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

To match or not to match... BS170s in your RXTX

Expand Messages
  • warrenallgyer
    I have spent the better part of a week s worth of free time looking at the benefits to matching the RXTX BS170s as suggested often here. I have a drawer full
    Message 1 of 19 , Nov 28, 2012
    • 0 Attachment
      I have spent the better part of a week's worth of free time looking at the benefits to matching the RXTX BS170s as suggested often here. I have a drawer full of more than 100 BS170s from various manufacturers and I have gone through maybe half of them in various combinations in the Q5, Q7, and Q8 positions. This was on 10 meters. My criteria to be measured were power output, intermodulation level with a two tone test, and second harmonic level.

      I have to say that I have found no significant benefit from matching according to these criteria. Q7 and Q8 seem to perform just about the same, matched or unmatched, high match numbers or low.

      Q5 is another story however when it comes to its' affect on total power output. Regardless the combination used in the finals, the device chosen for Q5 dramatically affects the power output from the radio. With the same set of finals, same PS, and same drive levels, I experienced output power variances between 200 mw and 1.2 watts on 10 meters.

      IMD and harmonic content are much more affected by the operating power level than by the choice of devices in the final. The previously mentioned guidelines of 1 watt for CW and digital modes and 500 mw for SSB are very good for 12V operation. You can be a little more aggressive at 13.8 with 800 mw to even a full watt meeting the IMD requirements.

      So, my recommendation as to how to use those extra BS170s that Tony generously includes is to test them at Q5 and choose the one that gives you the power you are looking for.

      I am happy to make my test data available to anyone who is interested. Please contact me off the board.

      Warren Allgyer - W8TOD
    • John Greusel
      Warren, My question is: did you notice that one brand of Mosfet made a better Q5 component for power output than another? John KC9OJV  
      Message 2 of 19 , Nov 28, 2012
      • 0 Attachment
        Warren,
        My question is: did you notice that one brand of Mosfet made a better Q5 component for power output than another?

        John
        KC9OJV
         



        From: warrenallgyer <allgyer@...>
        To: softrock40@yahoogroups.com
        Sent: Wednesday, November 28, 2012 7:22 PM
        Subject: [softrock40] To match or not to match... BS170s in your RXTX

         
        I have spent the better part of a week's worth of free time looking at the benefits to matching the RXTX BS170s as suggested often here. I have a drawer full of more than 100 BS170s from various manufacturers and I have gone through maybe half of them in various combinations in the Q5, Q7, and Q8 positions. This was on 10 meters. My criteria to be measured were power output, intermodulation level with a two tone test, and second harmonic level.

        I have to say that I have found no significant benefit from matching according to these criteria. Q7 and Q8 seem to perform just about the same, matched or unmatched, high match numbers or low.

        Q5 is another story however when it comes to its' affect on total power output. Regardless the combination used in the finals, the device chosen for Q5 dramatically affects the power output from the radio. With the same set of finals, same PS, and same drive levels, I experienced output power variances between 200 mw and 1.2 watts on 10 meters.

        IMD and harmonic content are much more affected by the operating power level than by the choice of devices in the final. The previously mentioned guidelines of 1 watt for CW and digital modes and 500 mw for SSB are very good for 12V operation. You can be a little more aggressive at 13.8 with 800 mw to even a full watt meeting the IMD requirements.

        So, my recommendation as to how to use those extra BS170s that Tony generously includes is to test them at Q5 and choose the one that gives you the power you are looking for.

        I am happy to make my test data available to anyone who is interested. Please contact me off the board.

        Warren Allgyer - W8TOD



      • warrenallgyer
        I did not detect a pattern in this John. I think I ended up with a different brand in each of the three positions. I finally settled on a configuration that
        Message 3 of 19 , Nov 28, 2012
        • 0 Attachment
          I did not detect a pattern in this John. I think I ended up with a different brand in each of the three positions. I finally settled on a configuration that had a maximum output on 10 meters of 1.2 watts with 3rd order products at -22 dB and the second harmonic down at -50 dB. I needed to back it off to 800 mw to get the 3rd order products down to -30 (reference the single tone level). The total current draw PTT engaged, no drive, was a whopping 200 ma and 330 ma at full drive, full power.


