|
Will a pad give the nanovna more of a dynamic range to see? the notch depth? Iirc the dynamic range of a standard nvna is about 60 to 80db. So if I put a 10 or 15db pad that should help with tuning the notches?
Pads can be helpful on the generate side of a VNA providing a good resistive match to the device under test. Many years ago working at Hughes Aircraft we always terminated our generate cables with a 6dB pad.?
|
At 4/24/2025 08:14 AM, you wrote:
Pads can be helpful on the
generate side of a VNA providing a good resistive match to the device
under test. Many years ago working at Hughes Aircraft we always
terminated our generate cables with a 6dB pad.
In all the years I've used VNAs I've never done this except when the
nominal test port power (typically 0 dBm for an HP8510) would drive the
DUT into nonlinear operation.? The error correction applied
following calibration is supposed to take care of port match
issues.? Now if you're using something like an HP8410 w/o error
correction then I agree with using pads.
Bob NO6B
|
It will have the opposite effect.? To get more dynamic range you would really need an amplifier or a way to drive in more signal.? The pads are just to match the impedance as the output/input terminals may not be the nominal 50 ohms.? That is just for the notches.? If tuning the pass with a RLB do not use a pad as this will give a false match/very low RL .
Ralph ku4pt
On Thursday, April 24, 2025 at 11:28:09 AM EDT, Part 15 Engineer via groups.io <kc8gpd@...> wrote:
Will a pad give the nanovna more of a dynamic range to see? the notch depth? Iirc the dynamic range of a standard nvna is about 60 to 80db. So if I put a 10 or 15db pad that should help with tuning the notches?
Pads can be helpful on the generate side of a VNA providing a good resistive match to the device under test. Many years ago working at Hughes Aircraft we always terminated our generate cables with a 6dB pad.?
_._,_._,_
|
I think I finally got the pd526 tuned. It has 1db difference in sensitivity between direct input and going through duplexer and very little tx power loss coming out of duplexer. Probably only a few watts of loss. Tomorrow I will do some field testing to see if it gets put as far as the txrx 28-70-15h duplexer I was using. I'm not seeing any signs of desense yet.
It will have the opposite effect.? To get more dynamic range you would really need an amplifier or a way to drive in more signal.? The pads are just to match the impedance as the output/input terminals may not be the nominal 50 ohms.? That is just for the notches.? If tuning the pass with a RLB do not use a pad as this will give a false match/very low RL .
Ralph ku4pt
On Thursday, April 24, 2025 at 11:28:09 AM EDT, Part 15 Engineer via <kc8gpd= [email protected]> wrote:
Will a pad give the nanovna more of a dynamic range to see? the notch depth? Iirc the dynamic range of a standard nvna is about 60 to 80db. So if I put a 10 or 15db pad that should help with tuning the notches?
Pads can be helpful on the generate side of a VNA providing a good resistive match to the device under test. Many years ago working at Hughes Aircraft we always terminated our generate cables with a 6dB pad.?
|
Ok, I'm tuning the pd526 on the nvna. I can't seek to get everything to line up on both reject and pass simultaneously. It seems to be one or the other. I have attached pic of my latest plot. This is the best I got it so far. I did tune cavities individually then hook them together. And I got multiple peaks and valleys on the plot instead on one nice peak on the pass and valley on the reject. Does anyone have a link to the plot for the pd526 and what it should look?like?
Thanks 
I think I finally got the pd526 tuned. It has 1db difference in sensitivity between direct input and going through duplexer and very little tx power loss coming out of duplexer. Probably only a few watts of loss. Tomorrow I will do some field testing to see if it gets put as far as the txrx 28-70-15h duplexer I was using. I'm not seeing any signs of desense yet.
It will have the opposite effect.? To get more dynamic range you would really need an amplifier or a way to drive in more signal.? The pads are just to match the impedance as the output/input terminals may not be the nominal 50 ohms.? That is just for the notches.? If tuning the pass with a RLB do not use a pad as this will give a false match/very low RL .
Ralph ku4pt
On Thursday, April 24, 2025 at 11:28:09 AM EDT, Part 15 Engineer via <kc8gpd= [email protected]> wrote:
Will a pad give the nanovna more of a dynamic range to see? the notch depth? Iirc the dynamic range of a standard nvna is about 60 to 80db. So if I put a 10 or 15db pad that should help with tuning the notches?
Pads can be helpful on the generate side of a VNA providing a good resistive match to the device under test. Many years ago working at Hughes Aircraft we always terminated our generate cables with a 6dB pad.?
|
Here is the plot for a single cavity.
