Digitally Stabilized VFO For SDR
1. H&P VFO With Hard Locking Positions
2. AN OBSERVATION
3. VCO SIGNAL DYNAMICS?
* * * * * * * * * * * * * * * * * * *
Circuit Diagrams for the following circuit topics are in my "ka9rza"
folder at the "soft_radio" group files section of this group. Some
there are "Beta" testing ideas and maybe too preliminary and the
comments in the text are subject to great change to coin proper
definitive terms reflecting the facts of the state of the art as it
develops over time. Previous errors in experimental observations and
or assumptions of operation are subject to change for more accuracy as
we progress in this subject over time. There are still math terms to
coin, measurements to plot and graphical signal analysis to be made.
* * * * * * * * * * * * * * * * * * *
1. H&P VFO With Hard Locking Positions
{I see there are recent new soft_radio members so QSD here means
Quadrature Sampling Detector, a phase detector which uses two split RF
signals of a desired segment of RF, phase shifted by 90 degrees and
mixed with a Local Oscillator to produce two 90 degree phased, right
and left channel audio i.f. signals to the PC sound card. The RF then
is being converted down to a extremely low i.f. frequency in the audio
passband from 0~96 kHz there about, used then as a 48 kHz width of
tunable band spectrum. Software processes this wide swath of bandwidth
and detects the passband frequency location of each signal for an
oscillographic spectrum display with frequency readout, demodulates the
signals in AM, AM Synchronous, C Quam AM Stereo, SSB, CW, FM and DRM
modes. As well as adds automatic gain control, noise reduction, noise
blanking, band pass filtering and even a multiple notch filter.
All of these modulation detection modes are in the software and thus
replace allot of hardware circuits as is commonly found in high tech
communications receivers. Replacing also, the AGC and other circuits
where all of this is now implemented in software and thus makes the
receivers quite simple and reducible down into three or four chips and
a few external active and, or, passive circuit devices on a very small
circuit board.
H&P refers to the "Huff & Puff VFO Stabilizer" which is found on the
Internet. The H&P is a sort of Phase Lock Loop control of a VFO. Some
refer to it as Edge Trigger Lock rather than a Phase Trigger Lock and
now some are saying it is actually a Phase Lock still. Truthfully, it
is in the phase dither, of the rise of the leading edge of a wave
cycle, that the sampling occurs and is triggered. In this case however
the dither is extremely small as compared to a typical Phase Lock Loop.
Hence small FM shift and phase noise. All of this is made possible by
the complex logic circuitry found in the 74HC74 Edge Triggered Dual D
Flip Flop and other varieties of 74()()74 IC's in the 7400 Series.}
The use of a H&P VFO that creates rock solid locks in frequency
around every 1 kHz will be very useful in SDR. As you tune around you
will use a frequency counter as a built in frequency readout display to
note the VFO Frequency which is the "LO IN" into the "QSD" section.
Where there is a direct conversion scheme going on but; you may also
use a heterodyne band mixing scheme to cover numerous HF bands in one
receiver.
When you are in a band segment of interest and want to work stations
there. You will then enter the "LO IN" frequency seen on the VFO
frequency counter into the "Enter LO In" feature of SDRadio or Winrad
and then be able to tune the spectrum on the softwares spectrum
display. The graduated frequencies then seen on the dial of the
software will be accurate and you will know at what frequency each
station is on in this manner. The dial will be extremely accurate.
Also, the softwares are able to take a LO IN setting in terms of a
few figures to the right of the decimal point so you will have an
extremely fine resolution on the software dial. I your VFO LO IN
frequency locks on 7410.018 kHz then you can enter exactly that LO IN
frequency resolution into the software. You can enter in any LO IN
frequency from 0~30000 kHz, so it can be 455 kHz or 10700 kHz also.
The software is already set up for all of this.
The VFO must have a geared reduction drive to slow down the tuning so
you won't miss the locking positions, you can skip over some and land
on a far one before you know it. A two speed dial such as is on the
old Siltronics Model 90 VFO is the sort of thing you want to have.
