开云体育

I recently bought a bunch of parts and boards to build a multi band
transceiver including some of the new boards that Andy recently
produced. Amongst the various modules was an almost complete kit of
parts for a Mk1 BPF using the spectrum coils. Over the last couple of
days I put this together along with cheap ceramic capacitors that I
sourced locally in addition to some that were supplied with the boards
and other parts.

Once I fixed a minor problem where I had a shorted 390 ohm resistor in
one side of the diode switching all seems to function, sort of. I'm
using 1N4148 switching diodes with 100 ohm resistors as per the Mk1
BPF build instructions.

I have a VNWA3E and have used this to sweep and adjust the filters.
All 9 filters with the possible exception of 20m look to be quite
poor. Most having significant in band insertion loss, several up to
20dB! All filters do tune and I can approximate the curves shown in
the various documents though not exactly in all cases.

I am wondering if the locally supplied ceramic capacitors maybe of
such poor quality as to affect the filter Q or if this is a non issue.
If it is what is recommended. I would also like to know if anyone has
made comparisons against the published data and found it sufficiently
close.

I plan on building the later toroid based filter for my multi band t/r
but would like to use the canned coil version for other projects,
external switched filter for an SDR as an example but would hope it
has better performance than the current Atten-u-filter!

Any thoughts.

Thanks

Martin, HS0ZED

Hello Martin

I built the filters for my G6LBQ transceiver and aligned them using a
miniVNA and then a spectrum analyser.

The miniVNA I used (initially) needed calibrating as the first
measurements indicated poor performance of the filters. I also
discovered that the terminating resistors I used as a 50 Ohm load were
not 50 Ohm's! Once I had re-calibrated the miniVNA (and using a pair
of 100 Ohm resistors in parallel as a 50 Ohm load) I managed to get some
pretty good results. The insertion loss (if I remember correctly) was
around 0.5 - 1.0dB and the band pass filters had a very nice shape.

Of course some diodes generate noise when used as switching diodes which
may also be causing your VNA to indicate erroneous measurements(I have
experienced this as well). I have also had trouble with diode switching
in the past, with other projects and ended up aligning each filter
independently (ie no switching diodes connected) because of noise
generation, and then only to discover that some of the diodes had
failed.....

So I suppose I would check the calibration of your VNA, check that the
50 Ohm loads are actually 50 Ohm, then try each filter independently
(with no diodes connected you are just testing the L & C components as a
single band unswitched filter) before discounting the locally sourced
Capacitors as dodgy.

Measuring the G6LBQ filters on a spectrum analyser, indicates that the
measurements published in the build guides are accurate.

I hope that provides some info for you to consider.

Let me know if I can help further.

Regards

Adrian M1LCR




On 06/05/2017 02:42, Martin Sole wrote:

I recently bought a bunch of parts and boards to build a multi band
transceiver including some of the new boards that Andy recently
produced. Amongst the various modules was an almost complete kit of
parts for a Mk1 BPF using the spectrum coils. Over the last couple of
days I put this together along with cheap ceramic capacitors that I
sourced locally in addition to some that were supplied with the boards
and other parts.

Once I fixed a minor problem where I had a shorted 390 ohm resistor in
one side of the diode switching all seems to function, sort of. I'm
using 1N4148 switching diodes with 100 ohm resistors as per the Mk1
BPF build instructions.

I have a VNWA3E and have used this to sweep and adjust the filters.
All 9 filters with the possible exception of 20m look to be quite
poor. Most having significant in band insertion loss, several up to
20dB! All filters do tune and I can approximate the curves shown in
the various documents though not exactly in all cases.

I am wondering if the locally supplied ceramic capacitors maybe of
such poor quality as to affect the filter Q or if this is a non issue.
If it is what is recommended. I would also like to know if anyone has
made comparisons against the published data and found it sufficiently
close.

I plan on building the later toroid based filter for my multi band t/r
but would like to use the canned coil version for other projects,
external switched filter for an SDR as an example but would hope it
has better performance than the current Atten-u-filter!

Any thoughts.

Thanks

Martin, HS0ZED

Hi Adrian,

Thanks for the information. I have tried again and did get slightly
better results though still quite a way from those published in the
multiband directory files. I calibrate my VNWA3E directly at its ports
using an Amphenol SMA calibration kit. The connections to the board
points are through approximately 300mm long RG316 cables terminated in
SMA on one end and hard wired to the board at the other. I suppose there
could be some effects from the point of calibration but I would not
expect such a big change at these relatively low frequencies. I can add
50 ohm 10dB pads to the feed to and from the VNWA and see if that
improves things I guess.

This is an early version with the Spectrum coils not the toroids so I'm
not expecting great things.

During this week I'll try and plot some of the bands as best I can and
do so with reference to the published data for a good comparison.

Martin, HS0ZED




On 06/05/2017 10:12, Adrian Rees (MW1LCR) wrote:

Hello Martin

I built the filters for my G6LBQ transceiver and aligned them using a
miniVNA and then a spectrum analyser.

The miniVNA I used (initially) needed calibrating as the first
measurements indicated poor performance of the filters. I also
discovered that the terminating resistors I used as a 50 Ohm load were
not 50 Ohm's! Once I had re-calibrated the miniVNA (and using a pair
of 100 Ohm resistors in parallel as a 50 Ohm load) I managed to get some
pretty good results. The insertion loss (if I remember correctly) was
around 0.5 - 1.0dB and the band pass filters had a very nice shape.

Of course some diodes generate noise when used as switching diodes which
may also be causing your VNA to indicate erroneous measurements(I have
experienced this as well). I have also had trouble with diode switching
in the past, with other projects and ended up aligning each filter
independently (ie no switching diodes connected) because of noise
generation, and then only to discover that some of the diodes had
failed.....

So I suppose I would check the calibration of your VNA, check that the
50 Ohm loads are actually 50 Ohm, then try each filter independently
(with no diodes connected you are just testing the L & C components as a
single band unswitched filter) before discounting the locally sourced
Capacitors as dodgy.

Measuring the G6LBQ filters on a spectrum analyser, indicates that the
measurements published in the build guides are accurate.

I hope that provides some info for you to consider.

Let me know if I can help further.

Regards

Adrian M1LCR




On 06/05/2017 02:42, Martin Sole wrote:

I recently bought a bunch of parts and boards to build a multi band
transceiver including some of the new boards that Andy recently
produced. Amongst the various modules was an almost complete kit of
parts for a Mk1 BPF using the spectrum coils. Over the last couple of
days I put this together along with cheap ceramic capacitors that I
sourced locally in addition to some that were supplied with the boards
and other parts.

Once I fixed a minor problem where I had a shorted 390 ohm resistor in
one side of the diode switching all seems to function, sort of. I'm
using 1N4148 switching diodes with 100 ohm resistors as per the Mk1
BPF build instructions.

I have a VNWA3E and have used this to sweep and adjust the filters.
All 9 filters with the possible exception of 20m look to be quite
poor. Most having significant in band insertion loss, several up to
20dB! All filters do tune and I can approximate the curves shown in
the various documents though not exactly in all cases.

I am wondering if the locally supplied ceramic capacitors maybe of
such poor quality as to affect the filter Q or if this is a non issue.
If it is what is recommended. I would also like to know if anyone has
made comparisons against the published data and found it sufficiently
close.

I plan on building the later toroid based filter for my multi band t/r
but would like to use the canned coil version for other projects,
external switched filter for an SDR as an example but would hope it
has better performance than the current Atten-u-filter!

Any thoughts.

Thanks

Martin, HS0ZED