Topics

BITX40 LPF mod

Arv Evans
 

Hello

With all the filter discussion regarding uBITX I decided to take a look at design of
the output filter on the BITX-40.
LTspice IV - BITX_40__output_filter.asc_030.png
This LTSpice run shows that the stock LPF starts to roll off at about 8 MHz, which is
pretty good.  Problem is that the magic -43 db point does not come until about 21 MHz
or about the 3rd harmonic area.  Can this be imporoved...maybe.

LTspice IV - BITX_40__output_filter_with_parallel_resonance.asc_031.png
By adding just one capacitor across the output inductor we now get much more
attenuation at the 3rd harmonic point of 21 MHz.  By changing the value of C3
it is possible to move that point up or down in frequency.  This might be a way
to attenuate a specific harmonic or spur if it does not move about too much.

Next step is probably to scan the BITX-40 output to see if it really needs this type
of modification (how clean is it?).

It should also be possible to add a similar capacitor across the L6 inductor to have
another attenuation point.  That is a test for tomorrow because it is late and us old
guys need their beauty sleep.    8-)

NOTE: The upper trace in the pictures is frequency response.  The lower trace is phase
shift, which could be important if oscillation is encountered.  Note the sharp phase
shift at the parallel resonance point in the 2nd picture.

Arv  K7HKL
_._

Glenn
 

Arv,
I read somewhere that its a good idea to drop the inductor value by 25% or so, (cant recall exact %) then calculate the parallel Cap value for the required notch.  Otherwise i think the RL is compromised.
Or my memory is suspect, quite possible........
vk3pe

Arv Evans
 

Glen VK3PE

That is something that can be tested.  It doesn't show in the simulator, but real-world
may be different.  When I get to actually wiring this it will be easy to test.
Thanks for the info.

Arv
_._


On Mon, Aug 6, 2018 at 10:18 PM Glenn <glennp@...> wrote:
Arv,
I read somewhere that its a good idea to drop the inductor value by 25% or so, (cant recall exact %) then calculate the parallel Cap value for the required notch.  Otherwise i think the RL is compromised.
Or my memory is suspect, quite possible........
vk3pe

Jerry Gaffke
 

Don Cantrell ND6T documented this adjustment to the Bitx40 LPF in January of 2017:
    http://bitxhacks.blogspot.com/2017/01/nd6ts-suppression-of-pesky-2nd-harmonic.html


On Mon, Aug 6, 2018 at 09:14 PM, Arv Evans wrote:
With all the filter discussion regarding uBITX I decided to take a look at design of
the output filter on the BITX-40.

Arv Evans
 

Re.  BITX-40 LPF Mod
When I added the parallel capacitance to notch out the 3rd harmonic the simulator showed
a very narrow notch.  Today I tried adding a resistance in series with that parallel capacitor.
The result is a much broader notch at approximately 21 MHz, but still with 63 db of attenuation.
LTspice IV - BITX_40__output_filter_with_parallel_resonance_and_Series_Resistance.asc_032.png
The upper trace is frequency response and the lower trace shows phase shift.  Adding the resistor seems to have
tamed the phase shift so it does not show a dramatic transition at resonance of L7 and C3.

Next possibly illogical step is to notch out any 2nd harmonic energy at 14 MHz.  This was done with a series trap
(R3, C1, L4).  With R3 at 1.5 ohms this looks promising.
LTspice IV - BITX_40__output_filter_with_21MHz_parallel_resonance_and_14_MHz_Trap_with_Resistance.asc_035.png
Upper trace shows frequency response and lower trace is phase shift.   
Results look like (1) no additional insertion loss, (2) Reasonably broad notch at 14 MHz, and
The notch at 21 MHz is still there, and (4) overall LPF roll off still starts at 10 MHz.

Will this actually work in a real-world circuit?  It will be interesting to find out.  If it does work it
could insure that 2nd and 3rd harmonics can be reduced to way below FCC requirement for the
BITX-40 transceiver.  The trap idea is interesting because more traps could possibly be added
to take care of any 12 MHz IF leak-through, or any other spurious signal that looks like it might
be a problem. 

Could this approach be applied to the uBITX filter problem...maybe, but I will leave that to people
who are much smarter than myself.

Arv
_._



On Mon, Aug 6, 2018 at 10:22 PM Arv Evans <arvid.evans@...> wrote:
Glen VK3PE

That is something that can be tested.  It doesn't show in the simulator, but real-world
may be different.  When I get to actually wiring this it will be easy to test.
Thanks for the info.

