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

Re: [softrock40] RXTX v6.3 improved signal to noise

Expand Messages
  • John
    Alan, Those 10 ohm resistors before the mixer are dropping precious nanovolts of wanted signal. Worsening signal to noise ratio. John G3UGY From: Alan Sent:
    Message 1 of 19 , Oct 15, 2012
    • 0 Attachment
      Alan,
       
      Those 10 ohm resistors before the mixer are dropping precious nanovolts of wanted signal. Worsening signal to noise ratio.
       
      John G3UGY
       

      From: Alan
      Sent: Sunday, October 14, 2012 7:23 PM
      Subject: Re: [softrock40] RXTX v6.3 improved signal to noise

       


      ----- Original Message -----
      Subject: [softrock40] RXTX v6.3 improved signal to noise

      John,

      I'm no expert but what is the noise you see on HDSDR?
      I suspect it is op-amp noise.
      Certainly on lower frequencies antenna noise overcomes this.
      Removing the 10Rs will increase the front end gain without altering the op-amp noise.
      Seems to me that you are measuring the wrong SNR?

      73 Alan G4ZFQ

      >I was reading one of DG8SAQ's pages http://www.mydarc.de/dg8saq/switchimpedance/index.shtml where he measures the input impedance
      >to the RX mixer.
      On the same page he mentions replacing the two 10 Ohm resistors on the mixer inputs with shorting links.

      I have just tried this with the folowing result:

      With the two 10 Ohm resistors:

      Frequency 1.977 MHz
      With a 50 Ohm resistor connected to the SoftRock antenna input HDSDR noise peaks are about -125dB
      With a signal generator set to -93 dBm connected to the SoftRock antenna input the signal peak is about -110 dB

      Without the two 10 Ohm resisors:

      Frequency 1.977 MHz
      With a 50 Ohm resistor connected to the SoftRock antenna input HDSDR noise peaks are about -125dB
      With a signal generator set to -93 dBm connected to the SoftRock antenna input the signal peak is about -104 dB

      This appears to have improved signal + noise / noise ratio by 6 dB

    • John
      Hi David, The stock band pass filters are broad. The response on the spectrum analyzer and tracking generator looks the same shape as with the
      Message 2 of 19 , Oct 15, 2012
      • 0 Attachment
        Hi David,
         
        The stock band pass filters are <extremely> broad.
        The response on the spectrum analyzer and tracking generator looks the same shape as with the resistors.
         
        John G3UGY
         

        Sent: Sunday, October 14, 2012 7:12 PM
        Subject: Re: [softrock40] RXTX v6.3 improved signal to noise

         

        Hi John,


        Don't forget that changing these resistor values will alter the filter response.

        73 David AE9RB
        http://AE9RB.com/

        On Sun, Oct 14, 2012 at 10:01 AM, John <v321uk@...> wrote:
        I was reading one of DG8SAQ's pages http://www.mydarc.de/dg8saq/switchimpedance/index.shtml where he measures the input impedance to the RX mixer.
        On the same page he mentions replacing the two 10 Ohm resistors on the mixer inputs with shorting links.

      • Alan
        ... Subject: Re: [softrock40] RXTX v6.3 improved signal to noise ... John, It does not seem that simple to me. My initial thought was After many years of
        Message 3 of 19 , Oct 15, 2012
        • 0 Attachment
          ----- Original Message -----
          Subject: Re: [softrock40] RXTX v6.3 improved signal to noise


          >Those 10 ohm resistors before the mixer are dropping precious nanovolts of wanted signal. Worsening signal to noise ratio.

          John,

          It does not seem that simple to me.
          My initial thought was "After many years of Softrocks how could this be?"
          6dB is a big difference.
          From a practical point of view with a Softrock connected to a good antenna what difference will it make?
          On a sufficiently sensitive low noise receiver SNR is established at the antenna input. Signal/atmospheric etc. noise. Provided the
          receiver adds no noise the SNR is fixed. Indeed with the lower frequencies an attenuator can be used with no effect on SNR. Tony
          uses attenuators in the Ensemble RX.

          Now, maybe on the higher frequencies a Softrock might need to be a little more sensitive. I understand this is sometimes done by
          using a low noise card and reducing op-amp gain. So why, apart from Tom, has nobody suggested removing the resistors?
          What are they doing? Is there something else that is degraded if they are removed? Are they an integral part of the mixer?

