Topics

Quartz scale filter


Gerard
 

Hello,

Does anyone know how the quartz on the  scale filter  are chosen on production of µbitx?
This is an important point in the editing of my µbitx. 
There is a lot of literature on this subject.
Normally for a good scale filter, you will have to choose 8 quartz out of a mini hundred.
It's a hard work because many measurements are to be made on each quartz and use Dishal software
to adjust the capacitors and others to have the better narrow band.
see for example: http://www.f6evt.fr/f6evt_er/filtras-a-quartz-en-echelle-part1.pdf
I think you have the same in English.
That’s the whole question. Is it just a collection of quartz of similar frequencies?
I think they are tested on frequency to eliminate the worst, because my test some do not oscillate or unstable.
 
Anothers questions, shouldn’t they all be grounded with a rigid wire? 
I see better stability by testing and grounding a wire.
what exactly is the bandwidth of this filter?. I had ordered at the base of the 12mhz.
 
If you have any information.
Cdt


Arv Evans
 

Gerard

In much earlier times we did order same frequency crystals in 100 unit 
groups and tried to select from that to obtain the desired bandwidth and 
shape factors.  Later we discovered that Conn type filters could be made 
with just a few crystals that were labeled as the same frequency.  One or 
two of the selected crystals could be a bit off-frequency and would still 
work very well.  In Conn style filters the bandwidth is adjusted by increase
or decrease of the capacitors to ground between each crystal.  


image.png

Four crystals are usually used in modern Conn filters, mostly because 
that results in an adequate filter response, and because the components 
are not very expensive.

Some versions of this filter layouthave C2 as twice the value of C1 or C3.

There are a number of input and output layouts which can be used to 
make an improved impedance match and reduce ripple in the filter 
passband.

Arv
_._


On Wed, Aug 12, 2020 at 10:26 AM Gerard <kabupos@...> wrote:
Hello,

Does anyone know how the quartz on the  scale filter  are chosen on production of µbitx?
This is an important point in the editing of my µbitx. 
There is a lot of literature on this subject.
Normally for a good scale filter, you will have to choose 8 quartz out of a mini hundred.
It's a hard work because many measurements are to be made on each quartz and use Dishal software
to adjust the capacitors and others to have the better narrow band.
I think you have the same in English.
That’s the whole question. Is it just a collection of quartz of similar frequencies?
I think they are tested on frequency to eliminate the worst, because my test some do not oscillate or unstable.
 
Anothers questions, shouldn’t they all be grounded with a rigid wire? 
I see better stability by testing and grounding a wire.
what exactly is the bandwidth of this filter?. I had ordered at the base of the 12mhz.
 
If you have any information.
Cdt


Ashhar Farhan
 

the lucky thing about ubitx is that you can change the crystal frequency to anything, it is broadband in that segment, you will have to just retune your BFO settings.
that said, it is best to match the crystals within 10% of the bandwidth desired, that would be it should be 300 Hz for 3 KHz bandwidth. The more important thing is that the crystals need to be characterized by measuring their motional parameters. the details are in any ARRL handbook or the EMRFD. You will need a frequency counter that is good down to 100 hz accuracy or a spectrum analyzer or antuino to measure the frequency shift. The other way is to do it ear and a receiver. you should know how to detect zero-beat of the test oscillator. 
- f
- f

On Wed, Aug 12, 2020 at 10:19 PM Arv Evans <arvid.evans@...> wrote:
Gerard

In much earlier times we did order same frequency crystals in 100 unit 
groups and tried to select from that to obtain the desired bandwidth and 
shape factors.  Later we discovered that Conn type filters could be made 
with just a few crystals that were labeled as the same frequency.  One or 
two of the selected crystals could be a bit off-frequency and would still 
work very well.  In Conn style filters the bandwidth is adjusted by increase
or decrease of the capacitors to ground between each crystal.  


image.png

Four crystals are usually used in modern Conn filters, mostly because 
that results in an adequate filter response, and because the components 
are not very expensive.

Some versions of this filter layouthave C2 as twice the value of C1 or C3.

There are a number of input and output layouts which can be used to 
make an improved impedance match and reduce ripple in the filter 
passband.

Arv
_._


On Wed, Aug 12, 2020 at 10:26 AM Gerard <kabupos@...> wrote:
Hello,

Does anyone know how the quartz on the  scale filter  are chosen on production of µbitx?
This is an important point in the editing of my µbitx. 
There is a lot of literature on this subject.
Normally for a good scale filter, you will have to choose 8 quartz out of a mini hundred.
It's a hard work because many measurements are to be made on each quartz and use Dishal software
to adjust the capacitors and others to have the better narrow band.
I think you have the same in English.
That’s the whole question. Is it just a collection of quartz of similar frequencies?
I think they are tested on frequency to eliminate the worst, because my test some do not oscillate or unstable.
 
