Merry Christmas to all in the group! Blessings and best wishes to all.

 

I'll start with a short story which almost resulted in me posing an entirely different question to the group. I spent hours in the shop last night playing with a set of TX/RX 220mhz duplexer I recently acquired (odd, it has two model numbers listed on the tag - 43-54A-02479-A & 26-54A-02479A), trying to line them up on a 222/223 amateur pair. Following the manufacturer instructions, I would continuously come up with final results with the passes and notches off frequency by roughly 100khz. Current test equipment is an Aeroflex 3920 and Eagle RLB150x3 return loss bridge. After tuning with these instruments, I've gotten into the habit of using my RigExpert AA-1500 to verify proper tuning right before installation, with very good results - I find it to be quite accurate. Fast forward to this morning after a few fresh cups of coffee and at a complete loss for the continued disagreement, I remember I have another RLB150 and decide to swap it out. Bingo - a much deeper notch on the scope and now in agreement with my reference measurement. I guess it's a good reminder to always test your accessories before embarking upon these journeys. 

 

Now, on to my next question. With regard to pass tuning, I now can tune each can individually achieving acceptable RL (30+ db). When paired back together, the passes remain in alignment and present an acceptable RL (better than 15bd) but can be improved (closer to 30bd) by additional tuning rod adjustment. After this fine adjustment, I went back and swept both cans individually and found that it moved the passes pretty significantly, and now presented an unacceptable RL around 12 db. 

 

My question is: should I tune for highest RL with each cavity in cascade or accept the lower but acceptable RL achieved by tuning the passes individually then placing them in cascade with little/no further adjustment. 

 

Merry Christmas,

Jeremy K1LFK

On 12/24/2024 8:29 PM EST Kevin Custer <kuggie@...> wrote:

 

 

Merry Christmas Jeremy (and others).

I don't have an answer or comment about your first question, but it may be tied to what's in the answer for the second question.

As to your second question - a wise man (Jeff / WN3A) once told me, "Don't use your duplexer as an antenna tuner".  If you don't know what that means - let me explain....   I apologize, this is long.

When you tune a single cavity for resonance using return loss - it's correctly tuned.  When you add additional cavities in series (that have all been tuned correctly, and when the phasing cables are the correct length) - the tuning of the resonance doesn't change - therefore - you shouldn't need to re-peak anything.  If you do 'touch them up' - you can be compensating for imaginary problems and/or inducing new ones.  As such - it may look like you are making it better - when in reality - you are making it worse.  The imaginary problems can be a less than perfect calibration load, less than prefect calibration of the VNA, and less than perfect test cables.  It can also be the desire to make the return loss 'dips' align on top of one another - which there is no reason to do so. 

After cabling the duplexer together, re-adjusting the pass frequency adjustments may lead you to believe your making an improvement - because the display "looks better" - but is it really better?  The answer is - probably not, and more so - very likely not.  What you may be doing is compensating for imperfection where two mismatches end up looking better on the display as a (conjugate match).  This is the same effect as what a transmatch does to a non resonant antenna - only in this case, you're taking something that has already been tuned to resonance and you are mis-aligning of an apparent improvement - when in reality, there is none. 

I've seen YouTube videos that say to realign after re-cabling - and I don't agree with them.  A tiny bit of readjustment can add symmetry, but if you're looking at S11/S22 and S21/S12 anyway, you're going to be okay.

Remember:
1 - When you add cavities to the duplexer leg, the return loss bandwidth widens. 
     In other words - the RL "dips" may not (likely won't) line up on top of one another.  This is NOT necessarily a bad thing - and can be a good thing.
2 - When adding cavities to the leg, the return loss graph will become humpy (have ripple), and that's okay too. 
     Nothing says the return loss "dip" response has to have a nice smooth bottom to it.
3 - The cavity's resonance as revealed by RL (dip) and IL (peak) may not line up. 
     Use the center of the RL dip for the actual pass resonant frequency on your repeater's transmit and receive frequencies.
4 - When you measure IL - do it at your repeater frequencies - even though there may be a frequency along the graph that shows less loss.
     Do this because of #3.

#3 is a big one on VHF duplexers.  I can't stress enough that you don't tune a cavity's pass frequency to the frequency that provides the lowest insertion loss.  That's not the pass frequency - - - the frequency that provides the best return loss is the 'real' pass frequency.  Obviously then - when measuring insertion loss - do it at the real pass frequency - not the frequency that shows the lowest insertion loss. 

Tuning tips:
1 - Set all coupling loops for the same IL - while keeping the pass to notch frequency delta close to what you need.
2 - Couple the cavities together and verify that the IL is the sum of all cavities plus a little for cable / connector loss.
3 - Verify that you have at least -20 dB of RL on each of the pass frequencies (more is better).
4 - Do NOT be tempted to retune the pass frequencies or coupling loop rotations even though you believe you are making the tuning 'better'.

In summary - What's important is that you have 20dB or more return loss on your pass frequencies (more is better).  Also verify the insertion loss is the simple math sum of the cavities in series - plus a tiny bit for cable and connector losses.  Two cavities set for 0.6dB IL should work out to less than 1.5dB total insertion loss. 

After all of that is set correctly, cabled together, and verified - then tune the notches for the most rejection. 

