A two tone test generator is a very useful, simple circuit that you can build to test your circuits for distortion.
What is it
The principle is very simple, two crystal oscillators generate two signals that are close to each other. Both the oscillators are well isolated from each others output and then mixed together in a 6 db hybrid coupler. The combined output is filtered to remove the harmonics. Here is the circuit. It has just four transistors in it. You can build one in an evening.
How does it work?
Let’s imagine that the two crystals frequencies of your two tone generator (let’s call it TTG from now on) are 14 Mhz and 14.16 Mhz. They are have a difference of 160 Khz between them. If you feed the output of the TTG to an amplifier, you expect just those two frequencies to show up in the output. That only happens in an ideal world..
All real world amplifiers, mixers, even filters show distortion. The distortion will result in spurs (spurious signals) at 13.8 Mhz and 14. 4 Mhz as well.
Mathematically speaking given two signals are f1 and f2, the spurs will appear at 2f1 – f2, and 2f2 – f1.
These are called third-order distortions. We will deal with the math later. Enough to say for now that,
You inject the two tones, each at -10 dBm into a circut.
Measure the output tones. An amplifier will increase the output by it’s gain. In a test case, the amplifier output was +7dbm, the gain was 7 – (-10dBm) => +17 db
Measure the tone to distortion ratio. In this case, the distortion product was at -23dBm. Which is 30 db below the tone ( +7dbm – (-23dbm) => 30 dB)
The Third Order Output Intercept Point(OIP3) is output power + half of distortion ratio. In this case, it was +7dBm + 30dB/2 = +22 dbm.
The Third Order Input Intercept Point(IIP3) will be input power + half of the distortion ratio. In this case, it was -10dBm + 30dB/2 = +5 dbm
Each of the two crystal oscillators is a conventional crystal oscillator. The unusual part is that we take the output from a small (10 ohms) resistor that is between the crystal and the ground. This is a neat trick where we use the crystal as a crystal filter as well. This where the oscillator noise is the lowest. A small value (100 ohms) preset allows you to fix the exact power output at -10 dBm at the BNC jack.
The amplifier that follows is common-base amplifier, that provides very good input-output isolation. This is important as the outputs are mixed further down in the circuit and we don’t want one of the outputs leaking back into another oscillator and intermodulating it. We need almost zero inter-modulation out of this circuit so that the only intermods you can measure are from the circuit that this TTG will drive.
The 6 db pads at the output of the amplifiers provides stable driving impedance to the hybrid coupler. The hybrid coupler is deceptively simple circuit. It mixes the two oscillator outputs in such a way that there is (theoretically) no reflection from one oscillator output to another. If the hybrid coupler reminds you of a return loss bridge, your guess is correct (Not now, not now).
Any amplifier output has harmonics. These harmonics can interfere with the measurments. You can appreciate that if you remember that the third order intermods are caused by intermodulation of the second harmonic of one tone with another tone’s fundamental (2.f1 – f2). Hence, an extensive, 3 section Chebyschev low pass filter keeps the harmonics down.
As you can see, the circuit is more elaborate than two simple crystal oscillators with their outputs tied together. Test instruments can never be over-engineered. It is a good idea to throw as much effort and care into them as possible.
I constructed this in a small tin box that they gave on a flight with nuts in them. (No, not free, you have to buy the nuts). I built it ugly style using some stryoflex (also called polystrene) capacitors in the LPF The disc ceramics of the plain variety proved quite lossy.
Depending upon the crystals you have, you may have to change the low pass filter values. The ARRL Handbook or the Experimental Methods in RF Design (Also an ARRL publication) give you the formulae to do it easily.
With SW1 switched off, only one oscillator will work. Connect the output to a 50 ohms load and with an oscilloscope probe, adjust the RV1 until the peak to peak RF appears to be 200 mV.
Now, switch on both the oscillators, and ajust the other oscillator until the combined peaks measure 400 mv. You adjustments are done. See the pictures. Keep the oscilloscope time base to about 1usec to get the peaks all bunched together. You can also see the two tones generating a PSK31 like waveform. The peaks are when the amplitudes add up and the nulls are where the amplitudes cancel out.
Doing it without a spectrum analyzer
The two tone generator assumes that you have a way of measuring the two tones seperately, that usually means a spectrum analyzer. But you can use your amateur radio receiver with a step attenuator. Here is how:
Turn off the AGC. It is a good thing that many homebrew receivers don’t have it to begin with!