Ok, I'm tuning the pd526 on the nvna. I can't seek to get everything to line up on both reject and pass simultaneously. It seems to be one or the other. I have attached pic of my latest plot. This is the best I got it so far. I did tune cavities individually then hook them together. And I got multiple peaks and valleys on the plot instead on one nice peak on the pass and valley on the reject. Does anyone have a link to the plot for the pd526 and what it should look?like?
Thanks 
I think I finally got the pd526 tuned. It has 1db difference in sensitivity between direct input and going through duplexer and very little tx power loss coming out of duplexer. Probably only a few watts of loss. Tomorrow I will do some field testing to see if it gets put as far as the txrx 28-70-15h duplexer I was using. I'm not seeing any signs of desense yet.
It will have the opposite effect.? To get more dynamic range you would really need an amplifier or a way to drive in more signal.? The pads are just to match the impedance as the output/input terminals may not be the nominal 50 ohms.? That is just for the notches.? If tuning the pass with a RLB do not use a pad as this will give a false match/very low RL .
Ralph ku4pt
On Thursday, April 24, 2025 at 11:28:09 AM EDT, Part 15 Engineer via <kc8gpd= [email protected]> wrote:
Will a pad give the nanovna more of a dynamic range to see? the notch depth? Iirc the dynamic range of a standard nvna is about 60 to 80db. So if I put a 10 or 15db pad that should help with tuning the notches?
Pads can be helpful on the generate side of a VNA providing a good resistive match to the device under test. Many years ago working at Hughes Aircraft we always terminated our generate cables with a 6dB pad.?
|
Your picture tells a lot.? My guess is that the cans on the same side of the duplexer are over coupled because of improper coax length between the cans. In other words, with the proper 1/4 wavelength , a short on one end creates very high impedance on the other end which gives the most isolation or decoupling between cans on the same side of the duplexer. It may only be off less than 1 inch length at UHF.
Since the best return loss perfectly aligns with least insertion loss, the length of coax between the antenna T and the opposite side cans is the proper length, so that is good.
?
John
On 04/26/2025 6:51 AM EDT Part 15 Engineer via groups.io <kc8gpd@...> wrote:
?
?
Ok, I'm tuning the pd526 on the nvna. I can't seek to get everything to line up on both reject and pass simultaneously. It seems to be one or the other. I have attached pic of my latest plot. This is the best I got it so far. I did tune cavities individually then hook them together. And I got multiple peaks and valleys on the plot instead on one nice peak on the pass and valley on the reject. Does anyone have a link to the plot for the pd526 and what it should look?like?
?
Thanks
I think I finally got the pd526 tuned. It has 1db difference in sensitivity between direct input and going through duplexer and very little tx power loss coming out of duplexer. Probably only a few watts of loss. Tomorrow I will do some field testing to see if it gets put as far as the txrx 28-70-15h duplexer I was using. I'm not seeing any signs of desense yet.
?
It will have the opposite effect.? To get more dynamic range you would really need an amplifier or a way to drive in more signal.? The pads are just to match the impedance as the output/input terminals may not be the nominal 50 ohms.? That is just for the notches.? If tuning the pass with a RLB do not use a pad as this will give a false match/very low RL .
?
Ralph ku4pt
?
?
On Thursday, April 24, 2025 at 11:28:09 AM EDT, Part 15 Engineer via <kc8gpd= [email protected]> wrote:
?
?
Will a pad give the nanovna more of a dynamic range to see? the notch depth? Iirc the dynamic range of a standard nvna is about 60 to 80db. So if I put a 10 or 15db pad that should help with tuning the notches?
Pads can be helpful on the generate side of a VNA providing a good resistive match to the device under test. Many years ago working at Hughes Aircraft we always terminated our generate cables with a 6dB pad.?
?
?
|
A number of things: ? - You¡¯re sweeping 30 MHz wide with only 401 points.? You need to either narrow down your sweep just to the range of interest, use more points, or more likely a combination of both.? I¡¯d typically be using a 10 MHz span and 1601 points (I don¡¯t have a NanoVNA, I don¡¯t know what it¡¯s capabilities are).
- Something is either very wrong with the cavity, or the calibration.? It is showing you have 3.129 of insertion loss in the bandpass; it should be about 0.25 dB for a properly-functioning PD526 resonator.? Likewise, at the notch frequency, it¡¯s showing that you have 4.77 dB of return loss ¨C it should be a lot closer to zero.? If I were to take a WAG, you have 3 dB or more of cable loss that isn¡¯t compensated for, and/or something is wrong with your calibration kit.
