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

Switching BPFs

Expand Messages
  • Pete Smith
    I have a real problem with my attempts to switch bandpass filters on my 8.3+Xtall, in pursuit of my goal of single knob bandswitching of the SDR hardware.
    Message 1 of 12 , Jul 29, 2008
    • 0 Attachment
      I have a real problem with my attempts to switch bandpass filters on my
      8.3+Xtall, in pursuit of my goal of single knob bandswitching of the SDR
      hardware.

      After some discussion on this list, I decided to try to use 4PDT relays to
      switch 4 of the 5 lines feeding the BPFs. This followed advice from
      someone on this list that the center-taps of the transformers could be
      paralleled, and did not need to be switched, but that the other 4 leads
      should be. I implemented this using Omron small-signal relays and simply
      wired each of the 4 BPF boards back through the relays to the BPF socket on
      top of the 8.3 board, with lead lengths of 3-4 inches average.

      Now here's the problem. I plug the 20M BPF into the 8.3 board and I have
      lots of signals. The auto I/Q phase and gain balancing works fine, though
      it shows a surprisingly big disparity between channels - 6-7 degrees phase
      error and .600-.700 gain on one channel relative to the other. Still, it
      does auto-balance, and the resulting signals are pretty
      image-free. However, when I move the BPF to the remote location, I see
      relatively fewer and weaker signals (still some), and the auto I/Q balance
      won't work in either CW Skimmer or Rocky, even when I provide it a signal
      to work with from my transceiver.

      I'm not smart enough to understand what I'm dealing with. Are lead lengths
      on this order simply too long for higher frequencies? Is the remote
      location of the BPFs throwing the phase errors still further off, to the
      point where the auto I/Q balance can't do the job? Is this a fool's errand
      I'm on, because of the lead lengths? Do I need to implement a more compact
      solution, such as pin diode switching? Or is there another switchable BPF
      kit that I could use in front of the SoftRock in order to avoid these
      problems? I think I've read that the BPF design of the Softrocks
      incorporates an impedance tranfformation and some important phasing
      functions, so I'm not sure that outboarding the BPF function entirely would
      work either.

      I'd really appreciate any advice, particularly from the softRock design
      team, or anyone who has made BPF switching work with the 8.3+Xtall. Thanks
      in advance.

      73, Pete N4ZR

      73, Pete N4ZR
    • alexle188
      Hi, Is the common transformer center taps connected to the position of the center tap of the SR socket? Alex
      Message 2 of 12 , Jul 29, 2008
      • 0 Attachment
        Hi,

        Is the common transformer center taps connected to the position of the
        center tap of the SR socket?

        Alex
      • Alan
        ... From: Pete Smith ... Pete, I think you should look for a fault. Either in your wiring or a relay not working properly. A few extra inches are not going
        Message 3 of 12 , Jul 29, 2008
        • 0 Attachment
          ----- Original Message -----
          From: "Pete Smith"

          >I have a real problem with my attempts to switch bandpass filters on my
          > 8.3+Xtall, in pursuit of my goal of single knob bandswitching of the SDR
          > hardware.
          >

          Pete,

          I think you should look for a fault. Either in your wiring or a relay not
          working properly.

          A few extra inches are not going to make much difference. Certainly no
          difference in signal strength should be noticed and I doubt there would be
          any significant change in image rejection.

          What you are describing sounds more like getting the transformer secondaries
          mixed up, as happens quite regularly.

          Keep at it, I see no reason why it should not work.

          73 Alan G4ZFQ
        • Pete Smith
          Alex, you hit the nail on the head from 10,000 miles away - somehow I had reversed the plug. I ll correct it and retest. Thanks and 73, Pete
          Message 4 of 12 , Jul 30, 2008
          • 0 Attachment
            Alex, you hit the nail on the head from 10,000 miles away - somehow I had
            reversed the plug. I'll correct it and retest.

            Thanks and 73, Pete

            At 09:41 PM 7/29/2008, you wrote:
            >Hi,
            >
            >Is the common transformer center taps connected to the position of the
            >center tap of the SR socket?
            >
            >Alex
            >
            >
            >------------------------------------
            >
            >Yahoo! Groups Links
            >
            >
            >
          • Tony Parks
            Hi Pete, With the gain balance so far off it may be there is a soldering problem at U6, the FST3253. Please look carefully at the soldering of U6 pins 3, 6,
            Message 5 of 12 , Aug 1, 2008
            • 0 Attachment
              Hi Pete,
               
              With the gain balance so far off it may be there is a soldering problem at U6, the FST3253.  Please look carefully at the soldering of U6 pins 3, 6, 11 and 12.
               
              Can you measure the sensitivity of the receiver?
               