          --- In softrock40@yahoogroups.com, John Greusel <greusel@...> wrote:
          >
          > Warren,
          > My question is: did you notice that one brand of Mosfet made a better Q5 component for power output than another?
          >
          > John
          > KC9OJV
          >  
          >
          >
          >
          > ________________________________
          > From: warrenallgyer <allgyer@...>
          > To: softrock40@yahoogroups.com
          > Sent: Wednesday, November 28, 2012 7:22 PM
          > Subject: [softrock40] To match or not to match... BS170s in your RXTX
          >
          >
          >  
          > I have spent the better part of a week's worth of free time looking at the benefits to matching the RXTX BS170s as suggested often here. I have a drawer full of more than 100 BS170s from various manufacturers and I have gone through maybe half of them in various combinations in the Q5, Q7, and Q8 positions. This was on 10 meters. My criteria to be measured were power output, intermodulation level with a two tone test, and second harmonic level.
          >
          > I have to say that I have found no significant benefit from matching according to these criteria. Q7 and Q8 seem to perform just about the same, matched or unmatched, high match numbers or low.
          >
          > Q5 is another story however when it comes to its' affect on total power output. Regardless the combination used in the finals, the device chosen for Q5 dramatically affects the power output from the radio. With the same set of finals, same PS, and same drive levels, I experienced output power variances between 200 mw and 1.2 watts on 10 meters.
          >
          > IMD and harmonic content are much more affected by the operating power level than by the choice of devices in the final. The previously mentioned guidelines of 1 watt for CW and digital modes and 500 mw for SSB are very good for 12V operation. You can be a little more aggressive at 13.8 with 800 mw to even a full watt meeting the IMD requirements.
          >
          > So, my recommendation as to how to use those extra BS170s that Tony generously includes is to test them at Q5 and choose the one that gives you the power you are looking for.
          >
          > I am happy to make my test data available to anyone who is interested. Please contact me off the board.
          >
          > Warren Allgyer - W8TOD
          >
        • Taka Sugi
          Hi Warren I don t understand why Q5 dramatically affects output power. Because, Q5/Q4 circuit works as constant voltage circuit to feed bias voltage to PA:
          Message 4 of 19 , Nov 28, 2012
          • 0 Attachment
            Hi Warren

            I don't understand why Q5 dramatically affects output power.
            Because, Q5/Q4 circuit works as constant voltage circuit to feed bias voltage to PA: Q7,Q8.
            Q5 and Q7/Q8 are thermally connected, if Q5 is different maker/lot from Q7/Q8, then output power-temperature dependence will be changed.
          • Alan
            ... Subject: [softrock40] To match or not to match... BS170s in your RXTX ... Warren, Previous checks of BS170 matching have established that Q5 affects the
            Message 5 of 19 , Nov 29, 2012
            • 0 Attachment
              ----- Original Message -----
              Subject: [softrock40] To match or not to match... BS170s in your RXTX


              > Q5 is another story however when it comes to its' affect on total power output. Regardless the combination used in the finals, the
              > device chosen for Q5 dramatically affects the power output from the radio. With the same set of finals, same PS, and same drive
              > levels, I experienced output power variances between 200 mw and 1.2 watts on 10 meters.

              Warren,

              Previous checks of BS170 matching have established that Q5 affects the bias, as measured here
              http://www.wb5rvz.com/sdr/ensemble/10_pa.htm across R48 & 49. And my summary of the discussion.
              https://sites.google.com/site/g4zfqradio/matching-bs170-for-softrock
              Did you check this each time you changed Q5? As Q5 is a key component in the bias network it would be interesting to know if your
              selected Q5 was actually one that best matched the finals.
              Or whether some other way of matching could be devised? Not everyone has a large number of BS170's or the time to check them as you
              did.


              >
              > IMD and harmonic content are much more affected by the operating power level than by the choice of devices in the final. The
              > previously mentioned guidelines of 1 watt for CW and digital modes and 500 mw for SSB are very good for 12V operation.

              Since I joined the group this is what experienced members have suggested. Yes, 13V8 is preferable, but often this is what is meant
              by 12V.
              I suggest this guideline should be followed by any user who cannot check their distortion products.
              In any case an output stage should not be pushed to it's limit, keep a little in hand.
              A relatively small increase in power will make no noticeable difference in received signal but often distortion products increase
              dramatically once the limit is reached.

              73 Alan G4ZFQ
            • warrenallgyer
              Hi Alan No I didn t check the static voltage across R48 and 49 although that would have been of interest. I was solely interested in evaluating the three
              Message 6 of 19 , Nov 29, 2012
              • 0 Attachment
                Hi Alan

                No I didn't check the static voltage across R48 and 49 although that would have been of interest. I was solely interested in evaluating the three factors I listed, power, distortion, and harmonic content.

                One thing I did note was that the different Q5 devices resulted in dramatically different initial bias voltage levels at the diode (sorry, in an airport now without schematics in hand) and the higher bias levels correlated very well with better power performance.

                My recommendation for choosing Q5 would be to complete everything in the final stage except Q5, Q7, and Q8. Then I would connect a voltmeter to the diode in the bias circuit. I think it does not matter too much which side.... and temporarily hold each of the supplied BS170s, properly oriented, to make contact with their respective holes in the Q5 position. I have not tried this but I think it should work. I would choose the 170 that results in the highest bias voltage at the diode. Then I would choose the finals pretty much at random from the remainder. It certainly would do no harm to match them according to your paper.

                I first began by tack soldering the FETs in place and then connected a prototyping board with very short leads so I could plug each of the three in separately.

                The really solid way would be to build the bias circuit externally. There is no need for Q7/8 in the test because they draw no gate current. Whichever device produces the highest bias voltage would win in my contest.

                There very well may be consequences to this that I have not encountered yet. I know Taka has sent me a response and I will reply there with my operational analysis of the bias circuit. However I did note the first line of his post and I have to disagree. I do not think the purpose of Q5 is to supply fixed bias. Q5 is thermally coupled to Q7/8 in such a way that increased drain currents due to heat at any or all of the devices result in lower bias to the finals and in reducing the collector voltage on the driver. This is a very dynamic circuit that reacts to heat and protects the finals by reducing bias and drive. It is effectively a negative feedback loop around the finals. I have taken particular care to make sure the three devices are well coupled thermally with a thin layer of grease between the devices and the board, another thin layer on the flat faces, and yet another on the bottom of the heat sink. I think it is really important that these devices share their heat mutually in order to stabilize the circuit.