Connect the two tone output of the circuit to the receiver through an attenuator. Keep the attenuation to at least 40 db or you will blow the receiver!
Tune to the spur. Note the speaker audio level with an oscilloscope.
Now, tune to the tone. It will be very loud. Keep flip on more and more attenuation until the tone is just as loud as the spur. Measure the audio level on the oscilloscope. Now, you have the IMD ratio. You can back calculate the OIP3.
Given a signal x that goes into an amplifier, the output should y such that
y = ax (The ideal case, a is the amplification factor)
Instead, we get,
y = ax + bx2 + cx3 + dx4 ….
This is not nice. We would ideally like b and c and d etc. to be zero so that just get ax remains.
Our two tone generator helps in measuring the cx3. This is the most troublesome distortion because it produces spurs that are close to our signals. The spurs that are further away are filtered by the front-end filters or the transmitting low pass filters. Why is that so?
Let’s imagine that the signal consists of two signals actually: a and b.
Now, the term cx3 becomes c(a + b)3
If you remember your high school maths,
(a +b)3 = a3 + b3 + 3ab2 + 3a2b
The last two terms are multiplying the signals like : 3 . a . b . b and 3 . a . a . b
We know that when signals multiply, they generate ouptut frequencies at their the sums and differences.
Multiplying signals 3 . a . b . b will generate frequencies a ± b ± c which would be:
a – b – b = a – 2b (this is our offending spur) a + b – b = a (doesn’t matter, it is just a) a + b + b = a + 2b (this spur at 3 times the frequency, it will get filtered out) a – b + b = b (doesn’t matter, it is just b, again)
and multiplcation of 3. a. a. b will produce the frequencies at a ± a ± b which will be:
a – a – b = -b (just b) a – a + b = +b (just b, again) a + a – b = 2a – b (this is the offending spur) a + a + b = 2a + b This is three times the frequeniecs, it will get filtered out)
This is the horrible secret of the third order (beacuse it is the third term) distortion and why we struggle to overcome it. It can never be defeated but it can be controlled. Pour some more current into the amplifiers, Jake!
An easy to build spectrum analyzer for the homebrewers
Show me what you got!
For those who can’t wait to see the circuit diagram, here it is (the Arduino hookup is a little way down on the page):
Analog Circuit of the Specan
(A better, higher resolution diagram can be downloaded here: from my G Drive)
If you don’t want to read this long long write-up, here is a summary:
This spectrum analyzer costs less than $100 to make.
It is just a superhet radio that has a log amp where the speaker should have been.
You can hack it. It has an Arduino.
Works as well as a commercial spectrum analyzer. In some ways, even better.
Works only upto 70 MHz (good enough for HF work, for VHF just remove the LPF)
It has just two resolution bandwidths : 300 KHz and 1 KHz. Quite enough for our kind of radio work.
You have to align the filters. Easily done, but still, you have to do it.
You use parts of the specan to align other parts. In this sense, it aligns itself!
Works with a PC instead of an oscilloscope for display.
It has a built-in LCD for quick readings
+15dbm IIP3, spur free dynamic range of 80 db at 1 KHz resolution, 70 db at 300 KHz.
Tunable in 100 Hz steps.
All under GPL v2.0 license.
Why do I need a spectrum analyzer?
Most of the homebrewers have a very basic measurement setup. Often we get by with just a VOM, an RF probe and maybe a frequency counter. The more ambitious experimenters aspire to acquire an oscilloscope – an expensive item. These instruments are quite useful but not enough for RF tinkerers.
Often times, we homebrewers end up using the following:
An RF Probe
This costs nothing to make. Coupled with a dummy load, it can provide reasonable measurement of power levels of RF signals from a milliwatt upwards. But it offers no insight into the frequency of the RF signal
A Frequency Counter
For a few dollars, we can program a PIC and an LCD display to display the frequency of an RF signal. However, the frequency counter does a very poor job. Imagine feeding a diode modulator’s output to the frequency counter. What would it read? the Modulation frequency? The carrier frequency? The upper side-band or the lower side-band? Your guess is as good as mine. When it comes to RF signals that have multiple frequency components like most of our RF circuits do, the frequency counter becomes useless. At best, you can use it to measure the frequency of your new VFO’s oscillations, but that’s about all you can do with it with any accuracy.