Arv
_._

On Mon, Aug 6, 2018 at 10:18 PM Glenn <glennp@...> wrote:
Arv,
I read somewhere that its a good idea to drop the inductor value by 25% or so, (cant recall exact %) then calculate the parallel Cap value for the required notch.  Otherwise i think the RL is compromised.
Or my memory is suspect, quite possible........
vk3pe

Arv Evans
 

Jerry

I first came across that parallel cap and inductor trick when Dan Tayloe N7VE used it in the
BITX20A, and later the BITX17A.  It worked well there so I decided to try it with my early
version BITX40 from HFSignals.  No reason why it should not work with other transceivers
that use an LPF connected to the antenna. 

From LTSpice simulations this trick seems to work best with the inductor closest to the
antenna.  I did do several simulations with the parallel capacitor on the inductor closest
to the finals, but results were a bit discouraging.  My guess is that if the parallel LC is
on the RF PA side of the LPF then unwanted RF gets reflected back toward the finals and
does strange things.

This morning I posted more information for using an LCR series trap on the output to suck
out 2nd harmonic energy.  This too seems to work, at least in the LTSpice simulator.  Will
probably be trying this in the real circuit sometime in the next week or so. 

Arv
_._


On Mon, Aug 6, 2018 at 11:39 PM Jerry Gaffke via Groups.Io <jgaffke=yahoo.com@groups.io> wrote:
Don Cantrell ND6T documented this adjustment to the Bitx40 LPF in January of 2017:
    http://bitxhacks.blogspot.com/2017/01/nd6ts-suppression-of-pesky-2nd-harmonic.html


On Mon, Aug 6, 2018 at 09:14 PM, Arv Evans wrote:
With all the filter discussion regarding uBITX I decided to take a look at design of
the output filter on the BITX-40.

Jerry Gaffke
 

Yes, I've been following what you are doing with interest.
A significant contribution if knocking down two harmonics can be done with just a couple extra caps.

The original intent on the uBitx was to get by with fewer filters by assuming the push-pull final
would have a negligible second harmonic.  It seemed Farhan got that to work,
but it had to be adjusted very carefully. 
I'd expect the second harmonic to not be much of an issue.

On the Bitx40, the second harmonic is quite severe due to how unbalanced the final is.
With only 100ma of quiescent current, the IRF510 is not providing much ooomph on low going peaks.

I'm seriously considering putting together a signal analyzer along the lines of your earlier post.
    Step_Attenuator -- 50mhz_LPF -- diode mixer -- MMIC_Amp -- PX1002_SAW -- ad8307 -- Nano_ADC

In parallel with the AD8307, add an SA612A plus audio amp so I can hear it.
There's lots of information in that audio if you know what to expect.

Jerry, KE7ER
 


On Tue, Aug 7, 2018 at 09:19 AM, Arv Evans wrote:
This morning I posted more information for using an LCR series trap on the output to suck
out 2nd harmonic energy.  This too seems to work, at least in the LTSpice simulator.  Will
probably be trying this in the real circuit sometime in the next week or so. 

ajparent1/KB1GMX <kb1gmx@...>
 

Arv,
The problem is the range of harmonic frequencies:

The stock 80M filter allows for 60M so the third harmonic goes from
10.5 to over 15.9 mhz.  Same for the others but the third harmonic of 
higher than 10MHZ hits the first filter and the PA filter.

If the filters were not compromised by layout and relay issues they do much better.

If we didn't amplify square waves and filtered at the input it would less an issue
at the output.

Allison

Arv Evans
 

Jerry

Lots of ways to skin the cat...even Schroedinger's variety.

The little sweepy test set that I built started as an experiment, and just keeps evolving.
Looking at audio with fldgi or argo is revealing.  At present the mixer is just 4 diodes and
a pair of FT37-43 cores, but it would not take much work to change that to a Gilbert-cell.
I have several AD8307 detectors available and may add those to the design at a later date.
Eventually the design may stabilize enough to warrant a PCB, chassis, and technical
writeup, but I'm not sure when that will be.  It is just too much fun to keep experimenting.

The series RLC traps for taming BITX40 harmonics were a total unknown until yesterday
when I tried it in LTSpice.  Results are quite encouraging.  I'm assuming that power
handling in the traps will be dependent on how strong the harmonics might be...more
experimenting to be done in that area.

I'm slowly working on a Frankenstein version of Raduino.  My BITX40 is a very early
version with analog VFO that wants to drift a bit.  First change was to add Huff & Puff
stabilization but it really needs a digital frequency display.  Work direction for this is a
74HC4066 pre-scaler and Arduino Nano to time interrupts.  But the 12 MHz USB/TTL
converter oscillator may be problematic for a system with a 12 MHz BFO and IF filter.
Several Arduino Pro-Mini with real 16 MHz crystal are available so will probably port
the Nano code to the Pro-mini platform.  This eliminates the 12 MHz oscillator as long
as the programming dongle is unplugged, but that eliminates CAT capability unless
it is done with an older PC that still has RS-232 for talking directly to the Arduino.  That
then requires a TTL-to/from-RS-232 voltage converter.   Yet another project.