          73 Alan G4ZFQ
        • John
          Alan, If I get a chance I will put the resistors back in and measure the minimum detectable signal and compare. DG8SAQ also mentions that the impedance looking
          Message 4 of 19 , Oct 15, 2012
          • 0 Attachment
            Alan,
             
            If I get a chance I will put the resistors back in and measure the minimum detectable signal and compare.
             
            DG8SAQ also mentions that the impedance looking into the mixer is closer to 50 Ohms without the resistors than with them as measured on his vector network analyzer. 
             
            John
          • Milt Cram
            ... Hi All, Shouldn t impedance matching be done with reactive components (transformers, capacitors and inductors)? If the 10 ohm resistors are being used for
            Message 5 of 19 , Oct 15, 2012
            • 0 Attachment
              On 10/14/2012 12:01 PM, John wrote:
              I was reading one of DG8SAQ's pages http://www.mydarc.de/dg8saq/switchimpedance/index.shtml where he measures the input impedance to the RX mixer.
              On the same page he mentions replacing the two 10 Ohm resistors on the mixer inputs with shorting links.
               
              I have just tried this with the folowing result:
               
              With the two 10 Ohm resistors:
               
              Frequency 1.977 MHz
              With a 50 Ohm resistor connected to the SoftRock antenna input HDSDR noise peaks are about -125dB
              With a signal generator set to -93 dBm connected to the SoftRock antenna input the signal peak is about -110 dB 
               
               
              Without the two 10 Ohm resisors:
               
              Frequency 1.977 MHz
              With a 50 Ohm resistor connected to the SoftRock antenna input HDSDR noise peaks are about -125dB
              With a signal generator set to -93 dBm connected to the SoftRock antenna input the signal peak is about -104 dB
               
              This appears to have improved signal + noise / noise ratio by 6 dB
               
              Has anyone else tried this?
               
              John G3UGY

              Hi All,

              Shouldn't impedance matching be done with reactive components (transformers, capacitors and inductors)?  If the 10 ohm resistors are being used for impedance matching, they will both attenuate the signal, and contribute noise power.  This will reduce the S/N.  Removing them should improve S/N....?  If necessary, adjust the turns ratio of the input transformer to provide the impedance match.

              73, Milt
              W8NUE
            • John
              Agreed... John G3UGY Hi All, Shouldn t impedance matching be done with reactive components (transformers, capacitors and inductors)? If the 10 ohm resistors
              Message 6 of 19 , Oct 15, 2012
              • 0 Attachment
                Agreed... John G3UGY
                 
                 
                Hi All,

                Shouldn't impedance matching be done with reactive components (transformers, capacitors and inductors)?  If the 10 ohm resistors are being used for impedance matching, they will both attenuate the signal, and contribute noise power.  This will reduce the S/N.  Removing them should improve S/N....?  If necessary, adjust the turns ratio of the input transformer to provide the impedance match.

                73, Milt
                W8NUE
              • Milt Cram
                ... Additional comments regarding the 10 ohm resistors-- I m guessing that the 10 ohm resistors are also used to swamp the variations in the on-resistance of
                Message 7 of 19 , Oct 15, 2012
                • 0 Attachment
                  On 10/14/2012 12:01 PM, John wrote:
                  I was reading one of DG8SAQ's pages http://www.mydarc.de/dg8saq/switchimpedance/index.shtml where he measures the input impedance to the RX mixer.
                  On the same page he mentions replacing the two 10 Ohm resistors on the mixer inputs with shorting links.
                   
                  I have just tried this with the folowing result:
                   
                  With the two 10 Ohm resistors:
                   
                  Frequency 1.977 MHz
                  With a 50 Ohm resistor connected to the SoftRock antenna input HDSDR noise peaks are about -125dB
                  With a signal generator set to -93 dBm connected to the SoftRock antenna input the signal peak is about -110 dB 
                   
                   
                  Without the two 10 Ohm resisors:
                   
                  Frequency 1.977 MHz
                  With a 50 Ohm resistor connected to the SoftRock antenna input HDSDR noise peaks are about -125dB
                  With a signal generator set to -93 dBm connected to the SoftRock antenna input the signal peak is about -104 dB
                   
                  This appears to have improved signal + noise / noise ratio by 6 dB
                   
                  Has anyone else tried this?
                   