- No since in working on the full duplexer harnessed together until the above are first addressed and you know each cavity is properly tuned individually.
? --- Jeff WN3A ?
toggle quoted message
Show quoted text
From: [email protected] <[email protected]> On Behalf Of Part 15 Engineer via groups.io
Sent: Saturday, April 26, 2025 7:46 AM
To: [email protected]
Subject: Re: [repeater-builder] PD526 Duplexer Tuning? Here is the plot for a single cavity. ? Ok, I'm tuning the pd526 on the nvna. I can't seek to get everything to line up on both reject and pass simultaneously. It seems to be one or the other. I have attached pic of my latest plot. This is the best I got it so far. I did tune cavities individually then hook them together. And I got multiple peaks and valleys on the plot instead on one nice peak on the pass and valley on the reject. Does anyone have a link to the plot for the pd526 and what it should look?like? Thanks
 ? I think I finally got the pd526 tuned. It has 1db difference in sensitivity between direct input and going through duplexer and very little tx power loss coming out of duplexer. Probably only a few watts of loss. Tomorrow I will do some field testing to see if it gets put as far as the txrx 28-70-15h duplexer I was using. I'm not seeing any signs of desense yet. ? It will have the opposite effect.? To get more dynamic range you would really need an amplifier or a way to drive in more signal.? The pads are just to match the impedance as the output/input terminals may not be the nominal 50 ohms.? That is just for the notches.? If tuning the pass with a RLB do not use a pad as this will give a false match/very low RL . On Thursday, April 24, 2025 at 11:28:09 AM EDT, Part 15 Engineer via <kc8gpd=[email protected]> wrote: Will a pad give the nanovna more of a dynamic range to see? the notch depth? Iirc the dynamic range of a standard nvna is about 60 to 80db. So if I put a 10 or 15db pad that should help with tuning the notches? ? Pads can be helpful on the generate side of a VNA providing a good resistive match to the device under test. Many years ago working at Hughes Aircraft we always terminated our generate cables with a 6dB pad.?
|
Ignore the first two images close to the top. They are earlier plots.
On Sat, Apr 26, 2025, 11:33?AM Part 15 Engineer < kc8gpd@...> wrote: How does this look? It's a nanovna and nit a calibrated to nist piece of equipment so it's not going to be dead on even when you go through the calibration procedure for the nanovna.
I think this is the best it's going to get.
A number of things: ? - You¡¯re sweeping 30 MHz wide with only 401 points.? You need to either narrow down your sweep just to the range of interest, use more points, or more likely a combination of both.? I¡¯d typically be using a 10 MHz span and 1601 points (I don¡¯t have a NanoVNA, I don¡¯t know what it¡¯s capabilities are).
- Something is either very wrong with the cavity, or the calibration.? It is showing you have 3.129 of insertion loss in the bandpass; it should be about 0.25 dB for a properly-functioning PD526 resonator.? Likewise, at the notch frequency, it¡¯s showing that you have 4.77 dB of return loss ¨C it should be a lot closer to zero.? If I were to take a WAG, you have 3 dB or more of cable loss that isn¡¯t compensated for, and/or something is wrong with your calibration kit.
- No since in working on the full duplexer harnessed together until the above are first addressed and you know each cavity is properly tuned individually.
? --- Jeff WN3A ? ? Here is the plot for a single cavity. ? Ok, I'm tuning the pd526 on the nvna. I can't seek to get everything to line up on both reject and pass simultaneously. It seems to be one or the other. I have attached pic of my latest plot. This is the best I got it so far. I did tune cavities individually then hook them together. And I got multiple peaks and valleys on the plot instead on one nice peak on the pass and valley on the reject. Does anyone have a link to the plot for the pd526 and what it should look?like? Thanks
 ? I think I finally got the pd526 tuned. It has 1db difference in sensitivity between direct input and going through duplexer and very little tx power loss coming out of duplexer. Probably only a few watts of loss. Tomorrow I will do some field testing to see if it gets put as far as the txrx 28-70-15h duplexer I was using. I'm not seeing any signs of desense yet. ? It will have the opposite effect.? To get more dynamic range you would really need an amplifier or a way to drive in more signal.? The pads are just to match the impedance as the output/input terminals may not be the nominal 50 ohms.? That is just for the notches.? If tuning the pass with a RLB do not use a pad as this will give a false match/very low RL . On Thursday, April 24, 2025 at 11:28:09 AM EDT, Part 15 Engineer via <kc8gpd=[email protected]> wrote: Will a pad give the nanovna more of a dynamic range to see? the notch depth? Iirc the dynamic range of a standard nvna is about 60 to 80db. So if I put a 10 or 15db pad that should help with tuning the notches? ? Pads can be helpful on the generate side of a VNA providing a good resistive match to the device under test. Many years ago working at Hughes Aircraft we always terminated our generate cables with a 6dB pad.?
