Build A Tracking
Generator for the
Modularized Spectrum
Analyzer
You may have
become
familiar with the MSA, which is presented on theMain Page. This is a guide for
adding a
tracking generator to the 0 to 1000 MHz Modularized Spectrum
Analyzer. Two designs are presented. The original tracking
generator design is shown first, followed by the new TG design.
This Page was Started Aug. 17,
2006. The Original
Tracking Generator
information
was removed from the main page and put here. Updated Sept. 12,
2006. The New
Tracking Generator is advised if the builder is planning to evolve his
MSA into a VNA. Updated Sept. 21, 2006Updated block diagrams to include integrated
amplifiers within the VCO modules. What is a Tracking Generator? A tracking generator (TG) is an RF
signal
generator that will change frequency at the same rate as the Spectrum
Analyzer (SA). When the SA is commanded to look for a particular
frequency, the TG output frequency will be that same frequency.
Here is the basic MSA Block Diagram :
Block
Diagram of Basic MSA
This is the
same Block Diagram as
the one on the main page for the MSA. All construction
information for
the basic MSA is
presented on the Main
Page. The
frequency
agile oscillator is LO 1. VCO 1 will tune from 1013.3
MHz (which corresponds to
a 0 MHz input signal) to 2013.3 MHz (which corresponds to a 1000 MHz
input signal). The input signal is mixed with VCO 1 to create a
difference frequency of 1013.3 MHz. This 1st IF signal is tightly
filtered by the cavity filter and then presented to the second
mixer. Here, it is mixed with the VCO 2 (1024 MHz) to create the
final IF of 10.7 MHz. The final IF is amplified and filtered and
presented to the Log Detector for power to voltage conversion.
The Log Det output voltage is sent to the Analog to Digital converter
to be read by the PC. The basic MSA will read absolute Magnitude
(power) measurements vs. frequency. A signal reference is
required to
perform Magnitude gain/loss measurements. A Tracking Generator
makes a useful signal reference.
Original
Tracking
Generator Block diagram of the MSA with
the added
Original Tracking Generator, highlighted.:
Block Diagram of MSA/TG (original)
LO 3, which is
the
combination of PLL 3 and VCO 3, creates a fixed frequency. It is
the same frequency of LO 1 when the MSA is commanded to "0" MHz.
In this case, the frequency is 1013.3 MHz. As LO 1 is increased
in frequency, the Tracking Generator output is the difference frequency
of LO 1 and LO 3. Since LO 1 tunes from 1013.3 to 2013.3 MHz, the
TG output is 0 to 1000 MHz, thereby,
"tracking" the MSA. There is also a sum frequency output of the
Mixer 3, which is 2026.6 MHz to 3016.3 MHz. A low pass filter is
integrated into Mixer 3, to attenuate the sum
frequencies.
It is labeled as a 1000 LPF, but with the suggested components, the
actual 3 dB point is about 1300 MHz. The TG output level of -30
dBm is determined by the output level of VCO 3 and the amount of
attenuation in the path. In the above diagram, the attenuation is
-37 dB. Higher TG output level can be obtained, but with higher
interference.
The VCO 1 module is shown with an internal
amplifier, feeding Mixer 3. If the VCO 1 module does not have the
dual amplifier capability, an external amplifier must be added.
This is a very simple topology for a tracking
generator. However, it does have the drawback of potential
interference.
Since the LO 3 frequency is fixed at 1013.3 MHz,
the signal can get back into the MSA I.F. chain and cause
interference. It
can be significant if the design and construction is not optimum.
Here is how the signal will
interfere, using some component assumptions:
The Log
Detector is LM3356. 0 dBm is the saturation input level to the
log
det. It's minimum detectable signal is -90 dBm. Any signal
that is not of interest (SNOI) at this point, and greater than -90 dBm,
is deemed
"interference".
The I.F. Amplifier has 44 dB of gain.
The other MSA chain elements (mixers and filters) have a total loss of
-24
dB.
The total gain of the MSA chain is +20 dB (+44 -24); therefore, the
input dynamic range of the MSA is -20 dBm to -110 dBm.