Anothers questions, shouldn’t they all be grounded with a rigid wire? 
I see better stability by testing and grounding a wire.
what exactly is the bandwidth of this filter?. I had ordered at the base of the 12mhz.
 
If you have any information.
Cdt


Gerard
 

Thank you for your answers
µbitx is fantastic


Le mer. 12 août 2020 à 19:56, Ashhar Farhan <farhanbox@...> a écrit :
the lucky thing about ubitx is that you can change the crystal frequency to anything, it is broadband in that segment, you will have to just retune your BFO settings.
that said, it is best to match the crystals within 10% of the bandwidth desired, that would be it should be 300 Hz for 3 KHz bandwidth. The more important thing is that the crystals need to be characterized by measuring their motional parameters. the details are in any ARRL handbook or the EMRFD. You will need a frequency counter that is good down to 100 hz accuracy or a spectrum analyzer or antuino to measure the frequency shift. The other way is to do it ear and a receiver. you should know how to detect zero-beat of the test oscillator. 
- f
- f

On Wed, Aug 12, 2020 at 10:19 PM Arv Evans <arvid.evans@...> wrote:
Gerard

In much earlier times we did order same frequency crystals in 100 unit 
groups and tried to select from that to obtain the desired bandwidth and 
shape factors.  Later we discovered that Conn type filters could be made 
with just a few crystals that were labeled as the same frequency.  One or 
two of the selected crystals could be a bit off-frequency and would still 
work very well.  In Conn style filters the bandwidth is adjusted by increase
or decrease of the capacitors to ground between each crystal.  


image.png

Four crystals are usually used in modern Conn filters, mostly because 
that results in an adequate filter response, and because the components 
are not very expensive.

Some versions of this filter layouthave C2 as twice the value of C1 or C3.

There are a number of input and output layouts which can be used to 
make an improved impedance match and reduce ripple in the filter 
passband.

Arv
_._


On Wed, Aug 12, 2020 at 10:26 AM Gerard <kabupos@...> wrote:
Hello,

Does anyone know how the quartz on the  scale filter  are chosen on production of µbitx?
This is an important point in the editing of my µbitx. 
There is a lot of literature on this subject.
Normally for a good scale filter, you will have to choose 8 quartz out of a mini hundred.
It's a hard work because many measurements are to be made on each quartz and use Dishal software
to adjust the capacitors and others to have the better narrow band.
I think you have the same in English.
That’s the whole question. Is it just a collection of quartz of similar frequencies?
I think they are tested on frequency to eliminate the worst, because my test some do not oscillate or unstable.
 
Anothers questions, shouldn’t they all be grounded with a rigid wire? 
I see better stability by testing and grounding a wire.
what exactly is the bandwidth of this filter?. I had ordered at the base of the 12mhz.
 
If you have any information.
Cdt


MadRadioModder
 

There are some nice features of this filter arrangement.  Filter skirts can be adjusted by the capacitors, and in fact, you can create a variable pass band filter by using varactor diodes to replace the capacitors as is done in the Elecraft K2. 


MRM

 


On Aug 12, 2020, at 11:49 AM, Arv Evans <arvid.evans@...> wrote:


Gerard

In much earlier times we did order same frequency crystals in 100 unit 
groups and tried to select from that to obtain the desired bandwidth and 
shape factors.  Later we discovered that Conn type filters could be made 
with just a few crystals that were labeled as the same frequency.  One or 
two of the selected crystals could be a bit off-frequency and would still 
work very well.  In Conn style filters the bandwidth is adjusted by increase
or decrease of the capacitors to ground between each crystal.  


<image.png>


Four crystals are usually used in modern Conn filters, mostly because 
that results in an adequate filter response, and because the components 
are not very expensive.

Some versions of this filter layouthave C2 as twice the value of C1 or C3.

There are a number of input and output layouts which can be used to 
make an improved impedance match and reduce ripple in the filter 
passband.

Arv
_._


On Wed, Aug 12, 2020 at 10:26 AM Gerard <kabupos@...> wrote:
Hello,

Does anyone know how the quartz on the  scale filter  are chosen on production of µbitx?
This is an important point in the editing of my µbitx. 
There is a lot of literature on this subject.
Normally for a good scale filter, you will have to choose 8 quartz out of a mini hundred.
It's a hard work because many measurements are to be made on each quartz and use Dishal software
to adjust the capacitors and others to have the better narrow band.
I think you have the same in English.
That’s the whole question. Is it just a collection of quartz of similar frequencies?
I think they are tested on frequency to eliminate the worst, because my test some do not oscillate or unstable.
 