|
The first two plots are typical of each individual cavity. In this case the high side. The second two are with the two coupling loops in line on the 3 high side cavities minus the coupling tee. So I'm guessing like you guys stated earlier it's the coupling cables? I didn't tune the cavities anymore. Just tuned them individually and installed the cables and this is from input to output with two coupling loops installed. Does anyone have a good set of harness cables? I bought these from an individual on here but apparently they are not the right cables?
Thanks
Robert
A number of things: ? - You¡¯re sweeping 30 MHz wide with only 401 points.? You need to either narrow down your sweep just to the range of interest, use more points, or more likely a combination of both.? I¡¯d typically be using a 10 MHz span and 1601 points (I don¡¯t have a NanoVNA, I don¡¯t know what it¡¯s capabilities are).
- Something is either very wrong with the cavity, or the calibration.? It is showing you have 3.129 of insertion loss in the bandpass; it should be about 0.25 dB for a properly-functioning PD526 resonator.? Likewise, at the notch frequency, it¡¯s showing that you have 4.77 dB of return loss ¨C it should be a lot closer to zero.? If I were to take a WAG, you have 3 dB or more of cable loss that isn¡¯t compensated for, and/or something is wrong with your calibration kit.
- No since in working on the full duplexer harnessed together until the above are first addressed and you know each cavity is properly tuned individually.
? --- Jeff WN3A ? ? Here is the plot for a single cavity. ? Ok, I'm tuning the pd526 on the nvna. I can't seek to get everything to line up on both reject and pass simultaneously. It seems to be one or the other. I have attached pic of my latest plot. This is the best I got it so far. I did tune cavities individually then hook them together. And I got multiple peaks and valleys on the plot instead on one nice peak on the pass and valley on the reject. Does anyone have a link to the plot for the pd526 and what it should look?like? Thanks
 ? I think I finally got the pd526 tuned. It has 1db difference in sensitivity between direct input and going through duplexer and very little tx power loss coming out of duplexer. Probably only a few watts of loss. Tomorrow I will do some field testing to see if it gets put as far as the txrx 28-70-15h duplexer I was using. I'm not seeing any signs of desense yet. ? It will have the opposite effect.? To get more dynamic range you would really need an amplifier or a way to drive in more signal.? The pads are just to match the impedance as the output/input terminals may not be the nominal 50 ohms.? That is just for the notches.? If tuning the pass with a RLB do not use a pad as this will give a false match/very low RL . On Thursday, April 24, 2025 at 11:28:09 AM EDT, Part 15 Engineer via <kc8gpd=[email protected]> wrote: Will a pad give the nanovna more of a dynamic range to see? the notch depth? Iirc the dynamic range of a standard nvna is about 60 to 80db. So if I put a 10 or 15db pad that should help with tuning the notches? ? Pads can be helpful on the generate side of a VNA providing a good resistive match to the device under test. Many years ago working at Hughes Aircraft we always terminated our generate cables with a 6dB pad.?
|
You got the cables from me, and they are decidedly the correct cables.? The four 11.5¡± cables go between cavities, the 12¡± cables go to the triple-female tee that I sent. ? It has to be something wrong with either the test equipment, testcables/cal kit, or methodology.? Or you have one or more bad resonators in the duplexer itself. ? In your prior post you said to ¡°ignore the first two plots¡±, but I only see two plots in the email that preceded that, so I don¡¯t know what I should be looking at. ? In your-most recent email, if there were any plots attached, they didn¡¯t come through ¨C there is a box with an X that says ¡°The linked image cannot be displayed.? The file may have been moved, renamed, or deleted.? Verify that the link points to the correct file location.¡±? Don¡¯t send links, don¡¯t embed images HTML, send the sweep images as attachments. ? Let¡¯s go back to step 1.? ? - ?Set up the VNA correctly (again, I don¡¯t have a NanoVNA, so you¡¯ll have to figure out how to do this):
- 10 MHz span centered roughly mid-way between your Tx and Rx frequencies at some whole-MHz value
- 1601 data points
- A reasonably low bandwidth (10 kHz or less)
- No averaging, no smoothing
- Do a full 2-port calibration using your test cables and cal kit OSL, no adapters
- At the end of cabliration, send a sweep of what your VNA shows with the test cables connected to each other with the thru adapter (presumably an N-female ¡°barrel¡±)
- Tune one of the high-pass cavities:
- Tune the pass for best return loss (S11), ignore insertion loss (S21).