                As usual, an answer that is too long by half..... sorry. Now running for an airplane. Back on line in 14 hours or so from the other side of the planet.

                Warren Allgyer - W8TOD


                --- In softrock40@yahoogroups.com, "Alan" <alan4alan@...> wrote:
                >
                >
                > ----- Original Message -----
                > Subject: [softrock40] To match or not to match... BS170s in your RXTX
                >
                >
                > > Q5 is another story however when it comes to its' affect on total power output. Regardless the combination used in the finals, the
                > > device chosen for Q5 dramatically affects the power output from the radio. With the same set of finals, same PS, and same drive
                > > levels, I experienced output power variances between 200 mw and 1.2 watts on 10 meters.
                >
                > Warren,
                >
                > Previous checks of BS170 matching have established that Q5 affects the bias, as measured here
                > http://www.wb5rvz.com/sdr/ensemble/10_pa.htm across R48 & 49. And my summary of the discussion.
                > https://sites.google.com/site/g4zfqradio/matching-bs170-for-softrock
                > Did you check this each time you changed Q5? As Q5 is a key component in the bias network it would be interesting to know if your
                > selected Q5 was actually one that best matched the finals.
                > Or whether some other way of matching could be devised? Not everyone has a large number of BS170's or the time to check them as you
                > did.
                >
                >
              • warrenallgyer
                Hi Taka I agree with you partially and I answered somewhat in my post to Alan. I think you are absolutely that the output power-temperature dependence will
                Message 7 of 19 , Nov 29, 2012
                • 0 Attachment
                  Hi Taka

                  I agree with you partially and I answered somewhat in my post to Alan. I think you are absolutely that the output power-temperature dependence will change somewhat. However, because this is in effect a high gain negative feedback loop it will tend to stabilize.

                  What I have found is initial bias voltage is very dependent upon the selection of the Q5 device. As I mentioned in my reply to Alan, there is a very close correlation between the initial bias voltage and the final power output. It would seem there is a significant variance in the drain-source conductivity among devices for a given gate voltage.

                  That is my guess.... i need more work to verify. Thank you for your reply and let us chat more later when I have time.

                  Regards,

                  Warren Allgyer

                  --- In softrock40@yahoogroups.com, "Taka Sugi" <taka_sugi1206@...> wrote:
                  >
                  >
                  >
                  >
                  >
                  >
                  > Hi Warren
                  >
                  > I don't understand why Q5 dramatically affects output power.
                  > Because, Q5/Q4 circuit works as constant voltage circuit to feed bias voltage to PA: Q7,Q8.
                  > Q5 and Q7/Q8 are thermally connected, if Q5 is different maker/lot from Q7/Q8, then output power-temperature dependence will be changed.
                  >
                • Alan
                  ... Subject: [softrock40] Re: To match or not to match... BS170s in your RXTX ... Warren, Yes, but I think the R48/49 voltages are useful. They are a measure
                  Message 8 of 19 , Nov 29, 2012
                  • 0 Attachment
                    ----- Original Message -----
                    Subject: [softrock40] Re: To match or not to match... BS170s in your RXTX


                    >
                    > No I didn't check the static voltage across R48 and 49 although that would have been of interest. I was solely interested in
                    > evaluating the three factors I listed, power, distortion, and harmonic content
                    > One thing I did note was that the different Q5 devices resulted in dramatically different initial bias voltage levels at the diode

                    Warren,

                    Yes, but I think the R48/49 voltages are useful.
                    They are a measure of the bias voltage and how it affects the idle conditions of output devices.
                    And, as they are mentioned in Robby's pages, we have lots of questions when they differ from his quoted values.

                    As your substitutions were effectively changing this value it would be interesting to know your final value. And the bias applied to
                    Q6/7 gates.
                    Surely by making one of the resistors in the bias circuit variable the optimum could be determined without changing Q5? Possibly
                    R41 as W1SRB did.

                    It would also be useful to know if the reccommended value was optimum or whether a higher value might be desirable.

                    >I would choose the 170 that results in the highest bias voltage at the diode. Then I would choose the finals pretty much at random
                    >from the remainder. It certainly would do no harm to match them according to your paper.
                    >
                    > I first began by tack soldering the FETs in place and then connected a prototyping board with very short leads so I could plug
                    > each of the three in separately.
                    >
                    > The really solid way would be to build the bias circuit externally. There is no need for Q7/8 in the test because they draw no
                    > gate current. Whichever device produces the highest bias voltage would win in my contest.
                    >

                    It has been noted before that when Q5 is not matched to the output devices the bias voltage varies. When 3 are matched by the
                    process detailed on my page then close to 50mV is achieved.
                    So presumably this simple procedure could be used with a resistor change to the bias circuit to give a different reccommended value.
                    Otherwise if the Q5 randomly selected your way was one of the few that have has a much different gate threshold voltage than the
                    others maybe a dangerously high standing current could result?

                    73 Alan G4ZFQ





                    > There very well may be consequences to this that I have not encountered yet.


