic. See
                      > http://www.spin-it.com/sdr/ik1odo_sdr1.html and
                      > http://www.spin-it.com/sdr/ik1odo_sdr2.html
                      >
                      > 73 - Marco IK1ODO / AI4YF
                      >
                      Hi Marco et al,

                      I decided to add some modifications of that development work. See in
                      the Files section the simulation and schematic diagrams related to
                      Marco's excellent work:
                      IQSDsim.zip

                      If you have the free simulation program TINA from TI, you can use
                      your creative imagination and may see very interesting results.

                      73, Ahti OH2RZ
                    • Kees & Sandy
                      Ahti, Thank you for sending this informative note. I was not aware of the TINA-TI software and it looks very usefull. The other item is those fast break before
                      Message 10 of 19 , Jun 3, 2007
                      • 0 Attachment

                        Ahti,

                        Thank you for sending this informative note. I was not aware of the TINA-TI software and it looks very usefull. The other item is those fast break before make switches. Previously I had ordered some TS5A23157DGSR and TS5A3357DCUR switches which have "sample-and-hold" listed as an application (which in my mind is basically "signal gating/switching"). The 74VC1G series actually spells out applications like "signal chopping" and "signal multiplexing for A-D or D-A conversions" (which in my mind implies a design point directed towards what we need for SDR). The specifications support that. 

                        TI actually has a "potential use" category of "SDR" now

                        73 Kees K5BCQ

                        -- "aaintil" <aaintil@...> wrote:

                        --- In softrock40@yahoogro ups.com, Marco IK1ODO <ik1odo@...> wrote:
                        >
                        > The QSD mixer could be further improved; I'm working (currently
                        very
                        > part-time) on this topic. See
                        > http://www.spin- it.com/sdr/ ik1odo_sdr1. html and
                        > http://www.spin- it.com/sdr/ ik1odo_sdr2. html
                        >
                        > 73 - Marco IK1ODO / AI4YF
                        >
                        Hi Marco et al,

                        I decided to add some modifications of that development work. See in
                        the Files section the simulation and schematic diagrams related to
                        Marco's excellent work:
                        IQSDsim.zip

                        If you have the free simulation program TINA from TI, you can use
                        your creative imagination and may see very interesting results.

                        73, Ahti OH2RZ

                      • Jim Miller
                        74VC1G series is this a TI product? is it spelled right? i couldn t find any info on it. 73 jim ab3cv ... From: Kees & Sandy To: softrock40@yahoogroups.com
                        Message 11 of 19 , Jun 4, 2007
                        • 0 Attachment
                          74VC1G series
                           
                          is this a TI product? is it spelled right? i couldn't find any info on it.
                           
                          73
                           
                          jim ab3cv
                           
                          ----- Original Message -----
                          Sent: Sunday, June 03, 2007 5:48 PM
                          Subject: Re: [softrock40] Re: Switches vs Mixers

                          Ahti,

                          The 74VC1G series actually

                        • Kees & Sandy
                          74LVC1G3157DBV ...to be exact (DBV is the package and can vary) 73 Kees K5BCQ -- Jim Miller wrote: 74VC1G series is this a TI product? is
                          Message 12 of 19 , Jun 4, 2007
                          • 0 Attachment

                            74LVC1G3157DBV ...to be exact (DBV is the package and can vary)

                            73 Kees K5BCQ

                            -- "Jim Miller" <jim@...> wrote:

                            74VC1G series
                             
                            is this a TI product? is it spelled right? i couldn't find any info on it.
                             
                            73
                             
                            jim ab3cv
                             
                            ----- Original Message -----
                            Sent: Sunday, June 03, 2007 5:48 PM
                            Subject: Re: [softrock40] Re: Switches vs Mixers

                            Ahti,

                            The 74VC1G series actually

                          • aaintil
                            ... on it. ... Jim, Try this: Ahti OH2RZ
                            Message 13 of 19 , Jun 5, 2007
                            • 0 Attachment
                              --- In softrock40@yahoogroups.com, "Jim Miller" <jim@...> wrote:
                              >
                              > 74VC1G series
                              >
                              > is this a TI product? is it spelled right? i couldn't find any info
                              on it.
                              >
                              > 73
                              >
                              > jim ab3cv

                              Jim,
                              Try this:
                              <http://focus.ti.com/docs/prod/folders/print/sn74lvc1g3157.html>

                              Ahti OH2RZ
                            • Jim Miller
                              kees thanks. i thought i was missing another new part in the alphabet soup of 74xx offerings. 73 jim ab3cv
                              Message 14 of 19 , Jun 5, 2007
                              • 0 Attachment
                                kees
                                 
                                thanks. i thought i was missing another new part in the alphabet soup of 74xx offerings.
                                 
                                73
                                 
                                jim ab3cv
                              Your message has been successfully submitted and would be delivered to recipients shortly.