Assuming Mixer 1 has a -7 dB conversion loss, the total gain from Mixer
1 R port to the Log Det is +27 dB. Therefore, a 1013.3 MHz signal
(-117 dBm), at the R port output of Mixer 1, will
create a -90 dBm level at the input to the Log. Detector. A
signal from the TG's LO 3 that reaches Mixer 1, R port, and is greater
than
-117 dBm, is considered a SNOI
and will interfere with real signal measurement.
Now let us follow the 1013.3 MHz signal that is
created by the
TG's VCO 3. It
has an initial power level of +13 dBm. It passes through the 37
dB
attenuator to become a level of -24 dBm.
Mixers are ADE-11X, and have only 25 dB of isolation (data
sheet). Therefore, the -24 dBm SNOI at the R port of Mixer 3,
passes to the L
port at a level of -49 dBm. The SNOI then passes through the 6
dB attenuator and is on the output of the VCO 1 module, with a
level of
-55 dBm. The VCO amplifiers have about 25 dB of reverse
isolation. Therefore the SNOI is -80 dBm on the output of the VCO
1 module. The SNOI then travels into Mixer 1 module, through the
6 dB
attenuator. The SNOI level at the L port of Mixer 1 is now -86
dBm. The isolation of the ADE-11X is -25 dB, therefore,
the SNOI at the R port of Mixer 1 is -111 dBm. This level of -111
dBm will create a signal level of -84 dBm at the input to the Log.
Det. This is 6 dB above the interference level of -90 dBm,
and interference will result. What the display will look like,
is, the noise floor of the Graph will rise by 6 dB. The input
dynamic range will be degraded by 6 dB, resulting in an input dynamic
range of -104 dBm to -20 dBm. In many cases this is not
significant. The MSA still has an 84 dB dynamic range, a 6 dB
degradation.
When I first designed and buit the SSA Prototype
with this tracking generator, there was no interference problem in this
area. I was using the Watkins Johnson, W1J, which has 40 dB of
isolation. When designing the next version, the SSA Board, I
chose the ADE-11X because the mfg claimed at least 40 dB
isolation. Well, in reality, their 40 dB number is a "best case"
data point and, not where we operate this mixer, at 1013 MHz.
I have seen the
ADE-11X isolation range from -18 dB to -28 dB in the 3 units I
have. If the isolation is as poor as -18 dB, then the above
calculations would degrade another 14 dB (7dB x 2 mixers). With
this
case, the SNOI would be -97 dBm at the R port of Mixer
1. This would create a signal level of -70 dBm at the input to
the Log Det. This is a 20 dB degradation to the dynamic
range. Not only are the mixers not guaranteed to have a
specific isolation value, the Pill style amplifiers cannot guarantee
-25 dB isolation. Poor layout and "leaky" connections can also
contribute to degraded isolation.
Here is another significant interference from the
VCO 3's
1013.3 MHz, "SNOI": It will travel through Mixer 3, from the R
port to the I port. Numerically, the -24 dBm SNOI at the Mixer 3
R
port will be -49 dBm at the I port. This SNOI is only 19 dB lower
than the TG output's signal of interest of -30 dBm. In many
cases,
this is not of concern. For example, tuning of filters, where the
filter
under test will attenuate this 1013.3 MHz, SNOI. But, if a wide
band
amplifier or a mixer is the test subject, then significant
intermodulation (IMD) can result. There is a low pass filter on
the output of the TG, but it's purpose is to attenuate harmonically
generated spurs from Mixer 3. It will not attenuate the 1013.3
Mhz. A notch filter would work well here, but TG
frequencies in the 900 MHz area would be degraded.
To decrease the interference in both examples, the
Trk Gen output can be lowered from -30 dBm to a value that causes the
least interference. The -37 dB attenuator could be changed to a
-47 dB attenuator. The SNOI interference would decrease by 10 dB
in both cases. The best method of reducing the interference is to
choose a mixer 1 and mixer 3 that have the highest possible isolation
at
1013 MHz.
PLL 3 is operated in a fixed frequency mode, and
it's phase detector frequency (PDF) must be chosen so that the "N"
steps of the PLL will be a sub-multiple of the LO 3 frequency.
This PDF will probably be very low, resulting with a high phase noise
LO 3. This is not important as a tracking generator with the
MSA. The main consideration is getting the correct
frequency for LO 3. In the above examples, the frequency is
1013.3 MHz. This frequency is determined by the frequency of LO 2
and the center frequency of the Final Xtal Filter (1024 MHz - 10.7 MHz
= 1013.3 Mhz). The PDF of PLL 3 could be 100 KHz.