- Tune the notch (S21).? The pass shouldn¡¯t have moved if the notch was close to being correct already, but if it did move a bit, repeat a) and b).
- Send the sweep
- Tune one of the low-pass cavities per #2 above and send the sweep.
- Tune the remaining four cavities individually, and confirm that each one¡¯s sweep looks virtually identical to the sweeps from #2 and #3.? A single cavity should have right around 30 dB return loss and 0.25 dB insertion loss for the pass (and again, you tune the pass for maximum return loss, not least insertion loss although the two will be close).? The notch should be 40 dB +/- 2 dB at a 5 MHz split.? If any resonator doesn¡¯t meet those specs then it has a problem that needs to be addressed.
- Don¡¯t bother harnessing it up until we successfully get through the above.
? --- Jeff WN3A ?
|
I measured the cable lengths and it's rg214 and it's approx 11.5 in for jumpers but they might be a hair over 11.5 inches. I'm not sure how critical the length is or how exact it must be.
The first two plots are typical of each individual cavity. In this case the high side. The second two are with the two coupling loops in line on the 3 high side cavities minus the coupling tee. So I'm guessing like you guys stated earlier it's the coupling cables? I didn't tune the cavities anymore. Just tuned them individually and installed the cables and this is from input to output with two coupling loops installed. Does anyone have a good set of harness cables? I bought these from an individual on here but apparently they are not the right cables?
Thanks
Robert
A number of things: ? - You¡¯re sweeping 30 MHz wide with only 401 points.? You need to either narrow down your sweep just to the range of interest, use more points, or more likely a combination of both.? I¡¯d typically be using a 10 MHz span and 1601 points (I don¡¯t have a NanoVNA, I don¡¯t know what it¡¯s capabilities are).
- Something is either very wrong with the cavity, or the calibration.? It is showing you have 3.129 of insertion loss in the bandpass; it should be about 0.25 dB for a properly-functioning PD526 resonator.? Likewise, at the notch frequency, it¡¯s showing that you have 4.77 dB of return loss ¨C it should be a lot closer to zero.? If I were to take a WAG, you have 3 dB or more of cable loss that isn¡¯t compensated for, and/or something is wrong with your calibration kit.
- No since in working on the full duplexer harnessed together until the above are first addressed and you know each cavity is properly tuned individually.
? --- Jeff WN3A ? ? Here is the plot for a single cavity. ? Ok, I'm tuning the pd526 on the nvna. I can't seek to get everything to line up on both reject and pass simultaneously. It seems to be one or the other. I have attached pic of my latest plot. This is the best I got it so far. I did tune cavities individually then hook them together. And I got multiple peaks and valleys on the plot instead on one nice peak on the pass and valley on the reject. Does anyone have a link to the plot for the pd526 and what it should look?like? Thanks
![]() ? I think I finally got the pd526 tuned. It has 1db difference in sensitivity between direct input and going through duplexer and very little tx power loss coming out of duplexer. Probably only a few watts of loss. Tomorrow I will do some field testing to see if it gets put as far as the txrx 28-70-15h duplexer I was using. I'm not seeing any signs of desense yet. ? It will have the opposite effect.? To get more dynamic range you would really need an amplifier or a way to drive in more signal.? The pads are just to match the impedance as the output/input terminals may not be the nominal 50 ohms.? That is just for the notches.? If tuning the pass with a RLB do not use a pad as this will give a false match/very low RL . On Thursday, April 24, 2025 at 11:28:09 AM EDT, Part 15 Engineer via <kc8gpd=[email protected]> wrote: Will a pad give the nanovna more of a dynamic range to see? the notch depth? Iirc the dynamic range of a standard nvna is about 60 to 80db. So if I put a 10 or 15db pad that should help with tuning the notches? ? Pads can be helpful on the generate side of a VNA providing a good resistive match to the device under test. Many years ago working at Hughes Aircraft we always terminated our generate cables with a 6dB pad.?
|
It's only capable of 1024 points max. I think everything else I can set to your specs.