Just way too many projects to make any real progress on any specific one.   8-)

Arv
_._



On Tue, Aug 7, 2018 at 11:37 AM Jerry Gaffke via Groups.Io <jgaffke=yahoo.com@groups.io> wrote:
Yes, I've been following what you are doing with interest.
A significant contribution if knocking down two harmonics can be done with just a couple extra caps.

The original intent on the uBitx was to get by with fewer filters by assuming the push-pull final
would have a negligible second harmonic.  It seemed Farhan got that to work,
but it had to be adjusted very carefully. 
I'd expect the second harmonic to not be much of an issue.

On the Bitx40, the second harmonic is quite severe due to how unbalanced the final is.
With only 100ma of quiescent current, the IRF510 is not providing much ooomph on low going peaks.

I'm seriously considering putting together a signal analyzer along the lines of your earlier post.
    Step_Attenuator -- 50mhz_LPF -- diode mixer -- MMIC_Amp -- PX1002_SAW -- ad8307 -- Nano_ADC

In parallel with the AD8307, add an SA612A plus audio amp so I can hear it.
There's lots of information in that audio if you know what to expect.

Jerry, KE7ER
 

On Tue, Aug 7, 2018 at 09:19 AM, Arv Evans wrote:
This morning I posted more information for using an LCR series trap on the output to suck
out 2nd harmonic energy.  This too seems to work, at least in the LTSpice simulator.  Will
probably be trying this in the real circuit sometime in the next week or so. 

Arv Evans
 

Allison

Agreed.  In the BITX40 case the RF PA section is generating some uglies all on its own.
That is a bit different from the uBITX which has a relatively clean push-pull layout.  With
the push-pull layout it should be possible to insist on clean inputs to get clean outputs. 
Not the case with the BITX40 single-ended RF PA section.  I considered changing it to
a push-pull layout but laziness set in and working on the output filter seemed like more
fun.

Arv
_._


On Tue, Aug 7, 2018 at 1:06 PM ajparent1/KB1GMX <kb1gmx@...> wrote:
Arv,
The problem is the range of harmonic frequencies:

The stock 80M filter allows for 60M so the third harmonic goes from
10.5 to over 15.9 mhz.  Same for the others but the third harmonic of 
higher than 10MHZ hits the first filter and the PA filter.

If the filters were not compromised by layout and relay issues they do much better.

If we didn't amplify square waves and filtered at the input it would less an issue
at the output.

Allison

ajparent1/KB1GMX <kb1gmx@...>
 

On Tue, Aug 7, 2018 at 10:37 AM, Jerry Gaffke wrote:
I'm seriously considering putting together a signal analyzer along the lines of your earlier post.
    Step_Attenuator -- 50mhz_LPF -- diode mixer -- MMIC_Amp -- PX1002_SAW -- ad8307 -- Nano_ADC
Add a monchrome graphic display and you have a spectrum analyser! 
A 128 or 240x64 will do fine some need I2C other can be run with
4bit interface like the 1x16, 2x16 and 4x20 displays. or just get a
si5351 board from adafruit and a Nano and post to the computer
via the USB.   If you use the 5351 at another port you can do a
tracking generator.

Emits fx to port 0, measure fx...post to scale on local display or send via
USB to computer to pretty up.

If your really slick using the 5351 as both LO and tracking generator
its possible to emit at say the receiver input frequency and look at
the IF frequency (offset). Send here at Fa, listen there at Fa(+-n)! 
Many SAs can't do that!

Add the ADC8307 and audio baseband and you have a really useful tool.

Allison

Arv Evans
 

Correction...the pre-scaler is a 74HC4060. 


On Tue, Aug 7, 2018 at 1:47 PM Arv Evans <arvid.evans@...> wrote:
Jerry

Lots of ways to skin the cat...even Schroedinger's variety.

The little sweepy test set that I built started as an experiment, and just keeps evolving.
Looking at audio with fldgi or argo is revealing.  At present the mixer is just 4 diodes and
a pair of FT37-43 cores, but it would not take much work to change that to a Gilbert-cell.
I have several AD8307 detectors available and may add those to the design at a later date.
Eventually the design may stabilize enough to warrant a PCB, chassis, and technical
writeup, but I'm not sure when that will be.  It is just too much fun to keep experimenting.

The series RLC traps for taming BITX40 harmonics were a total unknown until yesterday
when I tried it in LTSpice.  Results are quite encouraging.  I'm assuming that power
handling in the traps will be dependent on how strong the harmonics might be...more
experimenting to be done in that area.