If you started out with each cavity showing 30+ dB of return loss - when cabled together, I'd expect to see 25dB or so when completely cabled together, the notches properly tuned, and the opposite port properly terminated.  Some degradation of the return loss amount as compared to a single cavity is to be expected.

Kevin W3KKC


On 12/24/2024 1:49 PM, Jeremy Hansen via groups.io wrote:

Merry Christmas to all in the group! Blessings and best wishes to all.

 

I'll start with a short story which almost resulted in me posing an entirely different question to the group. I spent hours in the shop last night playing with a set of TX/RX 220mhz duplexer I recently acquired (odd, it has two model numbers listed on the tag - 43-54A-02479-A & 26-54A-02479A), trying to line them up on a 222/223 amateur pair. Following the manufacturer instructions, I would continuously come up with final results with the passes and notches off frequency by roughly 100khz. Current test equipment is an Aeroflex 3920 and Eagle RLB150x3 return loss bridge. After tuning with these instruments, I've gotten into the habit of using my RigExpert AA-1500 to verify proper tuning right before installation, with very good results - I find it to be quite accurate. Fast forward to this morning after a few fresh cups of coffee and at a complete loss for the continued disagreement, I remember I have another RLB150 and decide to swap it out. Bingo - a much deeper notch on the scope and now in agreement with my reference measurement. I guess it's a good reminder to always test your accessories before embarking upon these journeys. 

 

Now, on to my next question. With regard to pass tuning, I now can tune each can individually achieving acceptable RL (30+ db). When paired back together, the passes remain in alignment and present an acceptable RL (better than 15bd) but can be improved (closer to 30bd) by additional tuning rod adjustment. After this fine adjustment, I went back and swept both cans individually and found that it moved the passes pretty significantly, and now presented an unacceptable RL around 12 db. 

 

My question is: should I tune for highest RL with each cavity in cascade or accept the lower but acceptable RL achieved by tuning the passes individually then placing them in cascade with little/no further adjustment. 

 

Merry Christmas,

Jeremy K1LFK

Really appreciate your clarifications, Kevin. Now to up the complexity of alignment. While Jeremy was wrestling coupling 2 BP/BR cans, I was doing the same with one and then 2 BP cans in series with just a tracking generator/ spectrum analyzer and alternating with a return loss bridge without the luxury of a VNA to see both at once.

 

My question is that with just one bandpass can, did I do the adjustments  properly?  I first adjusted the 2 loops on a single can for a setting that gave the least insertion loss at the point where the skirts just started to want to go steeper. I figured this is the point of critical coupling. Also the loops were approximately the same angular rotation spot as related to the center resonator, but of course on opposite sides, so like mirror images. Then went to return loss sweeps and found different return loss levels depending which port was input. After lots of back and forth swamping input/ output ports, if i set the input port to best return loss , and then switched input ports the return loss became worse (lower) when peaking the opposite as input, and it just went back and forth that way, so I finally concluded that the correct loop rotation tweaking was the loop position to make closest to the same rotation from the resonator on both and by moving the input loop that had the higher RL down in the direction that gave a middle RL reading and most made the loops the same.

distance from the resonator. Then the can had balanced input/ output ports so it made no RL difference which port was input.

 

Then after both cans were aligned that way, they were connected  with different lengths of coax between them. Until the exact optimum either electrical 1/4 or 3/4 wavelength, the position of the 2 RL dips moved all around as did what became  2 peaks. Finally with the proper coax length by trial and error, the 2 peaks of lowest loss became one and the 2 RL dips were then inside that peak. Interestingly, the loop lengths had to be a component of that electrical 1/4 or 3/4 wavelength because their inductance apparently needed to be tuned out by having the coax portion actually LONGER electrically than 1/4 or 3/4 wavelength, so as I think Jeff said, you sometimes just have to arrive at the best coax jumper length experimentally!

 

Seems to me in a duplexer with series cap and loop reject , the series cap tunes out the loop inductance on the reject frequency and probably mostly on the pass  frequency , so the coax jumpers themselves are quite close electrically to 1/4 or 3/4 wavelength. In my experience with a BP/BR duplexer one of the biggest reasons RL and peak do not line up is an improper coax length from the antenna port T to the opposite set of cans, which is easily tested if there is alignment between peak and RL without the antenna T.

 

John

On 12/24/2024 8:29 PM EST Kevin Custer <kuggie@...> wrote:

 

 

Merry Christmas Jeremy (and others).

I don't have an answer or comment about your first question, but it may be tied to what's in the answer for the second question.

As to your second question - a wise man (Jeff / WN3A) once told me, "Don't use your duplexer as an antenna tuner".  If you don't know what that means - let me explain....   I apologize, this is long.

When you tune a single cavity for resonance using return loss - it's correctly tuned.  When you add additional cavities in series (that have all been tuned correctly, and when the phasing cables are the correct length) - the tuning of the resonance doesn't change - therefore - you shouldn't need to re-peak anything.  If you do 'touch them up' - you can be compensating for imaginary problems and/or inducing new ones.  As such - it may look like you are making it better - when in reality - you are making it worse.  The imaginary problems can be a less than perfect calibration load, less than prefect calibration of the VNA, and less than perfect test cables.  It can also be the desire to make the return loss 'dips' align on top of one another - which there is no reason to do so. 