No need for an analog VFO dial unless you want it that way, since a
small $39.00 frequency counter from Copper Electronics in Louisville KY
will fit into the VFO cabinet (or receiver cabinet if the whole project
is located in such). The more the demand on these sort of frequency
counters than the cheaper they will become in mass production terms.
Special orders from the manufacturer for an out sourced prefabricated
frequency counter part can be obtained for making a SDR receiver in
mass production terms.
In super heterodyne receiver schemes a separate mixing together of
the VFO and the Band Select heterodyne oscillator to create a
fundamental frequency for the frequency counter can be used to show the
actual tuned to frequency. Extreme isolation of this mixer and its
output from the receivers front end is required to prevent feedback of
the fundamental frequency back into the receiver. This mixer should be
right inside a shielded frequency counter inside the receiver and its
output leads kept extremely short inside the frequency counter.
* I am going to re~explore the H&P version I built that is
conventional with other's designs, which produced the hard locks, and
note how far apart each lock is and then decide on what to do. I will
post that circuit again with these locking schemes I observed with it;
and you then will see how useful that will be, although it is not
incremental. Actually it does not need to be incremental, it only
needs to lock every 1 kHz or so and then that will work as best as the
current DDS schemes.
* The current DDS scheme I am seeing here in a commercial SDR
receiver, is to make DDS channel changes in 24 kHz steps. But the DDS
can tune smaller steps. You can manually input a "LO IN frequency"
into the DDS via its software and get any desired LO IN frequency you
want which means that the selected LO IN is the center frequency on the
spectrum display window.
With a DDS step change of 24 kHz the dial of the spectrum display
changes by 24 kHz when you tune past the far end limits.
The dial then moves half way into the next segment of a 48 kHz wide
display, thus it moves half way up or down the next spectrum display
window. Incremental tuning across the dial is done only in the
software until the spectrum band edges are reached and then the DDS
makes a frequency step change in a 24 kHz step. The software then
changes the DDS and the spectrum display to readout a new segment
changed by 24 kHz.
SDRadio and Winrad are just detection softwares and do not control
the DDS so you have to enter in the LO IN frequency so that they will
reflect the correct frequency on the spectrum dial. Winrad does have
an application for using it with a receiver or two where I believe it
changes the DDS and tunes the receiver. There is now a Winrad receiver
and I believe what is known as the SDR 14.
Anyways I think you will be very pleased to use a digitally
stabilized VFO in your home brew SDR projects. It is really going to
work good also, and you will have you something extremely useful to get
around with on the air waves.
More folk should be experimenting with the H&P circuits since they
really are going to get rid of the problem with the spurs and I am now
very familiar with the spurs: they are on the spectrum display and they
are bothersome "continuous carrier like tones" you have to notch out
with a notch filter. Some occur right on top of an AM station or in
its sidebands or cover up a sideband QSO on the dial. A software notch
filter helps and you can also use the opposite sideband of an AM
station if spurious tones appear in the other sideband. Some spurs
have images of themselves across the spectrum display. Other spurs
appear as wider segments of noise that are up to 5 kHz wide and a few
take up to 10 kHz of space on the dial. And a few segments of the dial
are completely covered up with spurs and that makes those few segments
virtually useless to hear anything in.
But that problem is now eliminated. It just has to become more
common place, I mean the H&P style of VFO.
I am thinking that the " MIL SPEC Version " of SDR Hardware will be a
system that lacks the spurs and the Military won't mind turning a VFO
dial to get around the bands. Unfortunately any band mixing scheme is
going to produce spurs in some places; what we better know as Birdies,
if we can rid ourselves of most of them then OK for that. Since our
VFO will be the LO IN, making a slight tuning change can move a birdy
someplace else and out of our way: in the case of a multi band
switching scheme that has a heterodyne mixer for band selection. "In
this idea the heterodyne mixer frequency may need to be slightly
variable to move the birdy with." Since it is the combination of the
band spread heterodyne mixer and the VFO that produces the birdy;
sometimes along with perhaps, some natural band noise when an antenna
is attached.