I'm slowly working on a Frankenstein version of Raduino.  My BITX40 is a very early
version with analog VFO that wants to drift a bit.  First change was to add Huff & Puff
stabilization but it really needs a digital frequency display.  Work direction for this is a
74HC4066 pre-scaler and Arduino Nano to time interrupts.  But the 12 MHz USB/TTL
converter oscillator may be problematic for a system with a 12 MHz BFO and IF filter.
Several Arduino Pro-Mini with real 16 MHz crystal are available so will probably port
the Nano code to the Pro-mini platform.  This eliminates the 12 MHz oscillator as long
as the programming dongle is unplugged, but that eliminates CAT capability unless
it is done with an older PC that still has RS-232 for talking directly to the Arduino.  That
then requires a TTL-to/from-RS-232 voltage converter.   Yet another project.

Just way too many projects to make any real progress on any specific one.   8-)

Arv
_._



On Tue, Aug 7, 2018 at 11:37 AM Jerry Gaffke via Groups.Io <jgaffke=yahoo.com@groups.io> wrote:
Yes, I've been following what you are doing with interest.
A significant contribution if knocking down two harmonics can be done with just a couple extra caps.

The original intent on the uBitx was to get by with fewer filters by assuming the push-pull final
would have a negligible second harmonic.  It seemed Farhan got that to work,
but it had to be adjusted very carefully. 
I'd expect the second harmonic to not be much of an issue.

On the Bitx40, the second harmonic is quite severe due to how unbalanced the final is.
With only 100ma of quiescent current, the IRF510 is not providing much ooomph on low going peaks.

I'm seriously considering putting together a signal analyzer along the lines of your earlier post.
    Step_Attenuator -- 50mhz_LPF -- diode mixer -- MMIC_Amp -- PX1002_SAW -- ad8307 -- Nano_ADC

In parallel with the AD8307, add an SA612A plus audio amp so I can hear it.
There's lots of information in that audio if you know what to expect.

Jerry, KE7ER
 

On Tue, Aug 7, 2018 at 09:19 AM, Arv Evans wrote:
This morning I posted more information for using an LCR series trap on the output to suck
out 2nd harmonic energy.  This too seems to work, at least in the LTSpice simulator.  Will
probably be trying this in the real circuit sometime in the next week or so. 

Jerry Gaffke
 

Yes, I meant spectrum analyzer.
And definitely would have the tracking generator,
trivial now with multichannel signal generators like the si5351.

Though I've been thinking would be nice to include a way to get phase information as per
a signal analyzer, perhaps following the lead of the W5BIT Vector Impedance Analyzer (well described).
That's a technique also used by the Kees' AQRP VIA, though the AQRP device is using square waves.
But VIA capability is likely best kept a separate project. 

For occasional use a UART link to a PC for display is good enough,
fancy touch screen LCD a possible future option if it takes off.
Definitely need a way to get data from this new device out to the a host computer of some sort anyway.

Perhaps the Si5338 instead of the Si5351, more stable and possible to adapt for use into UHF.

I'd tend to use a $5 ARM chip with a decent 12 to 16 bit ADC instead of the development boards.
The development boards come and go with time, more so than the chips.
And have way more stuff on them than I need.

But could also include hooks to use most any offboard computer instead of the ARM chip,
anything from a Nano to an RPi.

Jerry, KE7ER



On Tue, Aug 7, 2018 at 12:59 PM, ajparent1/KB1GMX wrote:
On Tue, Aug 7, 2018 at 10:37 AM, Jerry Gaffke wrote:
I'm seriously considering putting together a signal analyzer along the lines of your earlier post.
    Step_Attenuator -- 50mhz_LPF -- diode mixer -- MMIC_Amp -- PX1002_SAW -- ad8307 -- Nano_ADC
Add a monchrome graphic display and you have a spectrum analyser! 
A 128 or 240x64 will do fine some need I2C other can be run with
4bit interface like the 1x16, 2x16 and 4x20 displays. or just get a
si5351 board from adafruit and a Nano and post to the computer
via the USB.   If you use the 5351 at another port you can do a
tracking generator.

Emits fx to port 0, measure fx...post to scale on local display or send via
USB to computer to pretty up.

If your really slick using the 5351 as both LO and tracking generator
its possible to emit at say the receiver input frequency and look at
the IF frequency (offset). Send here at Fa, listen there at Fa(+-n)! 
Many SAs can't do that!

Add the ADC8307 and audio baseband and you have a really useful tool.

F1BFU - Fr - 79
 

Arv

I find an 

AD831 Low Distortion High Frequency Mixer Module Active 500 MHz Balanced Mixer on China shop.


for the spectrum analyser with VFO/SigGen QRP Labs.

Gilles


Le mar. 7 août 2018 à 21:47, Arv Evans <arvid.evans@...> a écrit :
Jerry

Lots of ways to skin the cat...even Schroedinger's variety.