              73,
              Tony KB9YIG
              ----- Original Message -----
              Cc: AQRP
              Sent: Friday, August 01, 2008 9:31 AM
              Subject: [softrock40] Switching BPFs

              At 09:01 AM 8/1/2008, John H. Fisher wrote:
              >We are providing three binary bits to be used by a hardware BPF
              >switching board. The three hardware outputs represent memory locations
              >100-199, 200-299, 300-399... depending on the most significant digit 1-8
              >= 000-111. The actual BPF hardware will use those three lines to select
              >a one of eight filters based on the memory location. This was an add on
              >Kees thought of to use the last three pins on the chip. Seems like a
              >good idea :-) Best of luck with your relay BPF board. I'm sorry the
              >Si570 controller can't help you out with that particular problem:-)

              What you have there is really neat, John - I am wondering if anyone has a
              hardware implementation of the BPF switching that will work with the Lite
              8.3, or if the 5-lead BPFs it uses are unique and difficult to interface
              to. My problem seems to be related to lead lengths (maybe), and
              secondarily to an apparent imbalance in both phase and gain in my Lite
              8.3. I can remotely locate the BPF on 160 and 80. By 40M I'm no longer
              able to get the I/Q balance corrected reliably - sometimes it works and
              sometimes not. Above 40, the auto balance routine won't even run. Here
              are some examples:

              All of the following tests were run using Rocky 3.5, eyeballing the signal
              levels off the spectrum display. All the image rejection tests were run
              with the fundamental at x025 KHz - 1825, 3525, etc. and a local oscillator
              frequency of x046 - 1846, 3546, etc. Levels expressed are high/low
              extremes across 96 KHz audio bandwidth

              Band BPF location Image Rej (dB) Phase error (deg) Gain ratio
              (one channel as % of other)

              160 orig. >60 -.93 to -.33 .62-.64
              remote >60 1.4 to 4.1 .61-.65

              80 orig. >60 6.2 to 7.1 .59-.61
              remote >60 4.1 to 7.2 .69-.70

              40 orig. >60 7.3 to 8.1 .70-.71
              remote no data at this and higher - auto I/Q balance
              would not work

              20 orig. >70 5.8 to 6.44 .66-.78

              15 orig. >60 3.9 to 5.2 .74-.76

              10 orig. >60 8.7 to 9.2 .77=.78

              73, Pete N4ZR

              ...who is going to put this aside for a while and see if he can get his
              scope working.

            • Christos Nikolaou
              Pete and Tony, Just popped into my mind. Sometimes so much deviation comes from sampling problems of cards. This can be rectified if you ommit one or two
              Message 6 of 12 , Aug 1, 2008
              • 0 Attachment
                Pete and Tony,

                Just popped into my mind.
                Sometimes so much deviation comes from sampling problems of cards.
                This can be rectified if you ommit one or two samples from one
                channel. Such compensation is there in Rocky but its only +/- 1 sample
                whereas it might need more.
                In the aforementioned PowerSDR version is provision for +/- 32 samples.

                Just give it a try.

                73
                Christos SV1EIA


                --- In softrock40@yahoogroups.com, "Tony Parks" <kb9yig@...> wrote:
                >
                > Hi Pete,
                >
                > With the gain balance so far off it may be there is a soldering
                problem at U6, the FST3253. Please look carefully at the soldering of
                U6 pins 3, 6, 11 and 12.
                >
                > Can you measure the sensitivity of the receiver?
                >
                > 73,
                > Tony KB9YIG
                > ----- Original Message -----
                > From: Pete Smith
                > To: softrock40@yahoogroups.com
                > Cc: AQRP
                > Sent: Friday, August 01, 2008 9:31 AM
                > Subject: [softrock40] Switching BPFs
                >
                >
                > At 09:01 AM 8/1/2008, John H. Fisher wrote:
                > >We are providing three binary bits to be used by a hardware BPF
                > >switching board. The three hardware outputs represent memory
                locations
                > >100-199, 200-299, 300-399... depending on the most significant
                digit 1-8
                > >= 000-111. The actual BPF hardware will use those three lines to
                select
                > >a one of eight filters based on the memory location. This was an
                add on
                > >Kees thought of to use the last three pins on the chip. Seems like a
                > >good idea :-) Best of luck with your relay BPF board. I'm sorry the
                > >Si570 controller can't help you out with that particular problem:-)
                >
                > What you have there is really neat, John - I am wondering if
                anyone has a
                > hardware implementation of the BPF switching that will work with
                the Lite
                > 8.3, or if the 5-lead BPFs it uses are unique and difficult to
                interface
                > to. My problem seems to be related to lead lengths (maybe), and
                > secondarily to an apparent imbalance in both phase and gain in my
                Lite
                > 8.3. I can remotely locate the BPF on 160 and 80. By 40M I'm no
                longer
                > able to get the I/Q balance corrected reliably - sometimes it
                works and
                > sometimes not. Above 40, the auto balance routine won't even run.
                Here
                > are some examples:
                >
                > All of the following tests were run using Rocky 3.5, eyeballing
                the signal
                > levels off the spectrum display. All the image rejection tests
                were run
                > with the fundamental at x025 KHz - 1825, 3525, etc. and a local
                oscillator
                > frequency of x046 - 1846, 3546, etc. Levels expressed are high/low
                > extremes across 96 KHz audio bandwidth
                >
                > Band BPF location Image Rej (dB) Phase error (deg) Gain ratio
                > (one channel as % of other)
                >
                > 160 orig. >60 -.93 to -.33 .62-.64
                > remote >60 1.4 to 4.1 .61-.65
                >
                > 80 orig. >60 6.2 to 7.1 .59-.61
                > remote >60 4.1 to 7.2 .69-.70
                >
                > 40 orig. >60 7.3 to 8.1 .70-.71
                > remote no data at this and higher - auto I/Q balance
                > would not work
                >
                > 20 orig. >70 5.8 to 6.44 .66-.78
                >
                > 15 orig. >60 3.9 to 5.2 .74-.76
                >
                > 10 orig. >60 8.7 to 9.2 .77=.78
                >
                > 73, Pete N4ZR
                >
                > ...who is going to put this aside for a while and see if he can
                get his
                > scope working.
                >
              • Edson Pereira
                Hi Christos, Does the sample compensation also work in trasmit? Regards, -- Edson, jf1afn
                Message 7 of 12 , Aug 1, 2008
                • 0 Attachment
                  Hi Christos,