Anothers questions, shouldn’t they all be grounded with a rigid wire? 
I see better stability by testing and grounding a wire.
what exactly is the bandwidth of this filter?. I had ordered at the base of the 12mhz.
 
If you have any information.
Cdt


--

…_. _._


Gerard
 

Hello,
Can you please highlight one point. 
On the documentation of µbitx 6 it is noted:
"

12 MHz SSB filter

 The ladder topology is now enhanced with the improvisation suggested by G3UUR. Paralleling up crystals at two ends of the regular ladder filter of Cohn topology really flattens out the response and even improves the losses. We use a six-section ladder filter here as we can afford the slightly higher losses given that we have had enough gain from the preceding stages. Microprocessor-grade crystals are available cheaply and are well suited for the purpose. The lower Q of these crystals results in higher losses. We can handle the higher losses by increasing the gain in the 2nd RF amps that in turn results in slightly lower IIP3 (it is about +5 dbm as measured) at close range. The 12 MHz filter needs 200 ohms termination at both ends. We achieve this through 1:4 transformers that have the robust 50 ohms terminations. Taking care to terminate filters properly is the secret to having a nice sounding radio."

Can you explain to me how the 1:4 ratio of transformers T5 and T6 (The one before and after the filter) is done.
For me all transformers are tricoils between 8 to 10 turns on FT37-43 as noted on the documentation: "All the RF transfomers are 8-10 trifilar turns on FT37-43"
On the diagram we use only 2 windings on the 3 of the tricoil.
There’s probably a technical explanation (?)
Thank's for your's answers

cdt




Gerard
 

It may be more T7 that has this relationship
If anyone can explain how it works.
thank you


Ashhar Farhan
 

thank you for getting interested in this.
consider that the current that comes out of one winding, goes into the the second too.
hence, both the windings have the same current and the voltage developed across each winding is the same. hence, across the two ends of the transformer, the voltage developed is twice that on the center tap. 
you can also consider the current being fed from the previous stage input that is flowing into either of the windings. just, simply apply kirchoff's voltage law, it is a nice little puzzle that is very rewarding. 
- f 

On Mon, Aug 17, 2020 at 12:36 AM Gerard <kabupos@...> wrote:
Hello,
Can you please highlight one point. 
On the documentation of µbitx 6 it is noted:
"

12 MHz SSB filter

 The ladder topology is now enhanced with the improvisation suggested by G3UUR. Paralleling up crystals at two ends of the regular ladder filter of Cohn topology really flattens out the response and even improves the losses. We use a six-section ladder filter here as we can afford the slightly higher losses given that we have had enough gain from the preceding stages. Microprocessor-grade crystals are available cheaply and are well suited for the purpose. The lower Q of these crystals results in higher losses. We can handle the higher losses by increasing the gain in the 2nd RF amps that in turn results in slightly lower IIP3 (it is about +5 dbm as measured) at close range. The 12 MHz filter needs 200 ohms termination at both ends. We achieve this through 1:4 transformers that have the robust 50 ohms terminations. Taking care to terminate filters properly is the secret to having a nice sounding radio."

Can you explain to me how the 1:4 ratio of transformers T5 and T6 (The one before and after the filter) is done.
For me all transformers are tricoils between 8 to 10 turns on FT37-43 as noted on the documentation: "All the RF transfomers are 8-10 trifilar turns on FT37-43"
On the diagram we use only 2 windings on the 3 of the tricoil.
There’s probably a technical explanation (?)
Thank's for your's answers

cdt




Gerard
 

Hello,
Test crystal filter 45mhz
i have tested this filter with générator and oscilloscope.
https://www.ebay.fr/itm/Motorola-45MHz-Cristal-de-Bande-Passante-Filtre-Neuf-Qty-2/161791701078?hash=item25ab897056:g:ppkAAMXQ8odRHtxp
The filter works fine, and I have the maximum amplitude at 45006900htz. However, an injected signal of 15mhz, passes. A harmonic effect? I think that should be normal.
Do you have a drawing for a serious test?
Thank you


Ashhar Farhan
 

You are seeing the third harmonic of the 15 mhz signal pass through. This is why an instrument lile Antuino is more complex. The detector has to be tuned to the frequency being scanned.


On Tue 18 Aug, 2020, 3:05 PM Gerard, <kabupos@...> wrote:
Hello,
Test crystal filter 45mhz
i have tested this filter with générator and oscilloscope.
https://www.ebay.fr/itm/Motorola-45MHz-Cristal-de-Bande-Passante-Filtre-Neuf-Qty-2/161791701078?hash=item25ab897056:g:ppkAAMXQ8odRHtxp
The filter works fine, and I have the maximum amplitude at 45006900htz. However, an injected signal of 15mhz, passes. A harmonic effect? I think that should be normal.
Do you have a drawing for a serious test?
Thank you