The little sweepy test set that I built started as an experiment, and just keeps evolving.
Looking at audio with fldgi or argo is revealing.  At present the mixer is just 4 diodes and
a pair of FT37-43 cores, but it would not take much work to change that to a Gilbert-cell.
I have several AD8307 detectors available and may add those to the design at a later date.
Eventually the design may stabilize enough to warrant a PCB, chassis, and technical
writeup, but I'm not sure when that will be.  It is just too much fun to keep experimenting.

The series RLC traps for taming BITX40 harmonics were a total unknown until yesterday
when I tried it in LTSpice.  Results are quite encouraging.  I'm assuming that power
handling in the traps will be dependent on how strong the harmonics might be...more
experimenting to be done in that area.

I'm slowly working on a Frankenstein version of Raduino.  My BITX40 is a very early
version with analog VFO that wants to drift a bit.  First change was to add Huff & Puff
stabilization but it really needs a digital frequency display.  Work direction for this is a
74HC4066 pre-scaler and Arduino Nano to time interrupts.  But the 12 MHz USB/TTL
converter oscillator may be problematic for a system with a 12 MHz BFO and IF filter.
Several Arduino Pro-Mini with real 16 MHz crystal are available so will probably port
the Nano code to the Pro-mini platform.  This eliminates the 12 MHz oscillator as long
as the programming dongle is unplugged, but that eliminates CAT capability unless
it is done with an older PC that still has RS-232 for talking directly to the Arduino.  That
then requires a TTL-to/from-RS-232 voltage converter.   Yet another project.

Just way too many projects to make any real progress on any specific one.   8-)

Arv
_._



On Tue, Aug 7, 2018 at 11:37 AM Jerry Gaffke via Groups.Io <jgaffke=yahoo.com@groups.io> wrote:
Yes, I've been following what you are doing with interest.
A significant contribution if knocking down two harmonics can be done with just a couple extra caps.

The original intent on the uBitx was to get by with fewer filters by assuming the push-pull final
would have a negligible second harmonic.  It seemed Farhan got that to work,
but it had to be adjusted very carefully. 
I'd expect the second harmonic to not be much of an issue.

On the Bitx40, the second harmonic is quite severe due to how unbalanced the final is.
With only 100ma of quiescent current, the IRF510 is not providing much ooomph on low going peaks.

I'm seriously considering putting together a signal analyzer along the lines of your earlier post.
    Step_Attenuator -- 50mhz_LPF -- diode mixer -- MMIC_Amp -- PX1002_SAW -- ad8307 -- Nano_ADC

In parallel with the AD8307, add an SA612A plus audio amp so I can hear it.
There's lots of information in that audio if you know what to expect.

Jerry, KE7ER
 

On Tue, Aug 7, 2018 at 09:19 AM, Arv Evans wrote:
This morning I posted more information for using an LCR series trap on the output to suck
out 2nd harmonic energy.  This too seems to work, at least in the LTSpice simulator.  Will
probably be trying this in the real circuit sometime in the next week or so. 

Arv Evans
 

Giles

Price is right.  Size is right.  Could be an interesting part of a lab-grade analyzer.

Arv
_._


On Tue, Aug 7, 2018 at 2:52 PM Gilles Delpech <gilles.f1bfu@...> wrote:
Arv

I find an 

AD831 Low Distortion High Frequency Mixer Module Active 500 MHz Balanced Mixer on China shop.


for the spectrum analyser with VFO/SigGen QRP Labs.

Gilles


Le mar. 7 août 2018 à 21:47, Arv Evans <arvid.evans@...> a écrit :
Jerry

Lots of ways to skin the cat...even Schroedinger's variety.

The little sweepy test set that I built started as an experiment, and just keeps evolving.
Looking at audio with fldgi or argo is revealing.  At present the mixer is just 4 diodes and
a pair of FT37-43 cores, but it would not take much work to change that to a Gilbert-cell.
I have several AD8307 detectors available and may add those to the design at a later date.
Eventually the design may stabilize enough to warrant a PCB, chassis, and technical
writeup, but I'm not sure when that will be.  It is just too much fun to keep experimenting.

The series RLC traps for taming BITX40 harmonics were a total unknown until yesterday
when I tried it in LTSpice.  Results are quite encouraging.  I'm assuming that power
handling in the traps will be dependent on how strong the harmonics might be...more
experimenting to be done in that area.

I'm slowly working on a Frankenstein version of Raduino.  My BITX40 is a very early
version with analog VFO that wants to drift a bit.  First change was to add Huff & Puff
stabilization but it really needs a digital frequency display.  Work direction for this is a
74HC4066 pre-scaler and Arduino Nano to time interrupts.  But the 12 MHz USB/TTL
converter oscillator may be problematic for a system with a 12 MHz BFO and IF filter.
Several Arduino Pro-Mini with real 16 MHz crystal are available so will probably port
the Nano code to the Pro-mini platform.  This eliminates the 12 MHz oscillator as long
as the programming dongle is unplugged, but that eliminates CAT capability unless
it is done with an older PC that still has RS-232 for talking directly to the Arduino.  That
then requires a TTL-to/from-RS-232 voltage converter.   Yet another project.