A few hundred dollars buys you an oscilloscope that can show you the exact waveform. Well, nearly exact. You could look at a sinewave and say that it is clean or not. But, consider this, a power amplifier with a second harmonic that is just 40 db down (and hence illegal to use on the air) will not register on the oscilloscope waveform at all. That is a variation of only %1 of the waveform from its expected shape. You won’t see that on your beloved ‘scope. You will more!
We, radio amateurs deal with signals in terms of their frequency spread. We talk about how much the carrier suppression is, how clean our signals are, how distortion free our receivers are. More often than not, we luck out with some circuits and fail to understand why others don’t work.
Enter the spectrum analyzer – it does the job of all the above and more. It unrolls even weak signal into its component frequencies and shows you the precise strength of each.
With it you can :
Show how well your DSB modulator is suppressing the carrier, down to 0.2db
Locate your VFO’s frequency and it’s cleanliness. With a little effort, even the phase noise.
Plot your antenna’s VSWR across the HF spectrum.
Check your QRP transmitter’s harmonics and other emissions.
Measure the input and output impedances, well actually, the impedance mismatch.
Measure distortion and overloading of the mixers, amps, receivers, transmitters.
Plot any filter’s passband shape.
The professional spectrum analyzers are quite expensive. The least expensive ones cost a few thousand dollars. The better ones cost tens of thousand of dollars. A few cheaper ones are now available – but they are more suited to tuning up commercial radios rather than lab quality work. The only alternative for the homebrewer is to brew their own.
We are fortunate! It is now possible to build one yourself for less than a hundred dollars in new parts. That’s a fraction of what an oscilloscope costs and it is entirely homebrewed.
If there’s one investment you need to make in test equipment, it is this. I didn’t know that until I built one.
The November 1998 QST carried an amazing series of articles by Wes Hayward, W7ZOI and Terry White, K7TAU. You can read them both here and here. They lay out the details of a spectrum analyzer that could be built at home. It is a classic. Even if you are not interested in spectrum analyzers, this series of articles forms the most compact introduction to high performance receiver design I have ever seen.
The W7ZOI/K7ATU analyzer made very intelligent trade-offs. It limited itself to HF work and it had just two resolution bandwidths. The 300 KHz bandwith is useful to see the entire HF spectrum to pick up spurs and harmonics and the 8 KHz bandwidth was for measuring IMD. This made the W7ZOI/K7ATU spectrum analyzer as complicated (or simple) as a simple SSB rig.
Almost 15 years later, I attempted to replicate their spectrum analyzer : improvising where I didn’t have the original parts and redoing others where newer technology was available. The really big change was incorporating an Arduino controller with an Si570 as the local oscillator. This article describes a new version of the W7ZOI/K7TAU spectrum analyzer, now simply called the Specan. I have included the small details that are essential to building a high performance spectrum analyzer.
The differences between my design and the original are as follows:
The local oscillator is an Si570. This allows us to tune finely down to 1 Hz steps.
Having a stable, finely tuned spectrum analyzer allows for very narrow resolution bandwidth. You can now measure the distortions inside the pass-band of a transmitter.
The Arduino also handles a front-panel LCD display and tuning knob making it useful as a measurment receiver as well.
Instead of an oscilloscope, the spectrum is plotted on the PC. This saves the cost of an oscilloscope.
The IIP3 of the Specan is about 10 db below that of the QST analyzer due to the ADE-1 mixers. However, narrower resolution bandwidth of 1 KHz gives us the desired 80 db spur free dynamic range for the close-in distortion measurments.
How it works
The Specan is actually a very simple but robustly built receiver. it is, in essence, a double conversion superhet receiver with 112 Mhz and 12 Mhz Intermediate frequencies. The first mixer uses an Si570 as the local oscillator. The second oscillator is a crystal controlled at 100 MHz; built with a common microprocessor crystal of 20 MHz. Unlike most radio receivers, the second IF has two filters : a narrow band crystal filter with 1 Khz bandwidth and a wide band LC filter with 300 Khz bandwidth. The detector converts the tuned signals into a log scale. The detector output is a decibel measure of the incoming signal.
The Specan is controlled with an Arduino board. The Arduino controls the Si570, handles the front panel, talks to the computer over the USB port, reads the detector and switches the filters.
In a very simple usage, the Specan can be tuned around like a regular radio. Instead of listening to the signals, you read their strength on the LCD display.