* Be sure to listen to your H&P carrier on another receiver to detect
noise and then observe what changes in your circuit results in the
elimination of that noise. You do not want the VCO control signal
modulating the VFO. After that, use some audio analysis software to
detect further noise with an AM detection diode set up.
* In my experimentation I discovered that by adding a capacitor
across the 74HC74 output got rid of the noise but it also distorted the
waveform but this led to incremental tuning! The output thus in not
necessarily a square wave that the Integrator network (VCO network) to
the Varicap wants to see. Experts in the H&P thought that my use of
the heavy capacitor at the output was a very wrong thing to do. Since
the output is converted down to a very low frequency the use of a
capacitor in the 100 uF or so range here will not offer allot of
reactance to the chip to load it too much. After this I looked again
at the waveform across the Integrator, it still looked the same
although the input waveform was now altered from being a square wave.
This also makes the 74HC74 able to output varying amplitude which it
is not suppose to do ordinarily. The square waves are normally of a
single amplitude and vary only in period or duration of peaks. The
74HC74 rise time characteristics produces a ramped rise on the edges
rather than a right angle rise and so, in this ramp region that is
angular in its steep, if the timing of the chip is such that it falls
on this ramped rise and is only passing that portion of the waveform in
its sync, then the output will exhibit amplitude variations. It might
sound tricky but the chip synchronizes itself and if needed it will
pass a complete square or even saw wave to use as needed. Whatever
waveform that results in a lock is what it comes to settle on in this
deliberate distortion of the output of the 74HC74. So its not a tricky
thing at all.
Loading of the Q output then with a capacitor takes away high
frequency noise and distorts the output into being pulsed signals
rather than square waves. Ok this is in reference to a 455 to 1455 kHz
VFO and many not hold up as being useful as one goes higher in HF
frequency. Later test at higher frequencies will tell us how useful
this is to what frequency extent.
>>> The bottom line however with a hard locking VFO scheme that locks
1 kHz or so, is that the locking steps are fine enough for general high
performance use and thus ideal in a home brew or commercial kit
project. Also the softwares are able to take a LO IN setting in terms
of a few figures to the right of the decimal point so you will have an
extremely fine resolution on the software dial. I your LO IN frequency
locks on 7410.018 then you can enter exactly that resolution into the
software. <<<
Experimentation takes time and you might even do something
unconventional along the way. Maybe discover something useful. Not
all of your ideas will work but in time you see what ones do and what
direction they seem to want to go in.
Happy Experimenting!
Dan ka9rza
* * * * * * * * * * * * * * * * * *
2. AN OBSERVATION
For this post readers I want to give you an observation and some
insight.
I noticed on the SDR softwares that noise sometimes comes in on only
one sideband of an AM carrier. And this noise is seen on other
carriers at the same time on one or the other side bands. This means
that the noise is phase shifted and hence only mixing at the moment
with RF that is in a phase relationship with it.
Although we have a noise blanker in the software we might try and use
an extra one in the hardware, in fact, two extra ones: one for each QSD
output section. This then might improve the noise reduction a little
more and would work independently on each side band as needed. Also
this needs to be a feature that you can turn off when you feel it is
not needed and want to rely only on the softwares noise blanker feature.
This then might be a high performance add on feature, a sort of bells
and whistles extra.
Note: I know that the automatic noise blanker in the software can
work really well so do not think I am saying it is not effective
because it is. The AGC is also very effective and all of this is
really a surprising technology to use. It is most effective and can be
quite sensitive also. I am thinking then about how far we can push
things here? What features we find that will be practical?
* * * * * * * * * * * * * * * * * * *
Circuit Diagrams for the following circuit topics are in my "ka9rza"
folder at the "soft_radio" group files section of this group. Some
there are "Beta" testing ideas and maybe too preliminary and the
comments in the text are subject to great change to coin proper
definitive terms reflecting the facts of the state of the art as it
develops over time. Previous errors in experimental observations and
or assumptions of operation are subject to change for more accuracy as
we progress in this subject over time. There are still math terms to
coin, measurements to plot and graphical signal analysis to be made.