On Sun, Apr 27, 2025, 8:41?AM Part 15 Engineer < kc8gpd@...> wrote: I measured the cable lengths and it's rg214 and it's approx 11.5 in for jumpers but they might be a hair over 11.5 inches. I'm not sure how critical the length is or how exact it must be.
The first two plots are typical of each individual cavity. In this case the high side. The second two are with the two coupling loops in line on the 3 high side cavities minus the coupling tee. So I'm guessing like you guys stated earlier it's the coupling cables? I didn't tune the cavities anymore. Just tuned them individually and installed the cables and this is from input to output with two coupling loops installed. Does anyone have a good set of harness cables? I bought these from an individual on here but apparently they are not the right cables?
Thanks
Robert
A number of things: ? - You¡¯re sweeping 30 MHz wide with only 401 points.? You need to either narrow down your sweep just to the range of interest, use more points, or more likely a combination of both.? I¡¯d typically be using a 10 MHz span and 1601 points (I don¡¯t have a NanoVNA, I don¡¯t know what it¡¯s capabilities are).
- Something is either very wrong with the cavity, or the calibration.? It is showing you have 3.129 of insertion loss in the bandpass; it should be about 0.25 dB for a properly-functioning PD526 resonator.? Likewise, at the notch frequency, it¡¯s showing that you have 4.77 dB of return loss ¨C it should be a lot closer to zero.? If I were to take a WAG, you have 3 dB or more of cable loss that isn¡¯t compensated for, and/or something is wrong with your calibration kit.
- No since in working on the full duplexer harnessed together until the above are first addressed and you know each cavity is properly tuned individually.
? --- Jeff WN3A ? ? Here is the plot for a single cavity. ? Ok, I'm tuning the pd526 on the nvna. I can't seek to get everything to line up on both reject and pass simultaneously. It seems to be one or the other. I have attached pic of my latest plot. This is the best I got it so far. I did tune cavities individually then hook them together. And I got multiple peaks and valleys on the plot instead on one nice peak on the pass and valley on the reject. Does anyone have a link to the plot for the pd526 and what it should look?like? Thanks
![]() ? I think I finally got the pd526 tuned. It has 1db difference in sensitivity between direct input and going through duplexer and very little tx power loss coming out of duplexer. Probably only a few watts of loss. Tomorrow I will do some field testing to see if it gets put as far as the txrx 28-70-15h duplexer I was using. I'm not seeing any signs of desense yet. ? It will have the opposite effect.? To get more dynamic range you would really need an amplifier or a way to drive in more signal.? The pads are just to match the impedance as the output/input terminals may not be the nominal 50 ohms.? That is just for the notches.? If tuning the pass with a RLB do not use a pad as this will give a false match/very low RL . On Thursday, April 24, 2025 at 11:28:09 AM EDT, Part 15 Engineer via <kc8gpd=[email protected]> wrote: Will a pad give the nanovna more of a dynamic range to see? the notch depth? Iirc the dynamic range of a standard nvna is about 60 to 80db. So if I put a 10 or 15db pad that should help with tuning the notches? ? Pads can be helpful on the generate side of a VNA providing a good resistive match to the device under test. Many years ago working at Hughes Aircraft we always terminated our generate cables with a 6dB pad.?
|
They came off other PD526¡¯s; how close the factory made them to 11.5¡±, and how accurately you can measure them after having been deformed by the tight bend radius on a 526, I can¡¯t say.? But I can promise you that even if they were off half an inch that it wouldn¡¯t result in that specific duplexer not meeting spec. ? ??????????????????????????????????????????????????????????????????????????????????????????????? --- Jeff ?WN3A ?
toggle quoted message
Show quoted text
From: [email protected] <[email protected]> On Behalf Of Part 15 Engineer via groups.io
Sent: Sunday, April 27, 2025 10:42 AM
To: [email protected]
Subject: Re: [repeater-builder] PD526 Duplexer Tuning? I measured the cable lengths and it's rg214 and it's approx 11.5 in for jumpers but they might be a hair over 11.5 inches. I'm not sure how critical the length is or how exact it must be. ? The first two plots are typical of each individual cavity. In this case the high side. The second two are with the two coupling loops in line on the 3 high side cavities minus the coupling tee. So I'm guessing like you guys stated earlier it's the coupling cables? I didn't tune the cavities anymore. Just tuned them individually and installed the cables and this is from input to output with two coupling loops installed. Does anyone have a good set of harness cables? I bought these from an individual on here but apparently they are not the right cables? ? A number of things: ? - You¡¯re sweeping 30 MHz wide with only 401 points.? You need to either narrow down your sweep just to the range of interest, use more points, or more likely a combination of both.? I¡¯d typically be using a 10 MHz span and 1601 points (I don¡¯t have a NanoVNA, I don¡¯t know what it¡¯s capabilities are).