                    > --- In softrock40@yahoogroups.com, "Alan" <alan4alan@...> wrote:
                    >>
                    >>
                    >> ----- Original Message -----
                    >> Subject: [softrock40] To match or not to match... BS170s in your RXTX
                    >>
                    >>
                    >> > Q5 is another story however when it comes to its' affect on total power output. Regardless the combination used in the finals,
                    >> > the
                    >> > device chosen for Q5 dramatically affects the power output from the radio. With the same set of finals, same PS, and same drive
                    >> > levels, I experienced output power variances between 200 mw and 1.2 watts on 10 meters.
                    >>
                    >> Warren,
                    >>
                    >> Previous checks of BS170 matching have established that Q5 affects the bias, as measured here
                    >> http://www.wb5rvz.com/sdr/ensemble/10_pa.htm across R48 & 49. And my summary of the discussion.
                    >> https://sites.google.com/site/g4zfqradio/matching-bs170-for-softrock
                    >> Did you check this each time you changed Q5? As Q5 is a key component in the bias network it would be interesting to know if your
                    >> selected Q5 was actually one that best matched the finals.
                    >> Or whether some other way of matching could be devised? Not everyone has a large number of BS170's or the time to check them as
                    >> you
                    >> did.
                  • warrenallgyer
                    Hi Alan ... Logically I would have to agree with you. However, when a mismatch appears to have insignificant effect on power, IMD, or harmonic content then I
                    Message 9 of 19 , Nov 29, 2012
                    • 0 Attachment
                      Hi Alan


                      > Yes, but I think the R48/49 voltages are useful.
                      > They are a measure of the bias voltage and how it affects the idle conditions of output devices.

                      Logically I would have to agree with you. However, when a mismatch appears to have insignificant effect on power, IMD, or harmonic content then I am not sure where to look for the benefit.

                      I think there is no harm in running the two finals at very different current levels. They are effectively parallel isolated amplifiers operating in opposite phase but the power is combined in parallel. There would appear to be no problem if one is contributing more than its's share of the total output. By closely coupling them thermally, if one is running significantly hotter than the other it will dominate in setting the operating point via Q5.These devices are rated at 500 ma and we are running them at worst case at 60% of the rating.

                      > Surely by making one of the resistors in the bias circuit variable the optimum could be determined without changing Q5? Possibly
                      > R41 as W1SRB did.

                      I think making R41 a 50 ohm variable would be very effective for setting the operating point of the stage. But "optimum' is definitely in the eyes of the beholder. The distortion levels in particular seem to be determined by the power level for a given PS voltage. The operating point does not seem to affect this equation so much. So, for me, 'optimum' would be setting the power output at one watt for CW/Digital and 500-800 mw for SSB, depending upon the PS voltage.

                      > Otherwise if the Q5 randomly selected your way was one of the few that have has a much different gate threshold voltage than the
                      > others maybe a dangerously high standing current could result?

                      I think the standing current is again not particularly important so long as it is well within the device rating. The only time the full standing current is truly "standing" is when PTT is enabled with no drive. Otherwise most if not all of this current becomes RF output. And, even if you were to run the amplifier for an extended period, PTT down and no drive, the thermal feedback through Q5 should reduce the bias voltage, thereby reducing the standing current, and stabilizing the temperature.

                      The more I look at this very simple circuit I am really impressed with the stability and protection it provides.

                      I did go back over my test notes and, with the Q5 I finally selected, the initial bias voltage measured at the R39 hairpin is 3.2 volts. I did not note it after an extended key-down period and this would be an interesting data point to know after the trio has reached a stable key down temperature.

                      I am 'on the road' now for two weeks so any further testing will be deferred until after the 13th.

                      Warren Allgyer - W8TOD
                    • Taka Sugi
                      Hello Warren Yes I agree with you, Q5 variability affects Bias voltage. Especially,Vth variability. ‡™Vbias/‡™Vth is about 1.0 ! If Vth@changes
                      Message 10 of 19 , Nov 29, 2012
                      • 0 Attachment
                        Hello Warren

                        Yes I agree with you, Q5 variability affects Bias voltage. Especially,Vth variability. ‡™Vbias/‡™Vth is about 1.0 ! If Vth@changes 0.2V, then Vbias changes 0.2V.
                        I mentioned before on this message board, I replaced R41(33.2ohm)
                        by 50 ohm trimmer for tuning "initial bias voltage/idle current of Q7/Q8" , to remove variability of Q5.

                        I selected pair BS170s:Q7/Q8 and tuned idle current 25mA for my
                        Ensemble 40/30/20m, and measured IMD3. IMD3 is as follows,...

                        PEP(w)@ 40m 30m 20m
                        0.1 43dB 41dB 44dB
                        0.3 41 38 33
                        0.5 39 36 28
                        0.7 37 34 26
                        1.0 35 32 22

                        I usually use less than 0.7W PEP.