Just way too many projects to make any real progress on any specific one.   8-)

Arv
_._



On Tue, Aug 7, 2018 at 11:37 AM Jerry Gaffke via Groups.Io <jgaffke=yahoo.com@groups.io> wrote:
Yes, I've been following what you are doing with interest.
A significant contribution if knocking down two harmonics can be done with just a couple extra caps.

The original intent on the uBitx was to get by with fewer filters by assuming the push-pull final
would have a negligible second harmonic.  It seemed Farhan got that to work,
but it had to be adjusted very carefully. 
I'd expect the second harmonic to not be much of an issue.

On the Bitx40, the second harmonic is quite severe due to how unbalanced the final is.
With only 100ma of quiescent current, the IRF510 is not providing much ooomph on low going peaks.

I'm seriously considering putting together a signal analyzer along the lines of your earlier post.
    Step_Attenuator -- 50mhz_LPF -- diode mixer -- MMIC_Amp -- PX1002_SAW -- ad8307 -- Nano_ADC

In parallel with the AD8307, add an SA612A plus audio amp so I can hear it.
There's lots of information in that audio if you know what to expect.

Jerry, KE7ER
 

On Tue, Aug 7, 2018 at 09:19 AM, Arv Evans wrote:
This morning I posted more information for using an LCR series trap on the output to suck
out 2nd harmonic energy.  This too seems to work, at least in the LTSpice simulator.  Will
probably be trying this in the real circuit sometime in the next week or so. 

ajparent1/KB1GMX <kb1gmx@...>
 

Jerry,

The basic theory is phase of source and phase of returned signal, S11 Rx,Jx is the result
or reflected (referenced to 50 ohms) voltage and phase.  To do that you need to do
phase comparison and mixers like DBM can do that.

There are two systems in use to do that one uses a current measurement and voltage 
and compared those to reference (source signal) the other uses directional couplers
and refernce their output to the source.  The source has to be clean sine.  Once you
have phase and amplitude data (basically S11) you can calculate the result and
plot it as smith, polar, or as desired.  Not trivial but not rocket science (logs trig and
a little vector math).

Both require a clean sine source as harmonics will render teh results useless.  Simple
example is MFJ259B.  FYI if the source has issues in the 259B the readings are bad
and part of calibrating it is verifying the output has minimum (-40dbc) harmonics. 
I've replaced the diodes in mine and recalibrated it.  

Allison

Jerry Gaffke
 

It is possible to use a square wave as the signal source in a VNA.

I looked hard at the N2PK VNA a few years ago, that is a sine wave source
into the DUT, and the signal out of the DUT is inspected by two mixers with a LO
at the same freq as the DUT source but in quadrature, outputs at DC.  (Now that's baseband!)


The W5BIG AIM430 VIA creates 2 sine waves from DDS chips for DUT source and an LO to the mixers,
the two mixers effectively measure current and voltage from the DUT into a load resistor.
The mixers get sampled with ADC's and that data is passed on to a PC for an FFT to 
find the amplitude and phase of the two 1khz audio signals (which represent RF voltage and current)
He's got it pretty well described here:  http://w5big.com/QST_Article.pdf

The DG8SAQ VNWA is similar, but he's got that thing wound up to work well at some
very high frequencies, far higher than any of the data sheets for the parts being used
would suggest possible.  Well regarded, and very accurate.  Not cheap.
Mixer outputs are at a few khz, as I recall.
He's sellng lots of these, does not provide complete instructions on how to go into competition.
 
The AQRP VIA that Kees sells is is similar to the AIM430.  The big difference here is it
gets by using all square waves.  An FFT is carried out by an on-board ARM processor,
and by ignoring all but the 4khz difference frequency between DUT source and the LO to the mixers,
it can ignore the harmonics in the square wave.  It has recently been extended for use
as a two port VNA.  A nice self contained unit with touch screen at a good price.
    http://www.qsl.net/k5bcq/Kits/Kits.html




From post 56160:
>  I'm seriously considering putting together a signal analyzer along the lines of your earlier post.
>      Step_Attenuator -- 50mhz_LPF -- diode mixer -- MMIC_Amp -- PX1002_SAW -- ad8307 -- Nano_ADC
>  In parallel with the AD8307, add an SA612A plus audio amp so I can hear it.
>  There's lots of information in that audio if you know what to expect.

If instead of one SA612A we have two mixers driven with quadrature LO's, we could use SDR
techniques to separate upper and lower sidebands in an ARM processor.
But then we have pretty much all we need for an AQRP style VIA.
Hence the creeping featurism that prevents me from ever getting out the soldering iron.