You can switch between wide and narrow filters : using the wide filters to hunt for signals and then use the narrow filter to locate and measure them with greater precision. This in itself is quite a useful function. It is possible to measure intercepts, gains and harmonics without using a computer.
Under a PC’s control, the Specan can be made to step through any sequence of frequencies and plot them on the computer screen. The Specan understands a handful of text commands sent over the USB port of the Arduino.
The Specan also accomplishes a long standing personal goal – to make the homelab at VU2ESE entirely homebrew. The Specan can easily replace our aging TEK 465 scope and do even more.
This Specan can be built for far less than hundred dollars. Though it needs quite a few evenings of work. Construction is straightforward.You won’t need any special tools. Modules of the Specan are used to align the rest of it! The Specan serves as its own set of test and alignment gear!
What you need to build one
The Specan must be built one stage at a time. Over evenings the builder will bring each stage to life. Each module is calibrated using the previously built modules. As high-performance is a requirement, each stage needs extensive shielding.
Though the circuit is only as complicated a simple transceiver, each stage has to be built individually, tuned up and shielded properly. It is all done quite easily. It just takes a bit of effort to take care of the small things and not rush through it. You are building an instrument that will last a long time. Take your time!
You will need temporary access to a well calibrated 10 Mhz receiver or a frequency counter for a calibrating the Si570. The rest of the tune-up is done by the Specan itself.
A plug for another piece of test gear that I designed – The Sweeperino I highly recommend building the Sweeperino. It consists of the three easiest to build modules of the Specan (the Arduino, the power meter and the Si570). Hook up these three parts as a digital controlled sweeper. The Sweeperino will help you align the Specan. It is documented here.
How well does it work?
It has an input intercept of +15dbm and a display range of more than 80 db at -20dbm level when scanning with the narrow band filter and 70 db with the wide band filter. It works from DC to 70 Mhz. This means that you will be able to see your HF transmitters harmonics even if they are 70 db below the carrier : quite a useful thing. You will be able to measure the IIP3 of mixers, amplifiers with a few simple attenuators. If all this is confusing, think this way : this equipment will be enough for you to design and measure the performance of all your HF rigs, antennas and linears.
How does it work?
The Ardiuno that controls the Specan is a very flexible microcontroller board that you can program in simple C language and download through a USB port from your PC. It costs just a few dollars. This build used a free version of it called the Freeduino. The Arduino provides four functions
Tunes the Specan by controlling the frequency of the Si570 as the local oscillator
Switches between the two IF filters of 300 KHz and 1 KHz under program control
Handles manual tuning, display and the function switch on the front panel
Communicates commands and data with the PC over USB in a simple text format
The rest of the Specan is just a receiver that covers from 0 to 70 Mhz in two bandwidths : 300 Khz and 1 Khz. Unlike a regular receiver that demodulates the signal to audio, the Specan converts the amplitude of the tuned signal to its decibal value through an amazing little 8 pin chip – the AD8307 from Analog Devices.
The Specan’s tuning is done by setting the 1st oscillator frequency. The first oscillator is another 8 pin chip – the Si570 that has a spur free, low phase noise output all the way to 200 Mhz. It costs less than 15 dollars here. The Si570 is in turn controlled by the Arduino through a two digital lines following a protocol called I2C. You don’t have to figure this out, Arduino and Si570 already understand it, we just hook it up and use it.
Here is the block diagram of the Specan, it is just a usual double conversion receiver
The incoming signal is filtered through a 4 section low pass filter before entering the 1st Mixer. We used a two dollar ADE-1 mixer from Minicircuits. You could choose a higher level mixer in it’s place, but this is what was available in the junk box so this is what was used. The 1st mixer is driven by a local oscillator that consists of the LVDS version of Si570 amplified by an MAV-11 MMIC. The mixer is followed by a strong amplifier (AMP1) that also consists of another MAV-11. This amplifier helps terminate the first mixer properly. (Yes, we did try bypassing this amp for the heck of it, it brought the input intercept down to 0 dbm, you could substitute it with a 6db attenuator pad with a drop in the noise figure). The AMP1 also provides a useful gain of 10db. The signals are then filtered through a band pass filter centered around 112 MHz and mixed in a second ADE-1 mixer (MIX2) with a 100 Mhz crystal oscillator. The output of the second mixer is in HF region, two low-cost 2N3904s are used in parallel as a high intercept post-mixer amplifier (AMP2). The 12 Mhz IF signals are routed through either a narrow band crystal filter or a wide band LC filter. The IF is amplified by the second IF amp (AMP3) to a level suitable for the AD8307 log detector. The AD8307 generates a DC voltage between 250mv and 2.2V as the log of the RF input with a 25mv per db ramp.