But, the attractive point of the Basic MSA is, any Final Frequency
Filter can be used. What if the Final Xtal Filter were 9.954 MHz,
with a bandwidth of 2 KHz? The Tracking Generator's LO 3 must be
tuned to: 1024 - 9.954 = 1014.046 MHz. The obvious PDF would be 1
KHz. However, most Integer-N PLL chips can not divide that high
(1014.046 / .001 = 1,014,046). For sure, the LMX 2326
cannot. A very clever way to overcome this shortfall is to
use a PDF that is not quite so obvious. Since this is a fixed
frequency LO, it doesnt need to step to any other required
frequency. Use a PDF that is a numeric sub-multiple of both the
output frequency and the Master Clock Frequency. In the above
block diagram, the master clock is 64 MHz. What PDF will be a numeric sub-multiple of both
1014.046 MHz and 64 MHz ? With some mathmatics involved, a PDF of
.070098576 MHz will (closely) meet the requirements. The PLL's
divide by N counter will be 14466 (1014.046 MHZ / 14466 = 70,098.57597
Hz) and the divided by R counter will be 913 (64 MHz / 913 =
70,098.57612 Hz). The two frequencies are not exactly the
same. They are .00345 Hz different. Using the following
formula the output frequency of the LO 3 is calculated : VCO =
(clock/R) x N or (64 MHz/913) x 14466. This equals
1014,046,002.19 Hz. This is 2.19 Hz higher than the wanted
1014.046 MHz. However, 2 Hz is well within the bandpass of 2 KHz,
the BW of the Final Xtal Filter. The actual output frequency of
the Tracking Generator will always be off by 2.19 Hz. The PLL 3
can be designed with a loop filter for a PDF of 70 KHz. This 2 Hz
offset will not be a factor when tuning filters, or for gain/loss
measurements. As long as the final I.F. frequency is within the
bandpass of the Final Xtal Filter, all is well. This also holds
true when this Tracking Generator is used with the MSA/VNA, which is
discussed on a separate page.
If a Fractional-N PLL is used for PLL 3, there is
the same consideration of tuning LO 3 for the correct frequency.
But, a Fractional-N PLL will have a PDF that is 16 times it's step
frequency. If a step size of 1 KHz is required, the PDF will be
16 KHz, and this is easily accomplished with the LMX 2350 or LMX 2353.
Here
are some suggested loop component
values for the original PLL 3, for different PDF's.
PLL 3
Used as
Icp PDF
C1
R1 C2
R2
C3
ADF 4112
IntegerN 5.0
ma 5 KHz 120
nf 620 1.5
uf 1.1 K
47 nf
LMX2350/53 IntegerN 1.6
ma 5 KHz 39
nf 2.0 K .47
uf 3.6 K
15 nf
LMX2326 IntegerN
1.0
ma 5 KHz 27
nf 3.0 K .27
uf 5.6 K
8.2 nf
LMX2325 IntegerN
5.0
ma 10 KHz 68
nf 620 .68
uf 1.1 K
22 nf
ADF 4112 IntegerN
5.0
ma 10 KHz 68
nf 620 .68
uf 1.1 K
22 nf
LMX2350/53 IntegerN 1.6
ma 10 KHz 22
nf 2.0 K .22
uf 3.6 K
6.8 nf
LMX2326 IntegerN
1.0
ma 10 KHz 12
nf 3.0 K .15
uf 5.6 K
4.7 nf
LMX2350/53 FractionN 1.6
ma 80 KHz 2.7
nf 2.0 K 27 nf
3.6 K 820 pf
Notice that the LMX2350 and LMX2353 can be operated as a
Fractional-N PLL. This allows the PDF to run at a higher
frequency, and yet, step in 5 KHz increments.
Click here to see a schematic of PLL 2 using the LMX 2353.
Here
is a schematic and suggested values to build Mixer 3. It can be
built
on the Generic Mixer Module Board. The mixer can be an ADE-11X or
a
suitable mixer with the same footprint. Resistor values are shown
for 6 dB, 16 dB, and 37 dB pads.
Here is a schematic and suggested values for an amplifier with an
integrated high pass filter.