After cabling the duplexer together, re-adjusting the pass frequency adjustments may lead you to believe your making an improvement - because the display "looks better" - but is it really better?  The answer is - probably not, and more so - very likely not.  What you may be doing is compensating for imperfection where two mismatches end up looking better on the display as a (conjugate match).  This is the same effect as what a transmatch does to a non resonant antenna - only in this case, you're taking something that has already been tuned to resonance and you are mis-aligning of an apparent improvement - when in reality, there is none. 

I've seen YouTube videos that say to realign after re-cabling - and I don't agree with them.  A tiny bit of readjustment can add symmetry, but if you're looking at S11/S22 and S21/S12 anyway, you're going to be okay.

Remember:
1 - When you add cavities to the duplexer leg, the return loss bandwidth widens. 
     In other words - the RL "dips" may not (likely won't) line up on top of one another.  This is NOT necessarily a bad thing - and can be a good thing.
2 - When adding cavities to the leg, the return loss graph will become humpy (have ripple), and that's okay too. 
     Nothing says the return loss "dip" response has to have a nice smooth bottom to it.
3 - The cavity's resonance as revealed by RL (dip) and IL (peak) may not line up. 
     Use the center of the RL dip for the actual pass resonant frequency on your repeater's transmit and receive frequencies.
4 - When you measure IL - do it at your repeater frequencies - even though there may be a frequency along the graph that shows less loss.
     Do this because of #3.

#3 is a big one on VHF duplexers.  I can't stress enough that you don't tune a cavity's pass frequency to the frequency that provides the lowest insertion loss.  That's not the pass frequency - - - the frequency that provides the best return loss is the 'real' pass frequency.  Obviously then - when measuring insertion loss - do it at the real pass frequency - not the frequency that shows the lowest insertion loss. 

Tuning tips:
1 - Set all coupling loops for the same IL - while keeping the pass to notch frequency delta close to what you need.
2 - Couple the cavities together and verify that the IL is the sum of all cavities plus a little for cable / connector loss.
3 - Verify that you have at least -20 dB of RL on each of the pass frequencies (more is better).
4 - Do NOT be tempted to retune the pass frequencies or coupling loop rotations even though you believe you are making the tuning 'better'.

In summary - What's important is that you have 20dB or more return loss on your pass frequencies (more is better).  Also verify the insertion loss is the simple math sum of the cavities in series - plus a tiny bit for cable and connector losses.  Two cavities set for 0.6dB IL should work out to less than 1.5dB total insertion loss. 

After all of that is set correctly, cabled together, and verified - then tune the notches for the most rejection. 

If you started out with each cavity showing 30+ dB of return loss - when cabled together, I'd expect to see 25dB or so when completely cabled together, the notches properly tuned, and the opposite port properly terminated.  Some degradation of the return loss amount as compared to a single cavity is to be expected.

Kevin W3KKC


On 12/24/2024 1:49 PM, Jeremy Hansen via groups.io wrote:

Merry Christmas to all in the group! Blessings and best wishes to all.

 

I'll start with a short story which almost resulted in me posing an entirely different question to the group. I spent hours in the shop last night playing with a set of TX/RX 220mhz duplexer I recently acquired (odd, it has two model numbers listed on the tag - 43-54A-02479-A & 26-54A-02479A), trying to line them up on a 222/223 amateur pair. Following the manufacturer instructions, I would continuously come up with final results with the passes and notches off frequency by roughly 100khz. Current test equipment is an Aeroflex 3920 and Eagle RLB150x3 return loss bridge. After tuning with these instruments, I've gotten into the habit of using my RigExpert AA-1500 to verify proper tuning right before installation, with very good results - I find it to be quite accurate. Fast forward to this morning after a few fresh cups of coffee and at a complete loss for the continued disagreement, I remember I have another RLB150 and decide to swap it out. Bingo - a much deeper notch on the scope and now in agreement with my reference measurement. I guess it's a good reminder to always test your accessories before embarking upon these journeys. 

 

Now, on to my next question. With regard to pass tuning, I now can tune each can individually achieving acceptable RL (30+ db). When paired back together, the passes remain in alignment and present an acceptable RL (better than 15bd) but can be improved (closer to 30bd) by additional tuning rod adjustment. After this fine adjustment, I went back and swept both cans individually and found that it moved the passes pretty significantly, and now presented an unacceptable RL around 12 db. 

 

My question is: should I tune for highest RL with each cavity in cascade or accept the lower but acceptable RL achieved by tuning the passes individually then placing them in cascade with little/no further adjustment. 

 

Merry Christmas,

Jeremy K1LFK

Merry Christmas Jeremy (and others).

I don't have an answer or comment about your first question, but it may be tied to what's in the answer for the second question.

As to your second question - a wise man (Jeff / WN3A) once told me, "Don't use your duplexer as an antenna tuner".  If you don't know what that means - let me explain....   I apologize, this is long.