                  Does the sample compensation also work in trasmit?

                  Regards,

                  -- Edson, jf1afn

                  Christos Nikolaou wrote:

                  > Pete and Tony,
                  >
                  > Just popped into my mind.
                  > Sometimes so much deviation comes from sampling problems of cards.
                  > This can be rectified if you ommit one or two samples from one
                  > channel. Such compensation is there in Rocky but its only +/- 1 sample
                  > whereas it might need more.
                  > In the aforementioned PowerSDR version is provision for +/- 32 samples.
                  >
                  > Just give it a try.
                  >
                  > 73
                  > Christos SV1EIA
                  >
                  > --- In softrock40@yahoogroups.com
                  > <mailto:softrock40%40yahoogroups.com>, "Tony Parks" <kb9yig@...> wrote:
                  > >
                  > > Hi Pete,
                  > >
                  > > With the gain balance so far off it may be there is a soldering
                  > problem at U6, the FST3253. Please look carefully at the soldering of
                  > U6 pins 3, 6, 11 and 12.
                  > >
                  > > Can you measure the sensitivity of the receiver?
                  > >
                  > > 73,
                  > > Tony KB9YIG
                  > > ----- Original Message -----
                  > > From: Pete Smith
                  > > To: softrock40@yahoogroups.com <mailto:softrock40%40yahoogroups.com>
                  > > Cc: AQRP
                  > > Sent: Friday, August 01, 2008 9:31 AM
                  > > Subject: [softrock40] Switching BPFs
                  > >
                  > >
                  > > At 09:01 AM 8/1/2008, John H. Fisher wrote:
                  > > >We are providing three binary bits to be used by a hardware BPF
                  > > >switching board. The three hardware outputs represent memory
                  > locations
                  > > >100-199, 200-299, 300-399... depending on the most significant
                  > digit 1-8
                  > > >= 000-111. The actual BPF hardware will use those three lines to
                  > select
                  > > >a one of eight filters based on the memory location. This was an
                  > add on
                  > > >Kees thought of to use the last three pins on the chip. Seems like a
                  > > >good idea :-) Best of luck with your relay BPF board. I'm sorry the
                  > > >Si570 controller can't help you out with that particular problem:-)
                  > >
                  > > What you have there is really neat, John - I am wondering if
                  > anyone has a
                  > > hardware implementation of the BPF switching that will work with
                  > the Lite
                  > > 8.3, or if the 5-lead BPFs it uses are unique and difficult to
                  > interface
                  > > to. My problem seems to be related to lead lengths (maybe), and
                  > > secondarily to an apparent imbalance in both phase and gain in my
                  > Lite
                  > > 8.3. I can remotely locate the BPF on 160 and 80. By 40M I'm no
                  > longer
                  > > able to get the I/Q balance corrected reliably - sometimes it
                  > works and
                  > > sometimes not. Above 40, the auto balance routine won't even run.
                  > Here
                  > > are some examples:
                  > >
                  > > All of the following tests were run using Rocky 3.5, eyeballing
                  > the signal
                  > > levels off the spectrum display. All the image rejection tests
                  > were run
                  > > with the fundamental at x025 KHz - 1825, 3525, etc. and a local
                  > oscillator
                  > > frequency of x046 - 1846, 3546, etc. Levels expressed are high/low
                  > > extremes across 96 KHz audio bandwidth
                  > >
                  > > Band BPF location Image Rej (dB) Phase error (deg) Gain ratio
                  > > (one channel as % of other)
                  > >
                  > > 160 orig. >60 -.93 to -.33 .62-.64
                  > > remote >60 1.4 to 4.1 .61-.65
                  > >
                  > > 80 orig. >60 6.2 to 7.1 .59-.61
                  > > remote >60 4.1 to 7.2 .69-.70
                  > >
                  > > 40 orig. >60 7.3 to 8.1 .70-.71
                  > > remote no data at this and higher - auto I/Q balance
                  > > would not work
                  > >
                  > > 20 orig. >70 5.8 to 6.44 .66-.78
                  > >
                  > > 15 orig. >60 3.9 to 5.2 .74-.76
                  > >
                  > > 10 orig. >60 8.7 to 9.2 .77=.78
                  > >
                  > > 73, Pete N4ZR
                  > >
                  > > ...who is going to put this aside for a while and see if he can
                  > get his
                  > > scope working.
                  > >
                  >
                  >
                • Christos Nikolaou
                  Edson, It does Now! Just uploaded for your remark! :-) Thanks for pointing this out. 73 Christos SV1EIA ... samples. ... like a
                  Message 8 of 12 , Aug 1, 2008
                  • 0 Attachment
                    Edson,

                    It does Now!
                    Just uploaded for your remark! :-)
                    Thanks for pointing this out.