Of course, it would take me a year to figure out the DSP software involved.
Just listening to an AD8307 with an ADC makes the software trivial.
And gives us a spectrum analyzer perfectly suitable for poking at a uBitx with.

Jerry,  KE7ER



On Tue, Aug 7, 2018 at 04:30 PM, ajparent1/KB1GMX wrote:
The basic theory is phase of source and phase of returned signal, S11 Rx,Jx is the result
or reflected (referenced to 50 ohms) voltage and phase.  To do that you need to do
phase comparison and mixers like DBM can do that.

There are two systems in use to do that one uses a current measurement and voltage 
and compared those to reference (source signal) the other uses directional couplers
and refernce their output to the source.  The source has to be clean sine.  Once you
have phase and amplitude data (basically S11) you can calculate the result and
plot it as smith, polar, or as desired.  Not trivial but not rocket science (logs trig and
a little vector math).

Both require a clean sine source as harmonics will render teh results useless.  Simple
example is MFJ259B.  FYI if the source has issues in the 259B the readings are bad
and part of calibrating it is verifying the output has minimum (-40dbc) harmonics. 
I've replaced the diodes in mine and recalibrated it.  

Henning Weddig
 

Arv,

please re-simulate with real inductances i.e. with a reasonable Q  of 100.  This limited Q will already introduce a series resistance ( remember : Q = R /XL). Similar the caps also have a limited Q but "normally" larger, although I doubt if this is true for chap SMD caps.

Henning Weddig

DK5LV 


Am 07.08.2018 um 17:46 schrieb Arv Evans:

Re.  BITX-40 LPF Mod
When I added the parallel capacitance to notch out the 3rd harmonic the simulator showed
a very narrow notch.  Today I tried adding a resistance in series with that parallel capacitor.
The result is a much broader notch at approximately 21 MHz, but still with 63 db of attenuation.
LTspice IV -
BITX_40__output_filter_with_parallel_resonance_and_Series_Resistance.asc_032.png
The upper trace is frequency response and the lower trace shows phase shift.  Adding the resistor seems to have
tamed the phase shift so it does not show a dramatic transition at resonance of L7 and C3.

Next possibly illogical step is to notch out any 2nd harmonic energy at 14 MHz.  This was done with a series trap
(R3, C1, L4).  With R3 at 1.5 ohms this looks promising.
LTspice IV -
BITX_40__output_filter_with_21MHz_parallel_resonance_and_14_MHz_Trap_with_Resistance.asc_035.png
Upper trace shows frequency response and lower trace is phase shift.   
Results look like (1) no additional insertion loss, (2) Reasonably broad notch at 14 MHz, and
The notch at 21 MHz is still there, and (4) overall LPF roll off still starts at 10 MHz.

Will this actually work in a real-world circuit?  It will be interesting to find out.  If it does work it
could insure that 2nd and 3rd harmonics can be reduced to way below FCC requirement for the
BITX-40 transceiver.  The trap idea is interesting because more traps could possibly be added
to take care of any 12 MHz IF leak-through, or any other spurious signal that looks like it might
be a problem. 

Could this approach be applied to the uBITX filter problem...maybe, but I will leave that to people
who are much smarter than myself.

Arv
_._



On Mon, Aug 6, 2018 at 10:22 PM Arv Evans <arvid.evans@...> wrote:
Glen VK3PE

That is something that can be tested.  It doesn't show in the simulator, but real-world
may be different.  When I get to actually wiring this it will be easy to test.
Thanks for the info.

Arv
_._

On Mon, Aug 6, 2018 at 10:18 PM Glenn <glennp@...> wrote:
Arv,
I read somewhere that its a good idea to drop the inductor value by 25% or so, (cant recall exact %) then calculate the parallel Cap value for the required notch.  Otherwise i think the RL is compromised.
Or my memory is suspect, quite possible........
vk3pe

Arv Evans
 

Henning DK5LV

Will do...just as soon as I figure out how to add Q-factor to LTSpice components.
Part of the fun is learning how to better use Spice simulation.

Arv
_._


On Wed, Aug 8, 2018 at 1:19 AM Henning Weddig via Groups.Io <hweddig=aol.com@groups.io> wrote:

Arv,

please re-simulate with real inductances i.e. with a reasonable Q  of 100.  This limited Q will already introduce a series resistance ( remember : Q = R /XL). Similar the caps also have a limited Q but "normally" larger, although I doubt if this is true for chap SMD caps.

Henning Weddig

DK5LV 


Am 07.08.2018 um 17:46 schrieb Arv Evans:
Re.  BITX-40 LPF Mod
When I added the parallel capacitance to notch out the 3rd harmonic the simulator showed
a very narrow notch.  Today I tried adding a resistance in series with that parallel capacitor.
The result is a much broader notch at approximately 21 MHz, but still with 63 db of attenuation.