When you tune a single cavity for resonance using return loss - it's correctly tuned.  When you add additional cavities in series (that have all been tuned correctly, and when the phasing cables are the correct length) - the tuning of the resonance doesn't change - therefore - you shouldn't need to re-peak anything.  If you do 'touch them up' - you can be compensating for imaginary problems and/or inducing new ones.  As such - it may look like you are making it better - when in reality - you are making it worse.  The imaginary problems can be a less than perfect calibration load, less than prefect calibration of the VNA, and less than perfect test cables.  It can also be the desire to make the return loss 'dips' align on top of one another - which there is no reason to do so. 

After cabling the duplexer together, re-adjusting the pass frequency adjustments may lead you to believe your making an improvement - because the display "looks better" - but is it really better?  The answer is - probably not, and more so - very likely not.  What you may be doing is compensating for imperfection where two mismatches end up looking better on the display as a (conjugate match).  This is the same effect as what a transmatch does to a non resonant antenna - only in this case, you're taking something that has already been tuned to resonance and you are mis-aligning of an apparent improvement - when in reality, there is none. 

I've seen YouTube videos that say to realign after re-cabling - and I don't agree with them.  A tiny bit of readjustment can add symmetry, but if you're looking at S11/S22 and S21/S12 anyway, you're going to be okay.

Remember:
1 - When you add cavities to the duplexer leg, the return loss bandwidth widens. 
     In other words - the RL "dips" may not (likely won't) line up on top of one another.  This is NOT necessarily a bad thing - and can be a good thing.
2 - When adding cavities to the leg, the return loss graph will become humpy (have ripple), and that's okay too. 
     Nothing says the return loss "dip" response has to have a nice smooth bottom to it.
3 - The cavity's resonance as revealed by RL (dip) and IL (peak) may not line up. 
     Use the center of the RL dip for the actual pass resonant frequency on your repeater's transmit and receive frequencies.
4 - When you measure IL - do it at your repeater frequencies - even though there may be a frequency along the graph that shows less loss.
     Do this because of #3.

#3 is a big one on VHF duplexers.  I can't stress enough that you don't tune a cavity's pass frequency to the frequency that provides the lowest insertion loss.  That's not the pass frequency - - - the frequency that provides the best return loss is the 'real' pass frequency.  Obviously then - when measuring insertion loss - do it at the real pass frequency - not the frequency that shows the lowest insertion loss. 

Tuning tips:
1 - Set all coupling loops for the same IL - while keeping the pass to notch frequency delta close to what you need.
2 - Couple the cavities together and verify that the IL is the sum of all cavities plus a little for cable / connector loss.
3 - Verify that you have at least -20 dB of RL on each of the pass frequencies (more is better).
4 - Do NOT be tempted to retune the pass frequencies or coupling loop rotations even though you believe you are making the tuning 'better'.

In summary - What's important is that you have 20dB or more return loss on your pass frequencies (more is better).  Also verify the insertion loss is the simple math sum of the cavities in series - plus a tiny bit for cable and connector losses.  Two cavities set for 0.6dB IL should work out to less than 1.5dB total insertion loss. 

After all of that is set correctly, cabled together, and verified - then tune the notches for the most rejection. 

If you started out with each cavity showing 30+ dB of return loss - when cabled together, I'd expect to see 25dB or so when completely cabled together, the notches properly tuned, and the opposite port properly terminated.  Some degradation of the return loss amount as compared to a single cavity is to be expected.

Kevin W3KKC


On 12/24/2024 1:49 PM, Jeremy Hansen via groups.io wrote:

Merry Christmas to all in the group! Blessings and best wishes to all.

 

I'll start with a short story which almost resulted in me posing an entirely different question to the group. I spent hours in the shop last night playing with a set of TX/RX 220mhz duplexer I recently acquired (odd, it has two model numbers listed on the tag - 43-54A-02479-A & 26-54A-02479A), trying to line them up on a 222/223 amateur pair. Following the manufacturer instructions, I would continuously come up with final results with the passes and notches off frequency by roughly 100khz. Current test equipment is an Aeroflex 3920 and Eagle RLB150x3 return loss bridge. After tuning with these instruments, I've gotten into the habit of using my RigExpert AA-1500 to verify proper tuning right before installation, with very good results - I find it to be quite accurate. Fast forward to this morning after a few fresh cups of coffee and at a complete loss for the continued disagreement, I remember I have another RLB150 and decide to swap it out. Bingo - a much deeper notch on the scope and now in agreement with my reference measurement. I guess it's a good reminder to always test your accessories before embarking upon these journeys. 

 

Now, on to my next question. With regard to pass tuning, I now can tune each can individually achieving acceptable RL (30+ db). When paired back together, the passes remain in alignment and present an acceptable RL (better than 15bd) but can be improved (closer to 30bd) by additional tuning rod adjustment. After this fine adjustment, I went back and swept both cans individually and found that it moved the passes pretty significantly, and now presented an unacceptable RL around 12 db. 

 

My question is: should I tune for highest RL with each cavity in cascade or accept the lower but acceptable RL achieved by tuning the passes individually then placing them in cascade with little/no further adjustment. 

 

Merry Christmas,

Jeremy K1LFK

Kevin,

As Gary mentioned, I plan to keep a hard copy of your detailed response. Thanks for the gift of sharing some very valuable knowledge.