Building the Specan
(This photograph is with the shields removed. the Si570 shown here was CMOS version without the MAV11 amp)
Here are a number of things you must know before starting to build the Specan
The Specan is should be built on a large unetched board of 12″ by 12″. The original was built on a smaller, congested board.
Each module is assembled on smaller unetched boards and tested. They are tack-soldered onto the large board by small wires.
The front panel is also made of unetched copper clad board as well. The front panel is soldered to the base board.
Many of these modules need to be shielded. The original article recommended Hammond boxes for shielding. These are expensive. Instead, the Specan builds shields around each of the module by soldering walls of 1 inch high copper sheeting available in the hardware stores. The sheeting is easily cut with scissors.
A lid is fashioned out of more of the same sheeting and tack soldered to the walls of the shield. It works very well.
Many of the modules use SMD components. In the absence of a PCB, a hobby knife can be used to cut pads outof the un-etched double side copper clad board. The components are mounted between the pads and between the pads and the ground plane.
The ground plane on the top side is connected to the bottom plane by drilling a number of holes between them and soldering small pieces of wire.
Small pads are cut at the edge of the module boards for power and signal lines. The copper shielding has small apertures cut out to allow the lines through the modules. These apertures should be just wide and high enough to avoid shorting these pads to the ground. See the close-up pictures.
Keep the LM1117 regulator far away from the Si570 to avoid tempurature drift. Use a well regulated linear supply with the Specan.
In case you don’t hear it right that last point, let me repeat i: Use a separate, regulated, linear supply. DO NOT use the same power supply as the one for the device under test. I’d recommend building one inside the cabinet of the Specan. I didn’t do it and this build suffers from RF leaks when I accidentally use the same power supply for the units under test and the Specan.
Mount the LPF module such that the shield directly solders to front-panel to avoid picking up the RF from Arduino and the Si570 wires.
1 – Hooking up the Arduino
To use the Arduino in the Specan you don’t need to know any programming! You just download the Arduino code from the net and upload it to the Arduino.
We should all be very grateful for the Arduino board. It is a two dollar microcontroller on a board. The really amazing thing about the Arduino is that you program it in simple C language without messing about with assembly language. For writing these programs, a very easy-to-use IDE (Integrated Development Environment) is freely available from www.arduino.cc. The entire program of the Specan for the Arduino is available under General Public License. You can download it from https://github.com/afarhan/specan.
The Arduino has a two rows of female connectors on either side. Here is what it looks like:
The Specan uses Pin 2 to Pin 7 from the top row of connections and Analog pins 0 to 5 in the lower row (see the above picture). In addition to that, the USB port is used to talk to the PC. The Arduino is usually set to take power from the USB port. We want the Arduino to work even without a PC connected, hence we power it from an external 5v regulator. This is how you prepare the Arduino :
Move the PWR jumper on the Arduino from USB to EXT.
Mount the Arduino on one corner of the Specan’s box. Our Specan used 1/2 inch brass stand-offs to mount the Arduino.
The mounted Arduino should have its USB port peek out of the front panel through a small aperture (to avoid RF leaks)
The display should be mounted right on top of the Arduino so that the wire lengths are short.
The Arduino takes a general purpose board with male pins that mate with the sockets on the Arduino. You only need two rows : for pins 2-7 and pins A0-A5. This general purpose board also has the 5v regulator for the Arduino.
Solder or bolt a shield around the Arduino and the LCD display. A wall of thin copper sheeting (used in transformers) was soldered around the Arduino to prevent the RFI from the Arduino from getting into the rest of the Specan. Small apertures at the bottom of the shielding were left out to allow the control and power lines to run to the Specan.
Here is circuit for the Arduino’s interface with the Specan.
Once your Arduino is all wired up, you can load the Specan program into it. Here are the steps:
Install the Arduino IDE on your PC from www.arduino.com.
Download the specan.ino file from here.
Connect an Arduino through the USB cable to the PC and power it up.
From the Tools > Board choose the appropriate model of Arduino board that you have.