This amplifier should have an average gain of +20 dB
from 800 Mhz to 2000 Mhz, and should saturate with an output of +13
dBm. 5 volt operation is not a necessity. The 30 ohm bias
resistor can be changed for other voltage regulators. The
SGA-4586 is a suggestion, and a variety of other amplifiers will work
well. Shown, is a single stage amplifier board that Cash Olsen
has produced. It has room on the input for a resistive attenuator
or a pass filter.
The PLL 3 Module is identical to the MSA's PLL 2
module and can be seen on the main page. Likewise, the VCO 3
Module is identical to the MSA's VCO 2 Module.
New Tracking
Generator Block diagram of the MSA with
the New Tracking Generator, highlighted:
Block Diagram of MSA/TG (new)
This diagram highlights the area of the new Tracking
Generator design. It is substantially
different than the old TG design, and has improved characteristics.
The similarity of this TG's LO 3 and the basic MSA's
LO 1 is not coincidental. They are identical, with one
exception. The DDS 3 in the TG is serially commanded, the DDS 1
is parallel commanded. To construct the LO3 chain, go to the Main Page and duplicate the LO1 chain.
LO 3 will command from 1024 Mhz to 2024 MHz.
The output from VCO 3 is amplified to a level of +13 dBm, and used as
the LO drive for Mixer 3. Medium level mixers only require about
+7 dBm for LO drive, thus the mixer module has space for a resistive
attenuator. The fixed frequency of 1024 MHz, from
LO
2, is input to the R port of Mixer 3. Mixer 3 output (I port)
will range from 0 MHz to 1000 MHz, thus, "tracking" the MSA. The
LO 1 frequency and LO 3 frequency will always have a differential
frequency equal to the frequency of the Final I.F. However, the
software can be told to "offset" the Tracking Generator output.
This is particularly useful when injecting a certain frequency into a
test device and a different frequency is the output of that device
(such as, a transciever). Notice, the
Mixer 3 module also contains a 1000 MHz low pass filter on the
TG
output. If the user is interested in using the higher frequency
mixing products, this filter can be deleted.
Like the Original TG,
this New TG will produce interference, but in a different way. There are two interfering
frequency sources
that have the potential of interfering with the MSA chain.
The
first is the Tracking Generator output frequency created by Mixer
3. Some of the internally generated TG Output
energy (0 MHz to 1000 MHz) will travel out of the R port of Mixer 3
back to VCO 2. The TG frequency will enter
Mixer 2, via the L port, causing some interference with the mixing
action of Mixer 2. This interference would be noticeable, only
when the
Tracking Generator is commanded to the same frequency as the Final I.F
(10.7 MHz). However, since the buffer amplifiers have excellent
isolation at low frequencies, there should be little effective
interference.
The second
interfering source is a bit more
complex. And, I didn't notice it in my original test
environment. This source is a combination of LO 1, LO 3, and
Mixer 2. Here is how it is created:
As stated before, the mixers have a certain amount of port to port
isolation. Mixer 1 allows LO 1 to travel from its L port to the R
port. The cavity filter will allow LO 1 frequencies from 1013.3
to
about 1015 MHz to freely pass to Mixer 2. The roll-off of the
cavity filter will attenuate frequencies above 1015 MHz.
Mixer 3 will
allow the L port energy of VCO 1 (1024
to 2024 MHz) to pass through to the R port. This signal will
travel back to VCO 2. Although there is isolation
in the LO 2
module, the signal will pass through the VCO 2 module and be present on
the L port of Mixer 2. The
two signals, 1013.3 MHz from VCO1/Mxr1 and 1024 MHz from VCO3/Mxr 3 are
combined in Mixer 2 to create 10.7 MHz. As the MSA/TG is swept,
the two frequencies are always 10.7 MHz in difference. As the
MSA/TG is swept above the bandwidth of the cavity filter, the
interference decreases to a level that is of no concern. The
bandwidth of the cavity filter will determine the cut-off frequency of
interference. In my MSA/TG the interference does not become
unnoticeable until the MSA is tuned above 3 MHz.
I am
showing an output of -10 dBm but that is dependent on the Mixer 3
loss, and the originating power level of the 1024 MHz sample from LO
2. This is much more power than is usually required to make
reference magnitude measurements. Therefore, the attenuator in
the Mixer 3, R port could be increased to reduce the Tracking Generator
output and, consequently, reduce the amount of interference.