I've read it over twice and will likely give it a third look today. I'm 90% tracking the theory, with my brain trying to wrap itself around the last 10% (mostly the comment about it being "ok"  for the RL notches not to line up after placing cavities in cascade. I'm getting there, especially with the help of people willing to share their knowledge and experience. Sincerely, thank you.

John, I'm also appreciative of your additional questions and commentary. Like you, I'm using a SA/TG to make these adjustments. Kevin's response is almost enough to push me into finally breaking down and purchasing a decent VNA (I've been following that thread too....)

Back to gifts and family time for now - more questions to follow.

Merry Christmas!
Jeremy K1LFK

On Wed, Dec 25, 2024, 9:39 AM Gary Cook via groups.io <gac=probemail.com@groups.io> wrote:

Thank you, Kevin (I printed your reply). And, Merry Christmas.

Gary

On 12/24/2024 7:29 PM, Kevin Custer wrote:

Merry Christmas Jeremy (and others).

I don't have an answer or comment about your first question, but it may be tied to what's in the answer for the second question.

As to your second question - a wise man (Jeff / WN3A) once told me, "Don't use your duplexer as an antenna tuner".  If you don't know what that means - let me explain....   I apologize, this is long.

When you tune a single cavity for resonance using return loss - it's correctly tuned.  When you add additional cavities in series (that have all been tuned correctly, and when the phasing cables are the correct length) - the tuning of the resonance doesn't change - therefore - you shouldn't need to re-peak anything.  If you do 'touch them up' - you can be compensating for imaginary problems and/or inducing new ones.  As such - it may look like you are making it better - when in reality - you are making it worse.  The imaginary problems can be a less than perfect calibration load, less than prefect calibration of the VNA, and less than perfect test cables.  It can also be the desire to make the return loss 'dips' align on top of one another - which there is no reason to do so. 

After cabling the duplexer together, re-adjusting the pass frequency adjustments may lead you to believe your making an improvement - because the display "looks better" - but is it really better?  The answer is - probably not, and more so - very likely not.  What you may be doing is compensating for imperfection where two mismatches end up looking better on the display as a (conjugate match).  This is the same effect as what a transmatch does to a non resonant antenna - only in this case, you're taking something that has already been tuned to resonance and you are mis-aligning of an apparent improvement - when in reality, there is none. 

I've seen YouTube videos that say to realign after re-cabling - and I don't agree with them.  A tiny bit of readjustment can add symmetry, but if you're looking at S11/S22 and S21/S12 anyway, you're going to be okay.

Remember:
1 - When you add cavities to the duplexer leg, the return loss bandwidth widens. 
     In other words - the RL "dips" may not (likely won't) line up on top of one another.  This is NOT necessarily a bad thing - and can be a good thing.
2 - When adding cavities to the leg, the return loss graph will become humpy (have ripple), and that's okay too. 
     Nothing says the return loss "dip" response has to have a nice smooth bottom to it.
3 - The cavity's resonance as revealed by RL (dip) and IL (peak) may not line up. 
     Use the center of the RL dip for the actual pass resonant frequency on your repeater's transmit and receive frequencies.
4 - When you measure IL - do it at your repeater frequencies - even though there may be a frequency along the graph that shows less loss.
     Do this because of #3.

#3 is a big one on VHF duplexers.  I can't stress enough that you don't tune a cavity's pass frequency to the frequency that provides the lowest insertion loss.  That's not the pass frequency - - - the frequency that provides the best return loss is the 'real' pass frequency.  Obviously then - when measuring insertion loss - do it at the real pass frequency - not the frequency that shows the lowest insertion loss. 

Tuning tips:
1 - Set all coupling loops for the same IL - while keeping the pass to notch frequency delta close to what you need.
2 - Couple the cavities together and verify that the IL is the sum of all cavities plus a little for cable / connector loss.
3 - Verify that you have at least -20 dB of RL on each of the pass frequencies (more is better).
4 - Do NOT be tempted to retune the pass frequencies or coupling loop rotations even though you believe you are making the tuning 'better'.

In summary - What's important is that you have 20dB or more return loss on your pass frequencies (more is better).  Also verify the insertion loss is the simple math sum of the cavities in series - plus a tiny bit for cable and connector losses.  Two cavities set for 0.6dB IL should work out to less than 1.5dB total insertion loss. 

After all of that is set correctly, cabled together, and verified - then tune the notches for the most rejection. 

If you started out with each cavity showing 30+ dB of return loss - when cabled together, I'd expect to see 25dB or so when completely cabled together, the notches properly tuned, and the opposite port properly terminated.  Some degradation of the return loss amount as compared to a single cavity is to be expected.

Kevin W3KKC


On 12/24/2024 1:49 PM, Jeremy Hansen via groups.io wrote:

Merry Christmas to all in the group! Blessings and best wishes to all.