From Tools > Serial Port choose the serial port of the Arduino.Choose the specan.ino from File>Open menu of the Arduino IDE and press the upload button.
Keep the contrast control at the half-way mark. You may have to turn it up or down to get the LCD display going.
That’s all you need to do. The Arduino is now ready to control the Specan.
When you power up the Arduino you should see “Specan v0.2” show up on the display.
The Arduino controller is under the veroboard. The log amp is on the right
2 – Building the Si570 Oscillator (Use only the LVDS version!)
In many ways, the Si570 oscillator is a key component of the Specan.
It is a reasonably low noise oscillator
It is easily controlled by a microcontroller like the Arduino
It can be soldered dead-bug style just like an LM386.
It works into the UHF range!
Unlike a DDS, it has no spurs
The circuit is very simple. Two lines of SCL and SDA control are used to program the frequency from the Arduino. Beyond that, there are just the power and ground connections. However, to keep the noise and the spurs down, you need to take some precautions.
Assemble the Si570 oscillator on a small one inch by one inch board.
Solder the Si570 upside down. The chip is held in place by a very short wire from pin 3 to ground and the two powe supply bypass caps mounted as close to pin 6 as possible.
Score off small pads (prehaps with a dremel) at the edge of the board to take in the lines from the Arduino and power.Make a separate pad for the RF out.
Make sure that the power to the Si570 board is indeed 3.3v (and not 12v!).
Mount the LM1117-3.3v regulator separately with sufficient heatsink. It will consume more than 1 watt of continuous power. The extensive bypassing on the power line of Si570 is to keep the digital noise of the Si570 from going out and the Arduino chatter from coming in.
Strap down the wire bundles going to arduino and power with small loops of wire soldered to the board and twisted tightly. Keep them away and perpendicular to the RF signal paths.
Testing the Si570 module
Attach the wires to the Arduino, power line and switch on the two modules.
Check that the pins 6, 7 and 8 report 3.3v.
Load the Arduino with Sweperino sketch. Download it from https://github.com/afarhan/sweeperino
Move the tuning knob around, you should see the frequency of the display change.
Set the Specan’s display to 10.0000 Mhz
Connect a frequency counter or monitor the output on a radio. Note the frequency. It will be close to 10 Mhz.
Take the actual frequency measured by your counter/receiver and plug it into this formula: (10.000 x 114326895)/(Freq in Mhz)
Now, open the Arduino’s sweeperino.ino and scroll down to the line that reads
#define FREQ_XTAL (114326895) and replace the number with the newly calculated number (it should have 9 digits as well). Upload this file again
Now, restart the Specan and tune up to 10 Mhz. The frequency should be spot-on in the receiver/counter.
Now, open the specan.ino source code file and replace the #define FREQ_XTAL with the same value as above, upload specan.ino to the Arduino again.
Your Si570 is now frequency aligned and ready for use. The 6 db pad in the output provides a well defined 50 ohms output impedance.
The Si570 without the MAV11 (these pictures were taken before I switched to the LVDS version of Si570)
3. Building the 8307 Log detector
The Specan has to display the power of RF signals in decibel scale. A great challenge in building a spectrum analyzer is to build a perfect log detector that can convert the RF power to a dbm readings. This problem is now eased with the availability of the AD 8307. This part is available for a couple of dollars. It provides a 90 db range with 0.2db accuracy! It is pretty simple to make. Here is the circuit :
Assemble the log amp in a small board using SMD components.
Cut pads into the unetched copper to take pins of the AD8307.
The regulator has to be on the board.
Bring out the RF in, Output and the power lines to pads at the edge of the board.
Solder a 1″ high ‘wall’ of copper sheeting all around the module, leaving small apertures for those tabs. A bit like dog flaps in the door.
Now, power up the AD8307 and test it.
With nothing connected to the RF input the output should read between 0.25v and 0.4v.
Connect the AD8307 to Arduino’s pin A2.
The Arduino should now start displaying the power readings.
Connect the output of the Si570 with a very short coax to the AD8307.
Note the reading. The Si570’s output is around -3 dbm. The display will show different reading. Just note the difference and mentally add that up to the readings until we fix this specan.ino source code.
The AD8307’s power reading are un-caliberated at the moment, but our initial alignment only needs relative readings.
4. Filters – Building the 70 Mhz low pass filter
The Specan has a total of four filters that need to be carefully built and shielded. The 70 MHz low pass filter is the simplest of them all.