                    73
                    Christos SV1EIA


                    --- In softrock40@yahoogroups.com, Edson Pereira <ewp_jp@...> wrote:
                    >
                    >
                    > Hi Christos,
                    >
                    > Does the sample compensation also work in trasmit?
                    >
                    > Regards,
                    >
                    > -- Edson, jf1afn
                    >
                    > Christos Nikolaou wrote:
                    >
                    > > Pete and Tony,
                    > >
                    > > Just popped into my mind.
                    > > Sometimes so much deviation comes from sampling problems of cards.
                    > > This can be rectified if you ommit one or two samples from one
                    > > channel. Such compensation is there in Rocky but its only +/- 1 sample
                    > > whereas it might need more.
                    > > In the aforementioned PowerSDR version is provision for +/- 32
                    samples.
                    > >
                    > > Just give it a try.
                    > >
                    > > 73
                    > > Christos SV1EIA
                    > >
                    > > --- In softrock40@yahoogroups.com
                    > > <mailto:softrock40%40yahoogroups.com>, "Tony Parks" <kb9yig@> wrote:
                    > > >
                    > > > Hi Pete,
                    > > >
                    > > > With the gain balance so far off it may be there is a soldering
                    > > problem at U6, the FST3253. Please look carefully at the soldering of
                    > > U6 pins 3, 6, 11 and 12.
                    > > >
                    > > > Can you measure the sensitivity of the receiver?
                    > > >
                    > > > 73,
                    > > > Tony KB9YIG
                    > > > ----- Original Message -----
                    > > > From: Pete Smith
                    > > > To: softrock40@yahoogroups.com <mailto:softrock40%40yahoogroups.com>
                    > > > Cc: AQRP
                    > > > Sent: Friday, August 01, 2008 9:31 AM
                    > > > Subject: [softrock40] Switching BPFs
                    > > >
                    > > >
                    > > > At 09:01 AM 8/1/2008, John H. Fisher wrote:
                    > > > >We are providing three binary bits to be used by a hardware BPF
                    > > > >switching board. The three hardware outputs represent memory
                    > > locations
                    > > > >100-199, 200-299, 300-399... depending on the most significant
                    > > digit 1-8
                    > > > >= 000-111. The actual BPF hardware will use those three lines to
                    > > select
                    > > > >a one of eight filters based on the memory location. This was an
                    > > add on
                    > > > >Kees thought of to use the last three pins on the chip. Seems
                    like a
                    > > > >good idea :-) Best of luck with your relay BPF board. I'm sorry the
                    > > > >Si570 controller can't help you out with that particular problem:-)
                    > > >
                    > > > What you have there is really neat, John - I am wondering if
                    > > anyone has a
                    > > > hardware implementation of the BPF switching that will work with
                    > > the Lite
                    > > > 8.3, or if the 5-lead BPFs it uses are unique and difficult to
                    > > interface
                    > > > to. My problem seems to be related to lead lengths (maybe), and
                    > > > secondarily to an apparent imbalance in both phase and gain in my
                    > > Lite
                    > > > 8.3. I can remotely locate the BPF on 160 and 80. By 40M I'm no
                    > > longer
                    > > > able to get the I/Q balance corrected reliably - sometimes it
                    > > works and
                    > > > sometimes not. Above 40, the auto balance routine won't even run.
                    > > Here
                    > > > are some examples:
                    > > >
                    > > > All of the following tests were run using Rocky 3.5, eyeballing
                    > > the signal
                    > > > levels off the spectrum display. All the image rejection tests
                    > > were run
                    > > > with the fundamental at x025 KHz - 1825, 3525, etc. and a local
                    > > oscillator
                    > > > frequency of x046 - 1846, 3546, etc. Levels expressed are high/low
                    > > > extremes across 96 KHz audio bandwidth
                    > > >
                    > > > Band BPF location Image Rej (dB) Phase error (deg) Gain ratio
                    > > > (one channel as % of other)
                    > > >
                    > > > 160 orig. >60 -.93 to -.33 .62-.64
                    > > > remote >60 1.4 to 4.1 .61-.65
                    > > >
                    > > > 80 orig. >60 6.2 to 7.1 .59-.61
                    > > > remote >60 4.1 to 7.2 .69-.70
                    > > >
                    > > > 40 orig. >60 7.3 to 8.1 .70-.71
                    > > > remote no data at this and higher - auto I/Q balance
                    > > > would not work
                    > > >
                    > > > 20 orig. >70 5.8 to 6.44 .66-.78
                    > > >
                    > > > 15 orig. >60 3.9 to 5.2 .74-.76
                    > > >
                    > > > 10 orig. >60 8.7 to 9.2 .77=.78
                    > > >
                    > > > 73, Pete N4ZR
                    > > >
                    > > > ...who is going to put this aside for a while and see if he can
                    > > get his
                    > > > scope working.
                    > > >
                    > >
                    > >
                    >
                  • Pete Smith
                    Tony, it s a good thing you re sworn to use your super-powers only for good - hi. I did in fact have a cold solder joint on pin 11 of the FST3253. After
                    Message 9 of 12 , Aug 1, 2008
                    • 0 Attachment
                      Tony, it's a good thing you're sworn to use your super-powers only for good
                      - hi. I did in fact have a cold solder joint on pin 11 of the
                      FST3253. After repairing it, the gain disparity between the two channels
                      is only 1.145-1.733. I assume that's within normal limits for this circuit
                      (?).