- Something is either very wrong with the cavity, or the calibration.? It is showing you have 3.129 of insertion loss in the bandpass; it should be about 0.25 dB for a properly-functioning PD526 resonator.? Likewise, at the notch frequency, it¡¯s showing that you have 4.77 dB of return loss ¨C it should be a lot closer to zero.? If I were to take a WAG, you have 3 dB or more of cable loss that isn¡¯t compensated for, and/or something is wrong with your calibration kit.
- No since in working on the full duplexer harnessed together until the above are first addressed and you know each cavity is properly tuned individually.
? --- Jeff WN3A ? ? Here is the plot for a single cavity. ? Ok, I'm tuning the pd526 on the nvna. I can't seek to get everything to line up on both reject and pass simultaneously. It seems to be one or the other. I have attached pic of my latest plot. This is the best I got it so far. I did tune cavities individually then hook them together. And I got multiple peaks and valleys on the plot instead on one nice peak on the pass and valley on the reject. Does anyone have a link to the plot for the pd526 and what it should look?like? ? I think I finally got the pd526 tuned. It has 1db difference in sensitivity between direct input and going through duplexer and very little tx power loss coming out of duplexer. Probably only a few watts of loss. Tomorrow I will do some field testing to see if it gets put as far as the txrx 28-70-15h duplexer I was using. I'm not seeing any signs of desense yet. ? It will have the opposite effect.? To get more dynamic range you would really need an amplifier or a way to drive in more signal.? The pads are just to match the impedance as the output/input terminals may not be the nominal 50 ohms.? That is just for the notches.? If tuning the pass with a RLB do not use a pad as this will give a false match/very low RL . On Thursday, April 24, 2025 at 11:28:09 AM EDT, Part 15 Engineer via <kc8gpd=[email protected]> wrote: Will a pad give the nanovna more of a dynamic range to see? the notch depth? Iirc the dynamic range of a standard nvna is about 60 to 80db. So if I put a 10 or 15db pad that should help with tuning the notches? ? Pads can be helpful on the generate side of a VNA providing a good resistive match to the device under test. Many years ago working at Hughes Aircraft we always terminated our generate cables with a 6dB pad.?
|
I ditto that. Every 526 I have tuned was done by individually tuning each cavity as per the instructions (there's more than one version of that) and then re-connect and each has been right on the money. First one in the 80's was done with a Measurements signal generator and Yaesu 708 portable as the receiver. Zero desense even with a GaAs FET preamp though it had its own preselector.
?
These are my preferred duplexers by far.?
GeorgeC W2DB
|
After hours of going at it. I have come to this point. I think it looks alot better, but still probably can be improved. 2 things. I can't set the nvna for more than 1024 points and the point rbw seems to unable to be changed from the 4000 hz setting. If someone knows how to change the rbw in nanovna app, immall ears. other points in the previous post were followed. I also added a 3db pad to the generator port and calibrated the nvna with the pad inline. Now I have both high and low sides looking similar. This is the high side. What should the total s21 gain be for each side approx.? I can't seem to get it lower than -0.8xx on each side. The s21 gain on reject averages about -70. The jumpers I'm using to tune will be the same ones used for the final installation. They are 2.0 meter new old stock 1/4 inch hardline with N connectors that came from cellular service. The nvna has sma to N connectors. I will follow up this post with pics of my test configuration.
I ditto that. Every 526 I have tuned was done by individually tuning each cavity as per the instructions (there's more than one version of that) and then re-connect and each has been right on the money. First one in the 80's was done with a Measurements signal generator and Yaesu 708 portable as the receiver. Zero desense even with a GaAs FET preamp though it had its own preselector.
?
These are my preferred duplexers by far.?
GeorgeC W2DB
|
On Mon, Apr 28, 2025, 4:41?AM Part 15 Engineer < kc8gpd@...> wrote: After hours of going at it. I have come to this point. I think it looks alot better, but still probably can be improved. 2 things. I can't set the nvna for more than 1024 points and the point rbw seems to unable to be changed from the 4000 hz setting. If someone knows how to change the rbw in nanovna app, immall ears. other points in the previous post were followed. I also added a 3db pad to the generator port and calibrated the nvna with the pad inline. Now I have both high and low sides looking similar. This is the high side. What should the total s21 gain be for each side approx.? I can't seem to get it lower than -0.8xx on each side. The s21 gain on reject averages about -70. The jumpers I'm using to tune will be the same ones used for the final installation. They are 2.0 meter new old stock 1/4 inch hardline with N connectors that came from cellular service. The nvna has sma to N connectors. I will follow up this post with pics of my test configuration.