Each of the coils is made by winding 8 turns on 1/4-20 bolt as a former. Remove the former after winding the coild. They are best made with silver wire, we used 22swg copper wire in the prototype.
A strip of five pads is cut in a double side PCB. The coils are soldered across the pads. The capacitors are soldered between the pads and the ground plane. Keep all the caps on the same side of the strip. Drill a number of tiny holes on the PCB and solder small pieces of wire to extend the ground plane to the bottom side as of the copper clad board as well.
The entire filter is shielded and a each coil is shielded from the other as well.
Small flaps are left out on the two ends of the shield to bring the input and output to the other parts of the Specan.
The LPF filter can be tested easily with the Sweeperino
Connect the sweeperino to the LPF’s ports
Download the specan.exe from https://github.com/afarhan/specan-ui
Now sweep the LPF with the specan.exe (See the videos for details on how to use it)
Your LPF is now tested.
Note : the coils are shielded from each other. the final shield around the BNC connector is not shown here
5. Filters – Building and Aligning the VHF bandpass filter
The original VHF bandpass filter proved a challenge. The coupling called for gimmick capacitors that were difficult to setup properly. Instead, a simple to make VHF bandpass filter was designed using only fixed value capacitors. Here is the circuit:
The circuit is assembled on a double side, unetched board with seven pads of 0.2 inches each made in a row with a box cuttter. The 2.2pfs are soldered between every pad except the end pads that have 4.7pfs. The 10 pf capacitors are soldered between every pad and the ground. Keep all the 10 pfs facing one side. Solder the coils’ ground ends on the same side as well. All capacitors are SMD.
Shielding the VHF filter is very important. Solder the a 1 inch walls of copper sheeting or copper clad between the coils. Take care to cut small apertures in the bottom of the shield to prevent the copper pads from being shorted to the shield.
There were no 4.7pf capacitors in the junk box, we used two 2.2pfs in parallel.
Aligning the filter is not difficult at all.
Connect the Sweeperino’s input and output together with a cable
Enter “f112000000” (without quotes) on the Serial Monitor of Arduino’s PC software to set the Si570 frequency to 112Mhz. Note thpower reading on the LCD as the reference reading.
Connect the filter to the Sweeperino.
Desolder the ground end of L6.
With a long nose plier, grab the ground end of L6 and drag it over the ground plane, expanding/contracting the turns for the best reading on the LCD display. Solder the ground end of L6 at the best spot.
Repeat this with L7 and L8.
Now solder the remaining side of the copper shielding around the filter.
Run a full sweep of the filter from 70 MHz to 130 MHz with specan.exe
Your total filter loss should be less than 6 db compared to the direct reading of the Si570 at 112 Mhz.
This completes the alignment of the VHF bandpass filter.
6. Filters – Building and aligning the two HF filters
The wide HF filter uses 5 resonators. The inductors are wound on T50-6 toroids. They can also be air wound if the one can expend the extra effort of shielding them individually. We used a copper clad board about 4 inches long and 2 inches wide to accommodate both the HF filters and the switching relays. The entire module of HF filters is shielded with copper sheeting. There are two outlets at the two ends for input and output. The power and filter selection lines are well bypassed while bringing them inside the shield to prevent stray RF pickup.
This wide band filter is assembled on a single strip of pads cut into a copper clad board similar to the VHF band pass filter. Aligning them is similar. Set the Si570’s output to 12 Mhz with ‘f12000000’ command from the Serial Monitor and peak the output by moving the turns on each of the them for the best response.
The narrow band filter needs no alignment. Just use 12 Mhz crystals that were matched to within 100 Hz of each other. All the nine crystals need to be matched to each other. Solder the bodies of the crystal to the copper board.
7. The 100 Mhz oscillator
Building overtone oscillators has been a challenge in the past. This new design is based on the work of John Stephenson, KD6OZH in the QST of November, 1999. It is a very low noise oscillator – exactly what is needed to complement the low noise of Si570 in the first oscillator. This version oscillates very easily and without needing a coil in parallel with the crystal.
Tuning up this overtone oscillator is an old trick from EMRFD. Replace the crystal with a 22 ohms resistor in series with a 0.01 uf capacitor. Switch on an FM radio nearby and tune it to exactly 100 Mhz (use a radio with digital display). Compress and expand L10 until you can hear the signal in the FM radio (the signal will appear to make the radio dead silent). Tune the radio around and confirm that the peak of the signal (maximum quiteness) is at 100 Mhz. Now, replace the resistor and capacitor with the crystal. You should be able to hear the signal again. You might have to (very lightly) touch up the L10 for proper oscillations.