                      Thanks again and 73, Pete N4ZR

                      At 10:11 AM 8/1/2008, Tony Parks wrote:
                      >Hi Pete,
                      >
                      >With the gain balance so far off it may be there is a soldering problem at
                      >U6, the FST3253. Please look carefully at the soldering of U6 pins 3, 6,
                      >11 and 12.
                      >
                      >Can you measure the sensitivity of the receiver?
                      >
                      >73,
                      >Tony KB9YIG
                      >----- Original Message -----
                      >From: <mailto:n4zr@...>Pete Smith
                      >To: <mailto:softrock40@yahoogroups.com>softrock40@yahoogroups.com
                      >Cc: <mailto:AQRP@yahoogroups.com>AQRP
                      >Sent: Friday, August 01, 2008 9:31 AM
                      >Subject: [softrock40] Switching BPFs
                      >
                      >At 09:01 AM 8/1/2008, John H. Fisher wrote:
                      > >We are providing three binary bits to be used by a hardware BPF
                      > >switching board. The three hardware outputs represent memory locations
                      > >100-199, 200-299, 300-399... depending on the most significant digit 1-8
                      > >= 000-111. The actual BPF hardware will use those three lines to select
                      > >a one of eight filters based on the memory location. This was an add on
                      > >Kees thought of to use the last three pins on the chip. Seems like a
                      > >good idea :-) Best of luck with your relay BPF board. I'm sorry the
                      > >Si570 controller can't help you out with that particular problem:-)
                      >
                      >What you have there is really neat, John - I am wondering if anyone has a
                      >hardware implementation of the BPF switching that will work with the Lite
                      >8.3, or if the 5-lead BPFs it uses are unique and difficult to interface
                      >to. My problem seems to be related to lead lengths (maybe), and
                      >secondarily to an apparent imbalance in both phase and gain in my Lite
                      >8.3. I can remotely locate the BPF on 160 and 80. By 40M I'm no longer
                      >able to get the I/Q balance corrected reliably - sometimes it works and
                      >sometimes not. Above 40, the auto balance routine won't even run. Here
                      >are some examples:
                      >
                      >All of the following tests were run using Rocky 3.5, eyeballing the signal
                      >levels off the spectrum display. All the image rejection tests were run
                      >with the fundamental at x025 KHz - 1825, 3525, etc. and a local oscillator
                      >frequency of x046 - 1846, 3546, etc. Levels expressed are high/low
                      >extremes across 96 KHz audio bandwidth
                      >
                      >Band BPF location Image Rej (dB) Phase error (deg) Gain ratio
                      >(one channel as % of other)
                      >
                      >160 orig. >60 -.93 to -.33 .62-.64
                      >remote >60 1.4 to 4.1 .61-.65
                      >
                      >80 orig. >60 6.2 to 7.1 .59-.61
                      >remote >60 4.1 to 7.2 .69-.70
                      >
                      >40 orig. >60 7.3 to 8.1 .70-.71
                      >remote no data at this and higher - auto I/Q balance
                      >would not work
                      >
                      >20 orig. >70 5.8 to 6.44 .66-.78
                      >
                      >15 orig. >60 3.9 to 5.2 .74-.76
                      >
                      >10 orig. >60 8.7 to 9.2 .77=.78
                      >
                      >73, Pete N4ZR
                      >
                      >...who is going to put this aside for a while and see if he can get his
                      >scope working.
                      >
                      >
                    • Pete Smith
                      I should have added, though, that the phase disparity is still 4.7 to 5.4 degrees. The auto I/Q balancing works fine with the BPF plugged right into the RX
                      Message 10 of 12 , Aug 1, 2008
                      • 0 Attachment
                        I should have added, though, that the phase disparity is still 4.7 to 5.4
                        degrees. The auto I/Q balancing works fine with the BPF plugged right into
                        the RX board, but will not compensate when the BPF is a few inches away (on
                        20M).

                        Any thoughts on why so large? I suppose it could be this sound card
                        (M-Audio Revolution 5.1). I'll try another just to be sure.