                        73, JA1TLH/Taka

                        > Hi Taka
                        >
                        > I agree with you partially and I answered somewhat in my post to Alan. I think you are absolutely that the output power-temperature dependence will change somewhat. However, because this is in effect a high gain negative feedback loop it will tend to stabilize.
                        >
                        > What I have found is initial bias voltage is very dependent upon the selection of the Q5 device. As I mentioned in my reply to Alan, there is a very close correlation between the initial bias voltage and the final power output. It would seem there is a significant variance in the drain-source conductivity among devices for a given gate voltage.
                        >
                        > That is my guess.... i need more work to verify. Thank you for your reply and let us chat more later when I have time.
                        >
                        > Regards,
                        >
                        > Warren Allgyer
                      • warrenallgyer
                        Hello Taka Congratulations on your excellent IMD3 results. I see you start to have a little bit of problem on 20 meters. I too have noticed this and my
                        Message 11 of 19 , Nov 29, 2012
                        • 0 Attachment
                          Hello Taka

                          Congratulations on your excellent IMD3 results. I see you start to have a little bit of problem on 20 meters. I too have noticed this and my 30/20/17 build shows even worse performance on 17 meters where I can find no power level that gives better than 20 dB for IMD3.

                          Alan has begun some research which indicates the driver stage may be introducing distortion at 17 and maybe at 20 as well. He is measuring this by using a scope probe to the RXII receiver and checking the driver input versus output. I have not had a chance to verify that my driver also has a problem but I suspect it does. I am wondering if this is something that begins to show at 20 meters.

                          This is one more thing I intend to investigate when I return home.

                          I like your idea of a 50 ohm trimmer for R41 and I plan to look for suitable components to convert my radios as well. I think this is much better than swapping different devices in the Q5 position.

                          Best regards,

                          Warren Allgyer - W8TOD
                          (now operating temporarily from the home QTH in Ohio)

                          --- In softrock40@yahoogroups.com, "Taka Sugi" <taka_sugi1206@...> wrote:
                          >
                          >
                          >
                          >
                          > Hello Warren
                          >
                          > Yes I agree with you, Q5 variability affects Bias voltage. Especially,Vth variability. ‡™Vbias/‡™Vth is about 1.0 ! If Vth@changes 0.2V, then Vbias changes 0.2V.
                          > I mentioned before on this message board, I replaced R41(33.2ohm)
                          > by 50 ohm trimmer for tuning "initial bias voltage/idle current of Q7/Q8" , to remove variability of Q5.
                          >
                          > I selected pair BS170s:Q7/Q8 and tuned idle current 25mA for my
                          > Ensemble 40/30/20m, and measured IMD3. IMD3 is as follows,...
                          >
                          > PEP(w)@ 40m 30m 20m
                          > 0.1 43dB 41dB 44dB
                          > 0.3 41 38 33
                          > 0.5 39 36 28
                          > 0.7 37 34 26
                          > 1.0 35 32 22
                          >
                          > I usually use less than 0.7W PEP.
                          >
                          > 73, JA1TLH/Taka
                          >
                          > > Hi Taka
                          > >
                          > > I agree with you partially and I answered somewhat in my post to Alan. I think you are absolutely that the output power-temperature dependence will change somewhat. However, because this is in effect a high gain negative feedback loop it will tend to stabilize.
                          > >
                          > > What I have found is initial bias voltage is very dependent upon the selection of the Q5 device. As I mentioned in my reply to Alan, there is a very close correlation between the initial bias voltage and the final power output. It would seem there is a significant variance in the drain-source conductivity among devices for a given gate voltage.
                          > >
                          > > That is my guess.... i need more work to verify. Thank you for your reply and let us chat more later when I have time.
                          > >
                          > > Regards,
                          > >
                          > > Warren Allgyer
                          >
                        • KQ8M
                          Warren, Where are you in Ohio? I am on the east side of Cleveland. 73, Tim Herrick, KQ8M Charter Member North Coast Contesters kq8m@kq8m.com AR-Cluster V6
                          Message 12 of 19 , Nov 29, 2012
                          • 0 Attachment

                            Warren,

                             

                            Where are you in Ohio? I am on the east side of Cleveland.

                             

                            73,

                            Tim Herrick, KQ8M

                            Charter Member North Coast Contesters

                            kq8m@...

                             

                            AR-Cluster V6 kq8m.no-ip.org

                            User Ports: 23, 7373  with local skimmer, 7374 without local skimmer

                            Server Ports: V6 3607, V4 Active 3605, V4 Passive 3606

                             

                            From: softrock40@yahoogroups.com [mailto:softrock40@yahoogroups.com] On Behalf Of warrenallgyer
                            Sent: Friday, November 30, 2012 1:37 AM
                            To: softrock40@yahoogroups.com
                            Subject: [softrock40] Re: To match or not to match... BS170s in your RXTX

                             

                             


                            Hello Taka

                            Congratulations on your excellent IMD3 results. I see you start to have a little bit of problem on 20 meters. I too have noticed this and my 30/20/17 build shows even worse performance on 17 meters where I can find no power level that gives better than 20 dB for IMD3.

                            Alan has begun some research which indicates the driver stage may be introducing distortion at 17 and maybe at 20 as well. He is measuring this by using a scope probe to the RXII receiver and checking the driver input versus output. I have not had a chance to verify that my driver also has a problem but I suspect it does. I am wondering if this is something that begins to show at 20 meters.

                            This is one more thing I intend to investigate when I return home.

                            I like your idea of a 50 ohm trimmer for R41 and I plan to look for suitable components to convert my radios as well. I think this is much better than swapping different devices in the Q5 position.