 

I'll start with a short story which almost resulted in me posing an entirely different question to the group. I spent hours in the shop last night playing with a set of TX/RX 220mhz duplexer I recently acquired (odd, it has two model numbers listed on the tag - 43-54A-02479-A & 26-54A-02479A), trying to line them up on a 222/223 amateur pair. Following the manufacturer instructions, I would continuously come up with final results with the passes and notches off frequency by roughly 100khz. Current test equipment is an Aeroflex 3920 and Eagle RLB150x3 return loss bridge. After tuning with these instruments, I've gotten into the habit of using my RigExpert AA-1500 to verify proper tuning right before installation, with very good results - I find it to be quite accurate. Fast forward to this morning after a few fresh cups of coffee and at a complete loss for the continued disagreement, I remember I have another RLB150 and decide to swap it out. Bingo - a much deeper notch on the scope and now in agreement with my reference measurement. I guess it's a good reminder to always test your accessories before embarking upon these journeys. 

 

Now, on to my next question. With regard to pass tuning, I now can tune each can individually achieving acceptable RL (30+ db). When paired back together, the passes remain in alignment and present an acceptable RL (better than 15bd) but can be improved (closer to 30bd) by additional tuning rod adjustment. After this fine adjustment, I went back and swept both cans individually and found that it moved the passes pretty significantly, and now presented an unacceptable RL around 12 db. 

 

My question is: should I tune for highest RL with each cavity in cascade or accept the lower but acceptable RL achieved by tuning the passes individually then placing them in cascade with little/no further adjustment. 

 

Merry Christmas,

Jeremy K1LFK

Kevin,

As Gary mentioned, I plan to keep a hard copy of your detailed response. Thanks for the gift of sharing some very valuable knowledge.

I've read it over twice and will likely give it a third look today. I'm 90% tracking the theory, with my brain trying to wrap itself around the last 10% (mostly the comment about it being "ok"  for the RL notches not to line up after placing cavities in cascade. I'm getting there, especially with the help of people willing to share their knowledge and experience. Sincerely, thank you.

John, I'm also appreciative of your additional questions and commentary. Like you, I'm using a SA/TG to make these adjustments. Kevin's response is almost enough to push me into finally breaking down and purchasing a decent VNA (I've been following that thread too....)

Back to gifts and family time for now - more questions to follow.

Merry Christmas!
Jeremy K1LFK

On Wed, Dec 25, 2024, 9:39 AM Gary Cook via groups.io <gac=probemail.com@groups.io> wrote:

Thank you, Kevin (I printed your reply). And, Merry Christmas.

Gary

On 12/24/2024 7:29 PM, Kevin Custer wrote:

Merry Christmas Jeremy (and others).

I don't have an answer or comment about your first question, but it may be tied to what's in the answer for the second question.

As to your second question - a wise man (Jeff / WN3A) once told me, "Don't use your duplexer as an antenna tuner".  If you don't know what that means - let me explain....   I apologize, this is long.

When you tune a single cavity for resonance using return loss - it's correctly tuned.  When you add additional cavities in series (that have all been tuned correctly, and when the phasing cables are the correct length) - the tuning of the resonance doesn't change - therefore - you shouldn't need to re-peak anything.  If you do 'touch them up' - you can be compensating for imaginary problems and/or inducing new ones.  As such - it may look like you are making it better - when in reality - you are making it worse.  The imaginary problems can be a less than perfect calibration load, less than prefect calibration of the VNA, and less than perfect test cables.  It can also be the desire to make the return loss 'dips' align on top of one another - which there is no reason to do so. 

After cabling the duplexer together, re-adjusting the pass frequency adjustments may lead you to believe your making an improvement - because the display "looks better" - but is it really better?  The answer is - probably not, and more so - very likely not.  What you may be doing is compensating for imperfection where two mismatches end up looking better on the display as a (conjugate match).  This is the same effect as what a transmatch does to a non resonant antenna - only in this case, you're taking something that has already been tuned to resonance and you are mis-aligning of an apparent improvement - when in reality, there is none. 

I've seen YouTube videos that say to realign after re-cabling - and I don't agree with them.  A tiny bit of readjustment can add symmetry, but if you're looking at S11/S22 and S21/S12 anyway, you're going to be okay.

Remember:
1 - When you add cavities to the duplexer leg, the return loss bandwidth widens. 
     In other words - the RL "dips" may not (likely won't) line up on top of one another.  This is NOT necessarily a bad thing - and can be a good thing.
2 - When adding cavities to the leg, the return loss graph will become humpy (have ripple), and that's okay too. 
     Nothing says the return loss "dip" response has to have a nice smooth bottom to it.
3 - The cavity's resonance as revealed by RL (dip) and IL (peak) may not line up. 
     Use the center of the RL dip for the actual pass resonant frequency on your repeater's transmit and receive frequencies.
4 - When you measure IL - do it at your repeater frequencies - even though there may be a frequency along the graph that shows less loss.
     Do this because of #3.

#3 is a big one on VHF duplexers.  I can't stress enough that you don't tune a cavity's pass frequency to the frequency that provides the lowest insertion loss.  That's not the pass frequency - - - the frequency that provides the best return loss is the 'real' pass frequency.  Obviously then - when measuring insertion loss - do it at the real pass frequency - not the frequency that shows the lowest insertion loss. 

Tuning tips:
1 - Set all coupling loops for the same IL - while keeping the pass to notch frequency delta close to what you need.
2 - Couple the cavities together and verify that the IL is the sum of all cavities plus a little for cable / connector loss.
3 - Verify that you have at least -20 dB of RL on each of the pass frequencies (more is better).
4 - Do NOT be tempted to retune the pass frequencies or coupling loop rotations even though you believe you are making the tuning 'better'.