                        73, Pete N4ZR

                        At 05:38 PM 8/1/2008, Pete Smith wrote:
                        >Tony, it's a good thing you're sworn to use your super-powers only for good
                        >- hi. I did in fact have a cold solder joint on pin 11 of the
                        >FST3253. After repairing it, the gain disparity between the two channels
                        >is only 1.145-1.733. I assume that's within normal limits for this circuit
                        >(?).
                        >
                        >Thanks again and 73, Pete N4ZR
                        >
                        >At 10:11 AM 8/1/2008, Tony Parks wrote:
                        > >Hi Pete,
                        > >
                        > >With the gain balance so far off it may be there is a soldering problem at
                        > >U6, the FST3253. Please look carefully at the soldering of U6 pins 3, 6,
                        > >11 and 12.
                        > >
                        > >Can you measure the sensitivity of the receiver?
                        > >
                        > >73,
                        > >Tony KB9YIG
                        > >----- Original Message -----
                        > >From: <mailto:n4zr@...>Pete Smith
                        > >To: <mailto:softrock40@yahoogroups.com>softrock40@yahoogroups.com
                        > >Cc: <mailto:AQRP@yahoogroups.com>AQRP
                        > >Sent: Friday, August 01, 2008 9:31 AM
                        > >Subject: [softrock40] Switching BPFs
                        > >
                        > >At 09:01 AM 8/1/2008, John H. Fisher wrote:
                        > > >We are providing three binary bits to be used by a hardware BPF
                        > > >switching board. The three hardware outputs represent memory locations
                        > > >100-199, 200-299, 300-399... depending on the most significant digit 1-8
                        > > >= 000-111. The actual BPF hardware will use those three lines to select
                        > > >a one of eight filters based on the memory location. This was an add on
                        > > >Kees thought of to use the last three pins on the chip. Seems like a
                        > > >good idea :-) Best of luck with your relay BPF board. I'm sorry the
                        > > >Si570 controller can't help you out with that particular problem:-)
                        > >
                        > >What you have there is really neat, John - I am wondering if anyone has a
                        > >hardware implementation of the BPF switching that will work with the Lite
                        > >8.3, or if the 5-lead BPFs it uses are unique and difficult to interface
                        > >to. My problem seems to be related to lead lengths (maybe), and
                        > >secondarily to an apparent imbalance in both phase and gain in my Lite
                        > >8.3. I can remotely locate the BPF on 160 and 80. By 40M I'm no longer
                        > >able to get the I/Q balance corrected reliably - sometimes it works and
                        > >sometimes not. Above 40, the auto balance routine won't even run. Here
                        > >are some examples:
                        > >
                        > >All of the following tests were run using Rocky 3.5, eyeballing the signal
                        > >levels off the spectrum display. All the image rejection tests were run
                        > >with the fundamental at x025 KHz - 1825, 3525, etc. and a local oscillator
                        > >frequency of x046 - 1846, 3546, etc. Levels expressed are high/low
                        > >extremes across 96 KHz audio bandwidth
                        > >
                        > >Band BPF location Image Rej (dB) Phase error (deg) Gain ratio
                        > >(one channel as % of other)
                        > >
                        > >160 orig. >60 -.93 to -.33 .62-.64
                        > >remote >60 1.4 to 4.1 .61-.65
                        > >
                        > >80 orig. >60 6.2 to 7.1 .59-.61
                        > >remote >60 4.1 to 7.2 .69-.70
                        > >
                        > >40 orig. >60 7.3 to 8.1 .70-.71
                        > >remote no data at this and higher - auto I/Q balance
                        > >would not work
                        > >
                        > >20 orig. >70 5.8 to 6.44 .66-.78
                        > >
                        > >15 orig. >60 3.9 to 5.2 .74-.76
                        > >
                        > >10 orig. >60 8.7 to 9.2 .77=.78
                        > >
                        > >73, Pete N4ZR
                        > >
                        > >...who is going to put this aside for a while and see if he can get his
                        > >scope working.
                        > >
                        > >
                        >
                        >
                        >------------------------------------
                        >
                        >Yahoo! Groups Links
                        >
                        >
                        >
                      • drmail377
                        Hi Pete, So you re using three sets of contacts per filter? Upload a schematic to the Files Temp area so we can take a look, hand-written is fine. Are you
                        Message 11 of 12 , Aug 2, 2008
                        • 0 Attachment
                          Hi Pete,

                          So you're using three sets of contacts per filter? Upload a schematic
                          to the Files > Temp area so we can take a look, hand-written is fine.

                          Are you using coax to interconnect? If so, what type? Make sure all
                          the wire and/or coax lengths are the same. 5 inches at 20 meters is
                          around 2.3 degrees in free space and 1.5 degrees in typical coax.

                          When you describe your phase error, do you mean it is 4.7 deg. with
                          one filter and 5.4 with a different one? Or is the total phase
                          disparity between 4.7 and 5.4 degrees, at different frequencies
                          perhaps. Or is it between a switched BPF and one plugged directly into
                          the main board? Please clarify.

                          Remember with coax, if it is not properly terminated it will introduce
                          reactance in likely significant amount even in inches of length (can't
                          remember off-hand how to calculate this). This causes some concern
                          because if you we 50 Ohm coax on the switched BPF transformer
                          secondary the termination is at-best around 25 Ohms, not 50, or maybe
                          not because of the commutating switch, hmmm...