The upper trace is frequency response and the lower trace shows phase shift.  Adding the resistor seems to have
tamed the phase shift so it does not show a dramatic transition at resonance of L7 and C3.

Next possibly illogical step is to notch out any 2nd harmonic energy at 14 MHz.  This was done with a series trap
(R3, C1, L4).  With R3 at 1.5 ohms this looks promising.

Upper trace shows frequency response and lower trace is phase shift.   
Results look like (1) no additional insertion loss, (2) Reasonably broad notch at 14 MHz, and
The notch at 21 MHz is still there, and (4) overall LPF roll off still starts at 10 MHz.

Will this actually work in a real-world circuit?  It will be interesting to find out.  If it does work it
could insure that 2nd and 3rd harmonics can be reduced to way below FCC requirement for the
BITX-40 transceiver.  The trap idea is interesting because more traps could possibly be added
to take care of any 12 MHz IF leak-through, or any other spurious signal that looks like it might
be a problem. 

Could this approach be applied to the uBITX filter problem...maybe, but I will leave that to people
who are much smarter than myself.

Arv
_._



On Mon, Aug 6, 2018 at 10:22 PM Arv Evans <arvid.evans@...> wrote:
Glen VK3PE

That is something that can be tested.  It doesn't show in the simulator, but real-world
may be different.  When I get to actually wiring this it will be easy to test.
Thanks for the info.

Arv
_._

On Mon, Aug 6, 2018 at 10:18 PM Glenn <glennp@...> wrote:
Arv,
I read somewhere that its a good idea to drop the inductor value by 25% or so, (cant recall exact %) then calculate the parallel Cap value for the required notch.  Otherwise i think the RL is compromised.
Or my memory is suspect, quite possible........
vk3pe

Henning Weddig
 

Arv,

please have a look into my folder (DK5LV stuff) in Groups.io, I just uploeaded the "original" and optimized lpf´s for 10 MHz and my optimized 7 MHz filter.

Simulationn in LTSPice: please use "add trace" and choose S21 and S11.

Henning

DK5LV


Am 08.08.2018 um 19:13 schrieb Arv Evans:

Henning DK5LV

Will do...just as soon as I figure out how to add Q-factor to LTSpice components.
Part of the fun is learning how to better use Spice simulation.

Arv
_._


On Wed, Aug 8, 2018 at 1:19 AM Henning Weddig via Groups.Io <hweddig=aol.com@groups.io> wrote:

Arv,

please re-simulate with real inductances i.e. with a reasonable Q  of 100.  This limited Q will already introduce a series resistance ( remember : Q = R /XL). Similar the caps also have a limited Q but "normally" larger, although I doubt if this is true for chap SMD caps.

Henning Weddig

DK5LV 


Am 07.08.2018 um 17:46 schrieb Arv Evans:
Re.  BITX-40 LPF Mod
When I added the parallel capacitance to notch out the 3rd harmonic the simulator showed
a very narrow notch.  Today I tried adding a resistance in series with that parallel capacitor.
The result is a much broader notch at approximately 21 MHz, but still with 63 db of attenuation.

The upper trace is frequency response and the lower trace shows phase shift.  Adding the resistor seems to have
tamed the phase shift so it does not show a dramatic transition at resonance of L7 and C3.

Next possibly illogical step is to notch out any 2nd harmonic energy at 14 MHz.  This was done with a series trap
(R3, C1, L4).  With R3 at 1.5 ohms this looks promising.

Upper trace shows frequency response and lower trace is phase shift.   
Results look like (1) no additional insertion loss, (2) Reasonably broad notch at 14 MHz, and
The notch at 21 MHz is still there, and (4) overall LPF roll off still starts at 10 MHz.

Will this actually work in a real-world circuit?  It will be interesting to find out.  If it does work it
could insure that 2nd and 3rd harmonics can be reduced to way below FCC requirement for the
BITX-40 transceiver.  The trap idea is interesting because more traps could possibly be added
to take care of any 12 MHz IF leak-through, or any other spurious signal that looks like it might
be a problem. 

Could this approach be applied to the uBITX filter problem...maybe, but I will leave that to people
who are much smarter than myself.

Arv
_._



On Mon, Aug 6, 2018 at 10:22 PM Arv Evans <arvid.evans@...> wrote:
Glen VK3PE

That is something that can be tested.  It doesn't show in the simulator, but real-world
may be different.  When I get to actually wiring this it will be easy to test.
Thanks for the info.

Arv
_._

On Mon, Aug 6, 2018 at 10:18 PM Glenn <glennp@...> wrote:
Arv,
I read somewhere that its a good idea to drop the inductor value by 25% or so, (cant recall exact %) then calculate the parallel Cap value for the required notch.  Otherwise i think the RL is compromised.
Or my memory is suspect, quite possible........
vk3pe