                  John G3UGY

                  Additional comments regarding the 10 ohm resistors--

                  I'm guessing that the 10 ohm resistors are also used to swamp the variations in the "on-resistance" of the switches.  At low HF frequencies, external noise probably dominates the S/N and the resistors "stabilize" the input impedance.  At higher frequencies, it would seem desirable to eliminate, as much as possible, resistors in the input circuitry.

                  73, Milt
                  W8NUE
                • Kees & Sandy
                  I agree. Any input impedance matching to the antenna will help and the VNA is a good tool to find out, but I assumed those resistors were in there also to
                  Message 8 of 19 , Oct 15, 2012
                  • 0 Attachment
                    I agree. Any input impedance matching to the antenna will help and the VNA is a good tool to find out, but I assumed those resistors were in there also to maintain balance between the two inputs. Since you can have some resistance variation between the two FET switches, windings, etc adding 10 ohms will swamp those values. Another approach might be to take a good commercial input RF transformer from someone like Mini-Circuits and use it. You can also parallel FETs to "average" the resistance variation. As Milt suggests, you can adjust the turns ratio and then reduce the 10 ohm resistance so you see 50 ohms with the VNA. Another item that I've found makes a difference is to use bifilar wire for the secondary. Regardless of how much/little you twist the wires, it's not as good as bifilar wire.
                     
                    73 Kees K5BCQ


                    ---------- Original Message ----------
                    From: Milt Cram <w8nue@...>
                    To: softrock40@yahoogroups.com
                    Subject: Re: [softrock40] RXTX v6.3 improved signal to noise
                    Date: Mon, 15 Oct 2012 09:21:55 -0500

                     

                    On 10/14/2012 12:01 PM, John wrote:
                    I was reading one of DG8SAQ's pages http://www.mydarc.de/dg8saq/switchimpedance/index.shtml where he measures the input impedance to the RX mixer.
                    On the same page he mentions replacing the two 10 Ohm resistors on the mixer inputs with shorting links.
                     
                    I have just tried this with the folowing result:
                     
                    With the two 10 Ohm resistors:
                     
                    Frequency 1.977 MHz
                    With a 50 Ohm resistor connected to the SoftRock antenna input HDSDR noise peaks are about -125dB
                    With a signal generator set to -93 dBm connected to the SoftRock antenna input the signal peak is about -110 dB 
                     
                     
                    Without the two 10 Ohm resisors:
                     
                    Frequency 1.977 MHz
                    With a 50 Ohm resistor connected to the SoftRock antenna input HDSDR noise peaks are about -125dB
                    With a signal generator set to -93 dBm connected to the SoftRock antenna input the signal peak is about -104 dB
                     
                    This appears to have improved signal + noise / noise ratio by 6 dB
                     
                    Has anyone else tried this?
                     
                    John G3UGY

                    Additional comments regarding the 10 ohm resistors--

                    I'm guessing that the 10 ohm resistors are also used to swamp the variations in the "on-resistance" of the switches.  At low HF frequencies, external noise probably dominates the S/N and the resistors "stabilize" the input impedance.  At higher frequencies, it would seem desirable to eliminate, as much as possible, resistors in the input circuitry.

                    73, Milt
                    W8NUE

                     

                     

                  • vbifyz
                    ... I agree with Milt. Shorting these resistors will improve SNR, but may degrade image rejection. There is a tradeoff, which is also frequency dependent. 73,
                    Message 9 of 19 , Oct 15, 2012
                    • 0 Attachment
                      --- In softrock40@yahoogroups.com, Milt Cram <w8nue@...> wrote:
                      >

                      > Additional comments regarding the 10 ohm resistors--
                      >
                      > I'm guessing that the 10 ohm resistors are also used to swamp the
                      > variations in the "on-resistance" of the switches. At low HF
                      > frequencies, external noise probably dominates the S/N and the resistors
                      > "stabilize" the input impedance. At higher frequencies, it would seem
                      > desirable to eliminate, as much as possible, resistors in the input
                      > circuitry.
                      >
                      > 73, Milt
                      > W8NUE
                      >

                      I agree with Milt. Shorting these resistors will improve SNR, but may degrade image rejection. There is a tradeoff, which is also frequency dependent.

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