8. Building the IF strip
The IF strip has more than 35db gain. As the AD8307 has a response from 0 to 500 Mhz, the output of the IF amplifier has to be tuned to 12 Mhz. Set the Si570 to 12 Mhz and inject the signal into the IF amplifier. Peak the L17 for maximum response.
9. All together
Once all the filters have been aligned, bringing the Specan together is a fairly straightforward and simple job of hooking it all up together. The rest of the modules : the mixers and the amplifiers are all broad-band and simple to make. They don’t need as much shielding or alignment.
Check the full circuit diagram and the pictures for the parts layout.
10. Using the Specan
The video shows how it can be used as a stand-alone measurement receiver or as a spectrum analyzer. Briefly this is how the controls work
11. Manual operation
The tuning knob tunes from 0 to 100 MHz in 100 KHz steps.
A click on the function button switches between fine tuning and coarse tuning. The fine tuning tunes in 100 Hz steps.
Press and hold the function button to switch between narrow and wide resolution bandwidths.
Using the PC
Connect the Arduino to the PC and fire up the specan.exe (Download it from https://github.com/afarhan/specan-ui.exe)
From Action > Connect set the specan.exe to use the correct serial port of the Specan’s Arduino.
Right click on the display to initiate a scan. The initial scan is set to 30 MHz sweep centered at 30 MHz (a total scan from 0 to 60 MHz)
Moving the cursor over the display measures the exact power level at the frequency under the cursor
You can left-click on any part of the display and choose a different sweep range to view that part of the scan better
There is a feature to place a ‘mark’ on a particular store a particular reading (like a maxima or a minima). The specan.exe displays the difference in frequency and power of the reading under the cursor and the previously marked reading
To capture a screenshot, press PrtScr key, open up your favourite paint program and paste the captured screen shot!
11. Final alignment
There are two independent calibrations to be made for the Specan.
A. The power caliberation
The two HF filters have different losses. Each of them has to be calibrated separately. The values of #define NB_POWER_CALIBERATION and #define WB_POWER_CALIBERATION need to be tweaked to get the power readings aligned properly. You will need a “calibrated signal source”. I just used a crystal oscillator for which I measured the output into a 50 ohm resistor with an oscilloscope.
B. Frequency caliberation
Connect the spectrum analyzer to an oscillator with accurately measured frequency. Initiate a scan with the narrow band filter selected. If the signal shows up at a different frequency on the Specan, adjust the #define IF_FREQ of specan.ino by the difference in between the Specan reading and the actual frequency of the oscillator. Save specan.ino and upload back to the Arduino.
I have provided this rather extensive documentation of this spectrum analyzer in the hope that many will try building it. In spite of its seeming complexity, it is in fact a quite straight forward circuit. For those of us who build our own rigs (and especially those who build multiple radios), a spectrum analyzer is a worthy instrument to build.
As useful as it is to have one in the lab, the building the specan has been a great education in construction of high performance general coverage rigs (the specan is a pretty decent receiver if you substitute an audio chaing for the log amp). So, if you have built a superhet in the past, this should be your next project. And don’t worry about the software part! it has all been worked out.
About the software. It is a little kludgy but it does what I wanted it to do. I had never used a spectrum analyzer before so I don’t know what a ‘real’ one does. I made it to my requirements. Feel free to download and modify. I can immediately think of having two separate knobs for coarse and fine tuning, a switch for selecting the narrow and wide filter. On the PC side, the connect dialog should only list the serial ports that are ‘live’. I hope someone hacks that for us. I’d prefer the PC software to be not written in C at all. It is possible to write it entirely as an app for a smart phone. That would clear up the PC from my bench!
No PCBs are available for this build (yet). If any kit supplier would like to offer them, please drop me a line. Join the EMRFD group to discuss the design.
This project wouldn’t have happened if the original hadn’t been designed by Wes and Terry. Over the last two years, Wes and others have mentored with this project. I am deeply indebted to William Meara for two reasons : For proof-reading this write-up (which is easy for him to do) and to keep it under wraps until I had it all sorted out (which is difficult for him to do given his gossiping ways at soldersmoke podcast!)