* * * * * * * * * * * * * * * * * * *
3. VCO SIGNAL DYNAMICS?
I am still Amazed like Ron Brink was in his post, there is yet some
amazing discoveries to be made with the H&P and I found one in the
files from Hans Summers site:
<http://www.hanssummers.com/radio/huffpuff/>
So I have "A reminder to look at the diodes signal voltage very closely
in a H&P design.
As a reference: I am looking at ideas from circuit diagram "PA0KSB
Driftcorrectie" Dated: February 17, 1997 (to obtain this text go to:
www.hanssummers.com) that was previously published in the Dutch
magazine "Electron" December 1996 and January 1997. The way he
connected the 74HC74 in that diagram is the way I found works best for
me. And you can go without his other transistor or the last IC if you
so choose. I have a similar model using one transistor feeding the
integrator circuit (the actual VCO network from the Q output of the
comparator section or 74HC74 to the Varactor or Varicap Diode).
So; I have sort of come full circle and now I better understand what
I am suppose to do in the circuit design. After a few hundred
experimental changes to the design to study what happens, it starts to
get clear. The control voltage has to be set for the operation range
of the diode, everything else pretty much works on its own within the
ICs. I think that the 74HC74 then does a good job at what it does as
long as the right inputs are used for the clock and the sampled signal.
So if the circuit design is correct, then replicating it can be easy
if the VCO voltage is correct for the diode being used. I realize now
that the circuit may work well one day and then not the next. If the
VCO voltage is not just right, it can work sometimes and not at others.
So I have found the circuit bug; run it down.
Also, you can have some mixing by-products occur between pins so that
has to be looked at. I think "circuit loading" is somewhat the matter
here. If needed, a resistor in between pins from the 74HC4060 to the
74HC74 can be added. I had a matter of hard mixing occur that I
discovered with test equipment and thought this initially might be the
reason for unreliable operation. It was the VCO voltage level all
along, that was giving me the lack of performance I was reading about
in the H&P models in the files. Anyways, I learned that I can filter
my inputs if needed.
Here I have some notes on:
1. VCO Control Signal Amplitude and Dynamics:
dynamics of the "range"
of the control signal
across the Varicap diode
"In Ball Park Figures"
2. My Diode Types Compared:
1N4007, 1N914, 1N270
and 1N4148
3. Volt Meters:
4. Diode Capacitance Measurements:
frequency change "versus"
diode voltage change
5. VCO Reset Idea:
* * * * * * * * * * * * * * * * * * * * * * *
1. VCO Control Signal Amplitude and Dynamics:
* * * * * * * * * * * * * * * * * * * * * * *
I was neglecting to go back and look at the idea of the voltage
control signal amplitude and dynamic range while looking at other
things going on in the H&P. The level of the VCO voltage as it is
commonly called since the VFO is now a "voltage controlled oscillator."
I was not seeing it right off. There is a certain needed voltage
amplitude to provide the dynamic range for peak positive to peak
negative control voltage on the diode. The parameter here is
"Capacitance Change Per Volt" with regard to the diode type. {1N4007
diodes do not come with specifications for using them as Varicaps so I
have figured out how to determine that parameter (in note 4.) when
needed for any diode type.}
*This is were a power transistor comes into play since one may need
to up the signal dynamics for their diode type. This also can be used
to change the 74HC74 Q output from a square wave into a more suitable
pulsed waveform.
The level of control voltage swing (or its meter deflection) leads to
looking at the bias voltage on the power amplifying transistor that
feeds the integrator. In the Class C amplifier regions the signal bias
may be too low and so Class AB to, regions near cutoff are best. Here
a 0.5 volt base to emitter bias for most transistors; is a good place
to start with in setting the bias for good operation. In some cases
this may be way in the Class Cutoff Region but also in your circuit
type it may be whats the most useful region. The transistor collector
load resistor value, sets the range of output signal swing. Adjustment
between the two, between the bias voltage and the collector load
resistor, without reaching the max transistor dissipation is required
to get the output's dynamic range or signal swing right for your
Varicap diode, especially if you are using ordinary rectifier diode
types. {A good transistor with more than enough driving power
capability is needed here since this is for Continuous Class Service or
CCS operation. An audio power transistor is fine, since the range of
frequencies in the circuit after IC binary division is only in the 10's
to 100's scale of the hertz frequency range.}
The adjustment of the VCO signal amplitude made all of the difference
in my experiments most recently. This was what I really needed in the
final analysis. A good over all control of the diode. And I really
only needed to use one of the available outputs of the 74HC74.