                            Best regards,

                            Warren Allgyer - W8TOD
                            (now operating temporarily from the home QTH in Ohio)

                            --- In softrock40@yahoogroups.com, "Taka Sugi" <taka_sugi1206@...> wrote:
                            >
                            >
                            >
                            >
                            > Hello Warren
                            >
                            > Yes I agree with you, Q5 variability affects Bias voltage. Especially,Vth variability. ‡™Vbias/‡™Vth is about 1.0 ! If Vth@changes 0.2V, then Vbias changes 0.2V.
                            > I mentioned before on this message board, I replaced R41(33.2ohm)
                            > by 50 ohm trimmer for tuning "initial bias voltage/idle current of Q7/Q8" , to remove variability of Q5.
                            >
                            > I selected pair BS170s:Q7/Q8 and tuned idle current 25mA for my
                            > Ensemble 40/30/20m, and measured IMD3. IMD3 is as follows,...
                            >
                            > PEP(w)@ 40m 30m 20m
                            > 0.1 43dB 41dB 44dB
                            > 0.3 41 38 33
                            > 0.5 39 36 28
                            > 0.7 37 34 26
                            > 1.0 35 32 22
                            >
                            > I usually use less than 0.7W PEP.
                            >
                            > 73, JA1TLH/Taka
                            >
                            > > Hi Taka
                            > >
                            > > I agree with you partially and I answered somewhat in my post to Alan. I think you are absolutely that the output power-temperature dependence will change somewhat. However, because this is in effect a high gain negative feedback loop it will tend to stabilize.
                            > >
                            > > What I have found is initial bias voltage is very dependent upon the selection of the Q5 device. As I mentioned in my reply to Alan, there is a very close correlation between the initial bias voltage and the final power output. It would seem there is a significant variance in the drain-source conductivity among devices for a given gate voltage.
                            > >
                            > > That is my guess.... i need more work to verify. Thank you for your reply and let us chat more later when I have time.
                            > >
                            > > Regards,
                            > >
                            > > Warren Allgyer
                            >

                          • warrenallgyer
                            Hi Tim I am in Gnadenhutten, OH. About 100 miles south of you on I77 Warren Allgyer -W8TOD
                            Message 13 of 19 , Nov 30, 2012
                            • 0 Attachment
                              Hi Tim

                              I am in Gnadenhutten, OH. About 100 miles south of you on I77

                              Warren Allgyer -W8TOD

                              --- In softrock40@yahoogroups.com, "KQ8M" <kq8m@...> wrote:
                              >
                              > Warren,
                              >
                              >
                              >
                              > Where are you in Ohio? I am on the east side of Cleveland.
                              >
                              >
                              >
                              > 73,
                              >
                              > Tim Herrick, KQ8M
                              >
                              > Charter Member North Coast Contesters
                              >
                              > kq8m@...
                              >
                              >
                              >
                              > AR-Cluster V6 kq8m.no-ip.org
                              >
                              > User Ports: 23, 7373 with local skimmer, 7374 without local skimmer
                              >
                              > Server Ports: V6 3607, V4 Active 3605, V4 Passive 3606
                              >
                              >
                              >
                              > From: softrock40@yahoogroups.com [mailto:softrock40@yahoogroups.com] On Behalf Of warrenallgyer
                              > Sent: Friday, November 30, 2012 1:37 AM
                              > To: softrock40@yahoogroups.com
                              > Subject: [softrock40] Re: To match or not to match... BS170s in your RXTX
                              >
                              >
                              >
                              >
                              >
                              >
                              > Hello Taka
                              >
                              > Congratulations on your excellent IMD3 results. I see you start to have a little bit of problem on 20 meters. I too have noticed this and my 30/20/17 build shows even worse performance on 17 meters where I can find no power level that gives better than 20 dB for IMD3.
                              >
                              > Alan has begun some research which indicates the driver stage may be introducing distortion at 17 and maybe at 20 as well. He is measuring this by using a scope probe to the RXII receiver and checking the driver input versus output. I have not had a chance to verify that my driver also has a problem but I suspect it does. I am wondering if this is something that begins to show at 20 meters.
                              >
                              > This is one more thing I intend to investigate when I return home.
                              >
                              > I like your idea of a 50 ohm trimmer for R41 and I plan to look for suitable components to convert my radios as well. I think this is much better than swapping different devices in the Q5 position.
                              >
                              > Best regards,
                              >
                              > Warren Allgyer - W8TOD
                              > (now operating temporarily from the home QTH in Ohio)
                              >
                              > --- In softrock40@yahoogroups.com <mailto:softrock40%40yahoogroups.com> , "Taka Sugi" <taka_sugi1206@> wrote:
                              > >
                              > >
                              > >
                              > >
                              > > Hello Warren
                              > >
                              > > Yes I agree with you, Q5 variability affects Bias voltage. Especially,Vth variability. ‡™Vbias/‡™Vth is about 1.0 ! If Vth@changes 0.2V, then Vbias changes 0.2V.
                              > > I mentioned before on this message board, I replaced R41(33.2ohm)
                              > > by 50 ohm trimmer for tuning "initial bias voltage/idle current of Q7/Q8" , to remove variability of Q5.
                              > >
                              > > I selected pair BS170s:Q7/Q8 and tuned idle current 25mA for my
                              > > Ensemble 40/30/20m, and measured IMD3. IMD3 is as follows,...
                              > >
                              > > PEP(w)@ 40m 30m 20m
                              > > 0.1 43dB 41dB 44dB
                              > > 0.3 41 38 33
                              > > 0.5 39 36 28
                              > > 0.7 37 34 26
                              > > 1.0 35 32 22
                              > >
                              > > I usually use less than 0.7W PEP.
                              > >
                              > > 73, JA1TLH/Taka
                              > >
                              > > > Hi Taka
                              > > >
                              > > > I agree with you partially and I answered somewhat in my post to Alan. I think you are absolutely that the output power-temperature dependence will change somewhat. However, because this is in effect a high gain negative feedback loop it will tend to stabilize.
                              > > >
                              > > > What I have found is initial bias voltage is very dependent upon the selection of the Q5 device. As I mentioned in my reply to Alan, there is a very close correlation between the initial bias voltage and the final power output. It would seem there is a significant variance in the drain-source conductivity among devices for a given gate voltage.
                              > > >
                              > > > That is my guess.... i need more work to verify. Thank you for your reply and let us chat more later when I have time.
                              > > >
                              > > > Regards,
                              > > >
                              > > > Warren Allgyer
                              > >
                              >
                            • warrenallgyer
                              Hi Taka Here is what I see in this circuit: There is a long voltage divider chain from the 12V rail consisting of Q3, R37(22.1K),Q5, and R41(33.2). As Q5 is
                              Message 14 of 19 , Nov 30, 2012
                              • 0 Attachment
                                Hi Taka