I ditto that. Every 526 I have tuned was done by individually tuning each cavity as per the instructions (there's more than one version of that) and then re-connect and each has been right on the money. First one in the 80's was done with a Measurements signal generator and Yaesu 708 portable as the receiver. Zero desense even with a GaAs FET preamp though it had its own preselector.
?
These are my preferred duplexers by far.?
GeorgeC W2DB
|
You shouldn't use an attenuator pad with a VNA.? If you do - at least two things happen, depending on where your calibration plane exists:
1 - When measuring return loss (match), the RL value will be at least twice as much as the loss value of the pad.? Why?? The signal travels through the pad twice.? A 3dB pad will show a 6dB return loss even if the opposite port of the pad is shorted or open.? If you calibrated at the output of the pad - all you did was reduce the ability to measure S11 by 6dB.
2 - You reduce the dynamic range capability by the amount of the value of the pad.? You're already running the NanoVNA out of dynamic range, so the last thing you need is to reduce it even more.
VNA's are calibrated devices (when you do the calibration), and pads are not only unnecessary, but undesired.
0.8dB of insertion loss for a fully coupled PD526 is about right. You have no idea how much rejection you actually have because the NanoVNA is incapable of measuring the notch depths of a properly tuned PD526.? See how grassy the notches look, that's not measuring signal - that's revealing the noise floor of the instrument.? If you disconnected the cables from everything - you'd see the same type of response.? In reality, you need about 10dB more dynamic range of an instrument to accurately measure response.? That means you'd need an instrument that has the capability of 130dB of dynamic range to completely and fully see the bottoms of the notches.? Your NanoVNA has about 90dB of dynamic range, allowing you to reveal about -80db (or so) of notch depth.
Kevin
toggle quoted message
Show quoted text
On 4/28/2025 6:41 AM, Part 15 Engineer via groups.io wrote:
After hours of going at it. I have come to this point. I think it looks alot better, but still probably can be improved. 2 things. I can't set the nvna for more than 1024 points and the point rbw seems to unable to be changed from the 4000 hz setting. If someone knows how to change the rbw in nanovna app, immall ears. other points in the previous post were followed. I also added a 3db pad to the generator port and calibrated the nvna with the pad inline. Now I have both high and low sides looking similar. This is the high side. What should the total s21 gain be for each side approx.? I can't seem to get it lower than -0.8xx on each side. The s21 gain on reject averages about -70. The jumpers I'm using to tune will be the same ones used for the final installation. They are 2.0 meter new old stock 1/4 inch hardline with N connectors that came from cellular service. The nvna has sma to N connectors. I will follow up this post with pics of my test configuration.
|
1 - When measuring return loss (match), the RL value will be at least
twice as much as the loss value of the pad. That's correct in theory, but a problem one would likely run into in the real world is that most run-of-the-mill pads aren't exactly 50 ohms. The result of putting the pad in line is that your S11 may look better or may look worse, or may shift it in frequency due to the cascaded mismatches. It's a crap shoot unless you characterize the pad separately and "do the math", or include the pad in the calibration which is generally a bad idea because it only degrades the accuracy in the same way as would adding adapters or using ratty test cables. No pads should ever be needed or used with a VNA when measuring passive devices. --- Jeff WN3A
|
I have a local friend with access to much better test equipment than I have access to. I'm trying to see if he will tweak it the rest of the way for me. I got it in the ballpark so maybe he can finish it up.
> 1 - When measuring return loss (match), the RL value will be at least
> twice as much as the loss value of the pad.?
That's correct in theory, but a problem one would likely run into in the real world is that most run-of-the-mill pads aren't exactly 50 ohms.? The result of putting the pad in line is that your S11 may look better or may look worse, or may shift it in frequency due to the cascaded mismatches.? It's a crap shoot unless you characterize the pad separately and "do the math", or include the pad in the calibration which is generally a bad idea because it only degrades the accuracy in the same way as would adding adapters or using ratty test cables.
No pads should ever be needed or used with a VNA when measuring passive devices.
? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? --- Jeff WN3A
|
Kevin Custer