"In Ball Park Figures"
In my case of using the 1N4007 diode the signal amplitude to the RC
integrator needs to be able to deflect to around 3V DC to 4V DC using a
meter with 10 to 11M ohm input impedance. I am referring to the signal
level going into the RC integrator. The amount of swing or deflection
can be checked by merely turning the VFO tuning dial. The typical
range of voltage across the Varicap diode mentioned in the files at
Hans site is mentioned to be from 1 to 5 volts range across the diode
so this is a ball park figure to use in getting the circuits up and
running. I find that the running signal voltage across the 1N4007
diode when the VFO is not being tuned around, is close to 2.8 volts. I
seem to have better performance by keeping the voltage on the 1N4007
diode above 1 volt, and running mostly at 2.8 volts. Any change great
or small made on the dial quickly results in the VFO coming to a lock
frequency in the new tuned into frequency. I am now happier to know
this and to have got around to looking at it since it gives me what I
am looking for in control.
I now have two reasons why a circuit might not perform right. This I
think is the most important one. It is all about getting the diode's
signal voltage right.
...So to start up a new circuit model, I believe I will keep the VCO
voltage above 2 volts initially so that I know it is of sufficient
level enough to get things initially operating as they should, and
adjust it more if needed after up and running...
* * * * * * * * * * * *
2. Diode Types Compared:
* * * * * * * * * * * *
I have used the 1N4007, the 1N914 and the 1N270 (germanium diode)
which appear to operate similar to each within the range of voltages
though I noticed slight performance differences which are expected but
they are similar in range; so far as I can tell to date. More specific
study is needed.
I tried the 1N4148 and need to look at it again since I believe it is
less on capacitance change per volt and may be then useful at higher HF
frequencies where there needs to be less of a capacitance change to
track the VFOs of those frequencies. So if thats true then that should
be kept in mind.
My view of this is that the voltage across the diode needs to be
dynamic in range for a good range of control to make immediate changes
with respect to positive, and then to negative control signal amplitude
directions and hence immediate changes from lesser to greater
capacitance across the PN junction. ...When a quick correction is
required... Before I was somewhat leery of using too much voltage
swing thinking it might make the capacitance tracking of the VFO over
shoot the frequency. Thankfully I discovered that I was not using
enough voltage across the diode when I began to look in the H&P files
of other's.
I found out that (pa0ksb) Klaus Spargaarten's last circuit idea was
using the same ideas I was trying and so I made a few corrections in my
circuits based on his in regard to the 74HC74. He too had a power
transistor at the Q output of the 74HC74.
The difference seems to be like going from moderate or mediocre
control to better if not ultra control.
The following by "Pat Hawker G3VA" goes into the idea of a Varicap's
capacitance range.
"I suggest the following guidelines for Varicap and integrator
values: the Varicaps pulling range, as before should be 2 to 3 kHz.
Most surprisingly, the integrator time constant should be 400 seconds.
This may seem extremely long, but it is necessary if smooth operation
is to be achieved. It also guarantees a very low level of frequency
modulation on the carrier, which can become a problem with shorter time
constants." I have found reasons to doubt this statement for a time
constant of 400 seconds since this results in very large capacitances
in some models and I have found that too much capacitance can result in
waiting forever for the circuit to settle down.
So far I have used two 1000uF electrolytic caps in the integrator and
it works.