In summary - What's important is that you have 20dB or more return loss on your pass frequencies (more is better).  Also verify the insertion loss is the simple math sum of the cavities in series - plus a tiny bit for cable and connector losses.  Two cavities set for 0.6dB IL should work out to less than 1.5dB total insertion loss. 

After all of that is set correctly, cabled together, and verified - then tune the notches for the most rejection. 

If you started out with each cavity showing 30+ dB of return loss - when cabled together, I'd expect to see 25dB or so when completely cabled together, the notches properly tuned, and the opposite port properly terminated.  Some degradation of the return loss amount as compared to a single cavity is to be expected.

Kevin W3KKC


On 12/24/2024 1:49 PM, Jeremy Hansen via groups.io wrote:

Merry Christmas to all in the group! Blessings and best wishes to all.

 

I'll start with a short story which almost resulted in me posing an entirely different question to the group. I spent hours in the shop last night playing with a set of TX/RX 220mhz duplexer I recently acquired (odd, it has two model numbers listed on the tag - 43-54A-02479-A & 26-54A-02479A), trying to line them up on a 222/223 amateur pair. Following the manufacturer instructions, I would continuously come up with final results with the passes and notches off frequency by roughly 100khz. Current test equipment is an Aeroflex 3920 and Eagle RLB150x3 return loss bridge. After tuning with these instruments, I've gotten into the habit of using my RigExpert AA-1500 to verify proper tuning right before installation, with very good results - I find it to be quite accurate. Fast forward to this morning after a few fresh cups of coffee and at a complete loss for the continued disagreement, I remember I have another RLB150 and decide to swap it out. Bingo - a much deeper notch on the scope and now in agreement with my reference measurement. I guess it's a good reminder to always test your accessories before embarking upon these journeys. 

 

Now, on to my next question. With regard to pass tuning, I now can tune each can individually achieving acceptable RL (30+ db). When paired back together, the passes remain in alignment and present an acceptable RL (better than 15bd) but can be improved (closer to 30bd) by additional tuning rod adjustment. After this fine adjustment, I went back and swept both cans individually and found that it moved the passes pretty significantly, and now presented an unacceptable RL around 12 db. 

 

My question is: should I tune for highest RL with each cavity in cascade or accept the lower but acceptable RL achieved by tuning the passes individually then placing them in cascade with little/no further adjustment. 

 

Merry Christmas,

Jeremy K1LFK

 

 

Hi,

 

I have a set of six six inch RFS cavities 3rd overtone for 470MHz so tune

magnificently on 147MHz.

These have access ports too small for N connectors but TNC are OK.

 

I have made two TX-RX style coupling loops. 

 

These cavities being for UHF, I have no inter-cavity cables.

 

I have to make EVERYTHING and this is common.

The cables get "stolen" for the connectors.

 

Can we have a good discussion on "tuning" the inter-cavity lines please?

 

Alan VK2ZIW

 