As previously
stated, the LO 3 chain is identical to the MSA's LO 1 chain. DDS
3 will "steer" PLL 3 as it's Reference Clock. The DDS 3 is
commanded serially from the J4 port of the Control Board. There
is a minor modification to the original
DDS module to allow for serial commanding. Two resistors are
added on the module's control connector, a pull up and a pull down
resistor. See the DDS schematic on the main
page, for the modification.
The VCO 3 has
two outputs, a high level output (+13 dBm) and a low level (-3
dBm). We are using the high level for LO injection for Mixer
3. The low level output can be loaded into 50 ohms, or used for
other purposes.
My main purpose for the secondary output will become apparent when you
read the web page on how to convert the MSA into a Vector Network
Analyzer.
This TG can be
used as a stand alone UHF synthesizer to create frequencies from about
950 MHz to 2050 MHz. The output can be multiplied, or it can be
combined
with other frequencies, using Mixer 3, R port. Mixer 3 can be
modulated from the I port (normally
the output port) and the output taken from the R port.
The Tracking
Generator is controlled by software commands via the MSA Control
Board. PLL 3 is controlled from Control Board J3. DDS 3 is
controlled from Control Board J4. Go to the main page to see a schematic
diagram of the MSA Control Board. Use your web browser's "Back"
button to return here.
The Tracking
Generator is never actually turned off. When the software
commands the "Track Gen Off" it simply commands the TG to idle at a
frequency that will not interfere with normal MSA operation. At
the present time, that frequency is 970 MHz. And, it can be
changed in the user's portion of the software code. When the
operator commands the "Track Gen On", the TG will
command to the same frequency that the MSA is tuned to.
The following is the same schematic to build Mixer 3 for the Original
TG. It
can be used for the New TG by changing the resistor values.
The VCO 3 Module is
identical to the MSA's VCO 1 Module, and can be seen on the main
page. Likewise, PLL 3 Module is identical to MSA's PLL 1
Module. The DDS 3's Square Ampl Module is the same as the MSA's
DDS's Square Ampl Module. DDS 3 Module is identical to the MSA's
DDS Module, with the resistor modification.
I hesitate to
recommend one TG over the other.
Both have their
pros and cons. The software will accomodate either.
However,
I do like the idea that the New TG can be
used as a general purpose, stand-alone signal generator. Also, it
less self generated interference than the old TG. Added note 9-12-06: After
much testing of the VNA option, I will change my recommendation.
The New Tracking Generator is the better choice for use with the
VNA. The reason is, the VNA phase measurements are truely
accurate, only, when the input to the MSA is 50 ohms. The Mixer 1
input and the Tracking Generator Mixer 3 outputs are not 50 ohms over
frequency. They can be brought closer to 50 ohms by adding 10 dB
pads to the ports. With the high level of -10 dBm output of the
new TG, there is still plenty of power available, after the addition of
the
10 dB pads.
I would like to point out to the reader, that, the
absolute frequency of the MSA and the Tracking Generator Output will
not be, precisely, the same frequency as the software command, even if
the Master Oscillator is precise. This is due to the DDS as a
steering frequency, for either MSA's LO1 and/or TG's LO 3. The
DDS is quite accurate, but it is not "phase locked" to the Master
Oscillator. It is treated as a frequency divider, and the
software will command the DDS to the closest significant bit.
Therefore, it's output frequency will be accurate within one bit, or
0.014
Hz. This seems pretty good, but remember, the DDS output is
"multiplied" in the PLL by anywhere from 93 to 190 times. This
equates to a frequency resolution of 1.3 Hz to 2.6 Hz when the PLL is
tuned from 1000 MHz to 2000 MHz. For use in a Spectrum Analyzer
or properly designed into a Vector Network Analyzer, this is not of
concern. But, it is just a general warning that the Tracking
Generator is not a precision frequency source. With certain
software conditions, it can be made a precision step frequency
source. However, the MSA does not do that. I just wanted to
make this statement before someone emails me and says, "Hey,
Scotty. You jerk, I commanded the MSA to zero MHz and the
Tracking Generator has an output of one fifth Hertz, what gives?"
Scotty
Sprowls, email is wsprowls (at) yahoo.com
End of page (so far).