* * * * * * * *
3. Volt Meters:
* * * * * * * *
I was able to use a 2k ohm per volt meter to measure a voltage across
the Varicap and that surprised me that such a low input impedance meter
will do that. It reads up to 5 to 9 volts AC at the integrator's input
node and hence is too wild in its AC swings, in reading the kilo hertz
range signal. It is a cigarette pack sized baby voltmeter that cost
$1.00 from the Dollar Mania Store. Such a meter can be used without
upsetting the circuit much, it will still seek and stay in frequency
lock with such a meter loading the integrator control loop to the
diode. Only the DC voltage scales appear to be similar in reading to a
10~11M ohm input meter in measuring the H&P signal to the diode. So
you can use such a meter to do work in these circuits.
I figure I should use both a 20k ohm per volt and 10~11M ohm input
meter to note measurements with in my final diagrams once they are
fully studied. Till then I will give 10M input measurements. I think
some folk might be using a 20k ohm per volt meter as I sometimes do so
they need to see something they can use in a circuit analysis that give
them a place to start to analyze things with.
* * * * * * * * * * * * * * * * *
4. Diode Capacitance Measurements:
* * * * * * * * * * * * * * * * *
To determine the Varicap range of even ordinary diodes. Connect the
diode to be used temporarily as a fine tuning diode. Using a
potentiometer and your voltmeter (10 to 11 mega ohm meter input if
possible) adjust the voltage and measure it across the diode and notice
also the frequency change associated with it in the VFO and jot down
notes as you check the range of frequency changes made. A range of
voltages from .5 to 3 volts across the Varicap is enough to plot here.
Using the values of your tuned circuits components as a guide, you can
calculate the capacitance per volt of change of the diode. When done
calculating, note at which level of voltages you have what amounts to a
2 and a 3 kHz frequency change. This is the range of voltage you have
to have in signal dynamics to provide good control to the VFO.
*You do not have to specifically calculate the capacitance here to
know how much voltage results in a frequency change of so many kilo
hertz. You can merely plot the frequency versus diode voltage settings
and use that information to analyze the parameters of your H&P control
signal.
* * * * * * * * * *
5. VCO Reset Idea:
* * * * * * * * * *
This idea is for use with models that provide fine incremental tuning
around 15 Hz or less.
If I can turn off the VCO control voltage and replace it with a
reference voltage for a moment, to tune the VFO to a new frequency and
then turn the control voltage back on in that spot, to make use of the
natural resonant frequency of the VFO. I feel the VFO will want to
operate more were it naturally wants to rest at in mechanical
resonance. I have seen the temperature pull the VFO a few hertz over
time. And maybe the actual place the VFO wants to resonate at will
pull the setting there over time also?
I see that if my average VCO voltage is 2.8 volts throughout most of
its range of operations or tuning band spread, then if I switch off the
VCO I must immediately replace it with that voltage as a fixed
reference and then tune the VFO to a new frequency. I have to retain
the capacitance of the diode when I turn the VCO voltage off and so I
have to substitute a voltage to keep that capacitance in the diode. If
I am lucky, when I switch the VCO back on, then I might see it come to
operate near to where I set it without the VCO control. If it wants to
lock somewhere up or down from there, then I can re tune it to the
natural lock frequency and turn on the VCO again and let it rest there.
The idea is to go with the VFO where it naturally wants to oscillate
and not use VCO control voltage extremes to pull the voltage really to
far one way or the other on the diode. Leaving a range for some head
room of operation. The diode then should operate with a good region
above and below the tuned to frequency.
It will end up merely being a switch to turn off the VCO control for
tuning and to turn the VCO on again once tuning is done.
Dan ka9rza
_____________________________________________________________________
PrivatePhone - FREE telephone number & voicemail.
A number so private, you can make it public.
http://www.privatephone.com
Yahoo! Groups Links
<*> To visit your group on the web, go to:
http://groups.yahoo.com/group/soft_radio/
<*> Your email settings:
Individual Email | Traditional
<*> To change settings online go to:
http://groups.yahoo.com/group/soft_radio/join
(Yahoo! ID required)
<*> To change settings via email:
mailto:[EMAIL PROTECTED]
mailto:[EMAIL PROTECTED]
<*> To unsubscribe from this group, send an email to:
[EMAIL PROTECTED]
<*> Your use of Yahoo! Groups is subject to:
http://docs.yahoo.com/info/terms/