                                Here is what I see in this circuit:

                                There is a long voltage divider chain from the 12V rail consisting of Q3, R37(22.1K),Q5, and R41(33.2). As Q5 is driven more and more into conduction by heat from the Q5/Q7/Q8 combo two things happen:

                                1) As discussed, current through Q5 drives Q4 and effectively reduces the resistance of R40. This lowers the bias level on the finals.

                                2) The increased current through Q5 also increases the voltage drop across R37 which reduces the collector voltage on Q6, the 2n2222 driver. This reduces the drive level to the finals, reducing the power output.

                                So variances in Q5 produce big variances in the Q6 drive level as well as the operating point of the finals. This is why the Q5 circuitry is so effective at stabilizing the finals and why a choice here affects the power performance of the radio.

                                This is my analysis. Do you agree?

                                Warren Allgyer - W8TOD

                                --- In softrock40@yahoogroups.com, "Taka Sugi" <taka_sugi1206@...> wrote:
                                >
                                >
                                >
                                >
                                >
                                >
                                > Hi Warren
                                >
                                > I don't understand why Q5 dramatically affects output power.
                                > Because, Q5/Q4 circuit works as constant voltage circuit to feed bias voltage to PA: Q7,Q8.
                                > Q5 and Q7/Q8 are thermally connected, if Q5 is different maker/lot from Q7/Q8, then output power-temperature dependence will be changed.
                                >
                              • KF4BQ
                                ... Hi Warren, Some bias point operation and other issues discussed at: http://groups.yahoo.com/group/softrock40/message/55570 73, Mike Collins KF4BQ
                                Message 15 of 19 , Nov 30, 2012
                                • 0 Attachment
                                  --- In softrock40@yahoogroups.com, "warrenallgyer" <allgyer@...> wrote:
                                  >
                                  > So variances in Q5 produce big variances in the Q6 drive level as well as the operating point of the finals. This is why the Q5 circuitry is so effective at stabilizing the finals and why a choice here affects the power performance of the radio.
                                  >
                                  > This is my analysis. Do you agree?
                                  >

                                  Hi Warren,

                                  Some bias point operation and other issues discussed at:
                                  http://groups.yahoo.com/group/softrock40/message/55570

                                  73, Mike Collins KF4BQ
                                • Taka Sugi
                                  Hi Warren Yes, almost I agree. ... 1. Ib(Q4)= Id(Q5) - Vbe(Q4)/R41 Here Vbe(Q4) is about 0.6V 2. Ic(Q4)=hfe(Q4) x Ib(Q4) 3. Vbias= V1 - (R39 xIc(Q4) + Vd(D3))
                                  Message 16 of 19 , Dec 2, 2012
                                  • 0 Attachment
                                    Hi Warren

                                    Yes, almost I agree.

                                    > 1) As discussed, current through Q5 drives Q4 and effectively
                                    >reduces the resistance of R40. This lowers the bias level on
                                    >the finals.

                                    1. Ib(Q4)= Id(Q5) - Vbe(Q4)/R41
                                    Here Vbe(Q4) is about 0.6V
                                    2. Ic(Q4)=hfe(Q4) x Ib(Q4)
                                    3. Vbias= V1 - (R39 xIc(Q4) + Vd(D3))
                                    Here V1= S12v - (R37 xIbias-circuit + Ic(Q6))
                                    If Id(Q5) is increased, then Ic(Q4) is increased, as a result
                                    Vbias is decreased( goes down).

                                    If Id(Q7,Q8) goes up by temperature, then Id(Q5, thermally connected
                                    to Q7,Q8) also goes up and Vbias goes down, then Id(Q7,Q8) goes down.
                                    This is a kind of "feedback".
                                    I am not sure about the function of R40(22.1k-ohm),but I think this
                                    is to avoide "local oscillation"?