                          Again, at least keep all cabling lengths the same at all points. I'm
                          wondering about the specifications of your relays. DIP relays are in
                          almost all cases not so good at RF. The fact that you're using OMRON
                          is a good sign, they're the only mfgr I've been able to find in a
                          quick search that has some decent RF-capable DIP relays with decent
                          isolation (around 60 dB if memory serves). What kind of relays are you
                          using?

                          I don't think the switch and switch wires to the relay coils should
                          matter, but you might want to try bypassing them anyway.

                          I presume the software you're using enables you to reset and restart
                          the phase error and phase tracking routine at-will. Don't tell the
                          software anything about frequency change when you change the SDR's
                          frequency and BPF. Then try the opposite, do tell it about the
                          frequency change. I'm concerned the software may be doing some sort of
                          scaling with frequency change. Dunno - reaching for straws.

                          Good Luck & 73's, David


                          --- In softrock40@yahoogroups.com, Pete Smith <n4zr@...> wrote:
                          >
                          > I should have added, though, that the phase disparity is still 4.7
                          to 5.4
                          > degrees. The auto I/Q balancing works fine with the BPF plugged
                          right into
                          > the RX board, but will not compensate when the BPF is a few inches
                          away (on
                          > 20M).
                          >
                          > Any thoughts on why so large? I suppose it could be this sound card
                          > (M-Audio Revolution 5.1). I'll try another just to be sure.
                          >
                          >
                          > 73, Pete N4ZR
                          >
                          > At 05:38 PM 8/1/2008, Pete Smith wrote:
                          > >Tony, it's a good thing you're sworn to use your super-powers only
                          for good
                          > >- hi. I did in fact have a cold solder joint on pin 11 of the
                          > >FST3253. After repairing it, the gain disparity between the two
                          channels
                          > >is only 1.145-1.733. I assume that's within normal limits for this
                          circuit
                          > >(?).
                          > >
                          > >Thanks again and 73, Pete N4ZR
                          > >
                          > >At 10:11 AM 8/1/2008, Tony Parks wrote:
                          > > >Hi Pete,
                          > > >
                          > > >With the gain balance so far off it may be there is a soldering
                          problem at
                          > > >U6, the FST3253. Please look carefully at the soldering of U6
                          pins 3, 6,
                          > > >11 and 12.
                          > > >
                          > > >Can you measure the sensitivity of the receiver?
                          > > >
                          > > >73,
                          > > >Tony KB9YIG
                          > > >----- Original Message -----
                          > > >From: <mailto:n4zr@...>Pete Smith
                          > > >To: <mailto:softrock40@yahoogroups.com>softrock40@yahoogroups.com
                          > > >Cc: <mailto:AQRP@yahoogroups.com>AQRP
                          > > >Sent: Friday, August 01, 2008 9:31 AM
                          > > >Subject: [softrock40] Switching BPFs
                          > > >
                          > > >At 09:01 AM 8/1/2008, John H. Fisher wrote:
                          > > > >We are providing three binary bits to be used by a hardware BPF
                          > > > >switching board. The three hardware outputs represent memory
                          locations
                          > > > >100-199, 200-299, 300-399... depending on the most significant
                          digit 1-8
                          > > > >= 000-111. The actual BPF hardware will use those three lines
                          to select
                          > > > >a one of eight filters based on the memory location. This was
                          an add on
                          > > > >Kees thought of to use the last three pins on the chip. Seems
                          like a
                          > > > >good idea :-) Best of luck with your relay BPF board. I'm sorry the
                          > > > >Si570 controller can't help you out with that particular problem:-)
                          > > >
                          > > >What you have there is really neat, John - I am wondering if
                          anyone has a
                          > > >hardware implementation of the BPF switching that will work with
                          the Lite
                          > > >8.3, or if the 5-lead BPFs it uses are unique and difficult to
                          interface
                          > > >to. My problem seems to be related to lead lengths (maybe), and
                          > > >secondarily to an apparent imbalance in both phase and gain in my
                          Lite
                          > > >8.3. I can remotely locate the BPF on 160 and 80. By 40M I'm no
                          longer
                          > > >able to get the I/Q balance corrected reliably - sometimes it
                          works and
                          > > >sometimes not. Above 40, the auto balance routine won't even run.
                          Here
                          > > >are some examples:
                          > > >
                          > > >All of the following tests were run using Rocky 3.5, eyeballing
                          the signal
                          > > >levels off the spectrum display. All the image rejection tests
                          were run
                          > > >with the fundamental at x025 KHz - 1825, 3525, etc. and a local
                          oscillator
                          > > >frequency of x046 - 1846, 3546, etc. Levels expressed are high/low
                          > > >extremes across 96 KHz audio bandwidth
                          > > >
                          > > >Band BPF location Image Rej (dB) Phase error (deg) Gain ratio
                          > > >(one channel as % of other)
                          > > >
                          > > >160 orig. >60 -.93 to -.33 .62-.64
                          > > >remote >60 1.4 to 4.1 .61-.65
                          > > >
                          > > >80 orig. >60 6.2 to 7.1 .59-.61
                          > > >remote >60 4.1 to 7.2 .69-.70
                          > > >
                          > > >40 orig. >60 7.3 to 8.1 .70-.71
                          > > >remote no data at this and higher - auto I/Q balance
                          > > >would not work
                          > > >
                          > > >20 orig. >70 5.8 to 6.44 .66-.78
                          > > >
                          > > >15 orig. >60 3.9 to 5.2 .74-.76
                          > > >
                          > > >10 orig. >60 8.7 to 9.2 .77=.78
                          > > >
                          > > >73, Pete N4ZR
                          > > >
                          > > >...who is going to put this aside for a while and see if he can
                          get his
                          > > >scope working.
                          > > >
                          > > >
                          > >
                          > >
                          > >------------------------------------
                          > >
                          > >Yahoo! Groups Links
                          > >
                          > >
                          > >
                          >
                        • Pete Smith
                          ... Schematic is now there - a little rough but readable. It s the only file currently in the folder. I am using 4PDT relays to do the switching - switching
                          Message 12 of 12 , Aug 2, 2008
                          • 0 Attachment
                            At 04:36 AM 8/2/2008, drmail377 wrote:
                            >Hi Pete,
                            >
                            >So you're using three sets of contacts per filter? Upload a schematic
                            >to the Files > Temp area so we can take a look, hand-written is fine.