 On Thu, 26 Dec 2024 16:35:09 -0500, Kevin Custer wrote
> On 12/26/2024 1:11 PM, Bob Dengler wrote:
> > At 12/25/2024 10:43 AM, you wrote:
> >
> >> Also take note to the violet graph in page two.  This shows the IL of
> >> the cascaded filters.  Note that from marker 2 to the right
> >> (increasing frequency) the loss gets increasingly less, likely beyond
> >> where the display stops.  I'd love for someone to tell me they could
> >> tune the pass frequency of this duplexer with a SA/TG looking at
> >> response only.
> >
> > Well, not with the vertical resolution set to 10 dB/div.
> >
> >> There's no way - it's impossible.  This shows that the cascaded
> >> cavity's resonance is not the frequency that revels the lowest
> >> insertion loss.  While there are duplexers that S11 and S21 align,
> >> this PD497 isn't one of them.
> >
> > Not disagreeing with the above, but...  how does one rationalize the
> > above with this PD-497 tuning instructions document from the manufacturer?
> >
> > <https://www.repeater-builder.com/antenna/phelps-dodge/pd-497-duplexer-instructions.pdf
> > >
> >
> > Granted PD doesn't specify a tracking gen., but they do instruct you
> > to tune the passes for max. signal.  If that method doesn't work why
> > is PD telling people to tune it that way?  I'm not saying this is the
> > preferred method - you'd rightfully point out that VNAs weren't
> > commonly available when that document was written.
> >
> > Bob NO6B
>
> Bob,
>
> You'd need to ask the manufacturer why they choose to tell the customer
> the wrong way to tune the duplexer.  It's as stupid as WACOM telling
> people to cut the TX cable to a "special" length between the TX and
> duplexer when the power loss is illogical.  At least we know better and
> can instruct people the right way.  Plus - now with amateur grade VNA's
> and RLB's that cost less than $100 - tuning by some any other method
> doesn't make any sense.
>
> I believe the manufacturer never used the instructions above for
> actually tuning duplexers that left the factory.  There have been ways
> to measure and calculate return loss since at least the invention of the
> "Smith Chart" by Phillip H Smith in 1939.  You don't need a VNA to
> measure return loss - a return loss bridge is adequate and that
> instrument has been available since well before the PD497 was
> conceived.  The problem is these instruments are not well understood -
> and that's what I'm attempting to circumvent. There are plenty of
> mis-tuned duplexers out there - but there shouldn't be - at least not today.
>
> And - even with the vertical resolution set to .5dB per division, it's
> still an increasing slope for hundreds of kHz, and that trace doesn't
> (ever) reveal the actual pass frequency concurrent with best return loss
> - never never never ever.  How do I know - let's look at another PD497
> retuned by Jeff.  This duplexer was expertly (ahem) tuned by a
> 'professional' two-way shop on a SA/TG.  The document is pretty self
> explanatory, but I'll include the email that Jeff wrote to the
> individual - and I was copied.  I'm removing the persons sensitive
> information.  Jeff's reply to the individual has many tips.  Like where
> the duplexer was tuned to a situation I call "short spaced" - a
> situation where reduced notch depths occur because the notches are too
> close to the cavities 'real' pass frequency.
>
> For you Bob, pay attention to pages 2 & 4 (S23 & S32).  Yes - it was
> tuned with a 3-port S-Parameter VNA - perfect for these common duplexers.
>
> Kevin W3KKC
>
> ------------------------------------------------------------------------
>
> (Person) (and copying Kevin as I know he likes to work on PD497's too):
>
> Retuned your duplexer.  As suspected, the previous tuning appears to have been done for minimum insertion loss at the pass frequency, rather than tuning each cavity individually with the pass tuned for maximum return loss. When you tune a PD497 (and many other duplexers as well, especially on highband) for minimum passband loss, especially when doing it with all cavities connected together rather than individually, the adverse effects are typically:
>
> a) Return loss is not optimal at the pass frequency, which results in a lot of interaction between cavities when connected in cascade.
>
> b) The notch depth is degraded because it is tuned "too close" to the anti-resonance bandpass frequency, resulting in overall reduced notch depth/isolation.
>
> c) The bandpass response of the resonator is degraded (it behaves more like a reject-only filter rather than pass/reject).  This is easy to see when you do an ultimate isolation measurement (i.e. from Tx port to Rx port with antenna port terminated) - the mid-band isolation goes down the tubes.  In the instant case, the mid-band isolation was originally only 11 dB, but shot up to 37 dB after retuning.  Like I said, the cavities behave more like notch-only rather than pass-notch due to this effect, so the mid-band isolation virtually disappears.  With only 11 dB of mid-band isolation, I would think this was your biggest cause of desense, with the degraded notch depth on the Tx leg (i.e. insufficient transmitter noise suppression) being a close second.
>
> I did make two changes to the duplexer.  First, the antenna tee was contributing a bit over a tenth of a dB of additional loss.  I couldn't find a manufacturer's part number on it, but eyeballing it, it appears to be from off-shore.  I replaced it with a real-deal Amphenol tee (F-M-F) and Amphenol barrel (both new) and the extra loss went away.  I don't have any Amphenol F-F-F tees, and for that matter, I don't think I've ever seen any.   I  guess if you were real ambitious you could replace the cables to the tee with new ones with PL-259's on one end and type N on the other end and then use a type N F-F-F tee, with the appropriate length corrections as   required...the exercise is left to the reader.
>
> Second, the insertion loss of the last cavity (i.e. furthest from the tee) on the receive side has about a tenth of a dB more loss than the others.  I removed the loop assembly, re-did the solder joints, even pulled out the capacitor "rod" and burnished it, with only minor improvement.   When measured outside the cavity, there is no excessive loss in the loop assembly, so I have to conclude that the resonator itself is a bit lossy, maybe had taken on some moisture or something at some point in time.  Given the age of the cavities (I'm guessing close to 40 years), I'm not going to sweat an extra tenth of a dB or two since the duplexer meets spec in all other aspects.
>
> Plots attached (both "before" and "after" measurements).   I apologize - when I swept the duplexer before retuning I had the VNA set for a 2 MHz wide sweep centered on 145 MHz, but when I swept it after retuning I had the VNA set for a 1 MHz wide sweep...
>
>                     --- Jeff WN3A
>
>


---------------------------------------------------
Alan VK2ZIW
Before the Big Bang, God, Sela.
OpenWebMail 2.53, nothing in the cloud.

Hi,

I have a set of six six inch RFS cavities 3rd overtone for 470MHz so tune

magnificently on 147MHz.

These have access ports too small for N connectors but TNC are OK.

I have made two TX-RX style coupling loops. 

These cavities being for UHF, I have no inter-cavity cables.

I have to make EVERYTHING and this is common.

The cables get "stolen" for the connectors.

Can we have a good discussion on "tuning" the inter-cavity lines please?

Alan VK2ZIW

On Friday, December 27, 2024, 11:24 AM, Kevin Custer <kuggie@...> wrote:

Bob,

You'd need to ask the manufacturer why they choose to tell the customer the wrong way to tune the duplexer.

Well, I reject the position that tuning the duplexer per the manufacturer's instructions is "wrong", so there is nothing to ask.

Okay -  there's nothing further to discuss.

And - since you refuse to believe the theory that has been presented by myself and Jeff - any further discussion is a waste of our time.

Kevin W3KKC