                                    > 2) The increased current through Q5 also increases the voltage
                                    > drop across R37 which reduces the collector voltage on Q6, the
                                    >2n2222 driver. This reduces the drive level to the finals, reducing
                                    >the power output.
                                    > So variances in Q5 produce big variances in the Q6 drive level as
                                    >well as the operating point of the finals. This is why the Q5
                                    >circuitry is so effective at stabilizing the finals and why a
                                    >choice here affects the power performance of the radio.
                                    >
                                    > This is my analysis. Do you agree?

                                    I think there is little influence to Ic of driver Tr:Q6(2N2222),
                                    unless Q5 varies drastically, because R37 is small ( 22.1 ohm).

                                    JA1TLH/Taka
                                  • Alan
                                    ... Subject: [softrock40] Re: To match or not to match... BS170s in your RXTX Taka, Did you read this? http://groups.yahoo.com/group/softrock40/message/55570
                                    Message 17 of 19 , Dec 3, 2012
                                    • 0 Attachment
                                      ----- Original Message -----
                                      Subject: [softrock40] Re: To match or not to match... BS170s in your RXTX


                                      Taka,

                                      Did you read this? http://groups.yahoo.com/group/softrock40/message/55570

                                      Mike points out that Q4 maintains a constant current through R41 and therefore through Q5.
                                      The current is constant in Q5, what varies is the voltage across it.
                                      This is passed to Q7/8 through D3.
                                      Q4 just maintains the constant current.

                                      Mike, are you there?

                                      73 Alan G4ZFQ



                                      > Yes, almost I agree.
                                      >
                                      >> 1) As discussed, current through Q5 drives Q4 and effectively
                                      >>reduces the resistance of R40. This lowers the bias level on
                                      >>the finals.
                                      >
                                      > 1. Ib(Q4)= Id(Q5) - Vbe(Q4)/R41
                                      > Here Vbe(Q4) is about 0.6V
                                      > 2. Ic(Q4)=hfe(Q4) x Ib(Q4)
                                      > 3. Vbias= V1 - (R39 xIc(Q4) + Vd(D3))
                                      > Here V1= S12v - (R37 xIbias-circuit + Ic(Q6))
                                      > If Id(Q5) is increased, then Ic(Q4) is increased, as a result
                                      > Vbias is decreased( goes down).
                                      >
                                      > If Id(Q7,Q8) goes up by temperature, then Id(Q5, thermally connected
                                      > to Q7,Q8) also goes up and Vbias goes down, then Id(Q7,Q8) goes down.
                                      > This is a kind of "feedback".
                                      > I am not sure about the function of R40(22.1k-ohm),but I think this
                                      > is to avoide "local oscillation"?
                                      >
                                      >> 2) The increased current through Q5 also increases the voltage
                                      >> drop across R37 which reduces the collector voltage on Q6, the
                                      >>2n2222 driver. This reduces the drive level to the finals, reducing
                                      >>the power output.
                                      >> So variances in Q5 produce big variances in the Q6 drive level as
                                      >>well as the operating point of the finals. This is why the Q5
                                      >>circuitry is so effective at stabilizing the finals and why a
                                      >>choice here affects the power performance of the radio.
                                      >>
                                      >> This is my analysis. Do you agree?
                                      >
                                      > I think there is little influence to Ic of driver Tr:Q6(2N2222),
                                      > unless Q5 varies drastically, because R37 is small ( 22.1 ohm).
                                    • Taka Sugi
                                      Hi Alan ... Yes, I knew. I analyed before and also simulated@by@LTspice. If you have an interest pse refer to...
                                      Message 18 of 19 , Dec 4, 2012
                                      • 0 Attachment
                                        Hi Alan
                                        > Did you read this? http://groups.yahoo.com/group/softrock40/message/55570

                                        Yes, I knew.
                                        I analyed before and also simulated@by@LTspice.
                                        If you have an interest pse refer to...

                                        http://dl.dropbox.com/u/85347415/HAM/temp/PA_Bias%20Circuit.doc

                                        JA1TLH/Taka
                                      • Alan
                                        ... Subject: [softrock40] Re: To match or not to match... BS170s in your RXTX Sorry Taka, I am too dumb to understand. But finally I see. I need simple
                                        Message 19 of 19 , Dec 4, 2012
                                        • 0 Attachment
                                          ----- Original Message -----
                                          Subject: [softrock40] Re: To match or not to match... BS170s in your RXTX


                                          Sorry Taka,

                                          I am too dumb to understand.

                                          But finally I see. I need simple explanations without mathematics.
                                          Q4 is configured to pass a constant current through R41. It does this by controlling Q5.
                                          So Q5 source/gate voltage is set by Q4 so it will pass that constant current keeping Q4 base at ~0V7.
                                          Q5 gate voltage will vary according to the temperature of Q5 (in thermal contact with Q7/8) and is sent to Q7/8 with D3 dropping
                                          ~0V7 to compensate for Q4 base voltage.

                                          Simple, but not as simple as me.

                                          73 Alan G4ZFQ
                                        Your message has been successfully submitted and would be delivered to recipients shortly.