                            Schematic is now there - a little rough but readable. It's the only file
                            currently in the folder. I am using 4PDT relays to do the switching -
                            switching pins 2 and 3 of P101 (the transformer output) and each pin of
                            P100 (antenna input and return). I have no doubt that PIN diodes would
                            make for a more compact switching scheme, but frankly didn't know how to
                            implement it - relays, I more or less understand.


                            >Are you using coax to interconnect? If so, what type? Make sure all
                            >the wire and/or coax lengths are the same. 5 inches at 20 meters is
                            >around 2.3 degrees in free space and 1.5 degrees in typical coax.


                            No, using #24 hookup wire. I did not pay any particular attention to
                            keeping lead lengths equal, and that may well be the source of my
                            problems. Tony has suggested testing with one BPF using RG-174 for the
                            outputs (pins 3 and 4 of P101, with the shields connected together to pin
                            1) and feeding the antenna straight to P100 on the BPF, to see if it is any
                            better behaved. I plan to do that next.


                            >When you describe your phase error, do you mean it is 4.7 deg. with
                            >one filter and 5.4 with a different one? Or is the total phase
                            >disparity between 4.7 and 5.4 degrees, at different frequencies
                            >perhaps. Or is it between a switched BPF and one plugged directly into
                            >the main board? Please clarify.

                            The numbers (4.7 to 5.4) define the range of values along the curve
                            produced by the phase display of the I/Q balance routine. In the message
                            you quoted, I recorded the ranges both for mounting on the RX and remote
                            mounting, on the bands where I was able to get the auto I/Q balancing to
                            work on the remote mounting. On 40 and above I could not.


                            >Remember with coax, if it is not properly terminated it will introduce
                            >reactance in likely significant amount even in inches of length (can't
                            >remember off-hand how to calculate this). This causes some concern
                            >because if you we 50 Ohm coax on the switched BPF transformer
                            >secondary the termination is at-best around 25 Ohms, not 50, or maybe
                            >not because of the commutating switch, hmmm...

                            Don't know, but if this is a problem it probably isn't practical to do
                            anything else.

                            >Again, at least keep all cabling lengths the same at all points. I'm
                            >wondering about the specifications of your relays. DIP relays are in
                            >almost all cases not so good at RF. The fact that you're using OMRON
                            >is a good sign, they're the only mfgr I've been able to find in a
                            >quick search that has some decent RF-capable DIP relays with decent
                            >isolation (around 60 dB if memory serves). What kind of relays are you
                            >using?


                            The relays I used are Omron, specifically designed for small signal use,
                            part no. G6A-474P-ST40-US. They are in 14-pin DIP packages.


                            >I don't think the switch and switch wires to the relay coils should
                            >matter, but you might want to try bypassing them anyway.
                            >
                            >I presume the software you're using enables you to reset and restart
                            >the phase error and phase tracking routine at-will. Don't tell the
                            >software anything about frequency change when you change the SDR's
                            >frequency and BPF. Then try the opposite, do tell it about the
                            >frequency change. I'm concerned the software may be doing some sort of
                            >scaling with frequency change. Dunno - reaching for straws.

                            If I tell the software what band, it goes back and finds the correction
                            curves for that band, which were stored the last time the band was
                            changed. When you start collecting phase and gain data again, it seemingly
                            starts at zero. I know that if the level difference between the
                            fundamental and the image signal is too close (say less than 10 dB) then
                            the auto balance routine cannot determine how to correct, because it
                            doesn't know which signal is which. but that's all I know about how it
                            does its magic.

                            73, Pete N4ZR
                          Your message has been successfully submitted and would be delivered to recipients shortly.