May 1, 2012
Join the "CWTD Yahoo Group" for email discussion in between our weekly sessions by clicking here.
Harmonics and Spurs and Parasitics (“Oh My!”)
… What you see is less than what you get
So how many of us have built an oscillator, a VFO, a DDS signal generator or even a full QRP transmitter, then put it on the air and tried to make a contact with it straight away? If it works, great, you keep using it! But if you don’t have some fancy ($$$$) test equipment, how do you know the quality of the signal you are putting on the air, or injecting into other circuits? An oscilloscope only goes so far, and having your buddy across town listen to your signal on the air is a little subjective.
Well … to get a better handle on your signal, it’s first a good idea to understand a little of the “mechanism” of what’s in a simple sinusoidal signal, and what affects it’s purity. And then, using some simple techniques implemented with standard analog components, look at all the signals your oscillator is generating to see if your transmitter is operating well enough.
That’s the goal of this week’s session in a nutshell!
73, George N2APB & Joe N2CX
Audio Recording ... (Listen to MP3 recording)
Text Log ...
<20:26:21>"Russ G3OTH": why use
a square wave for the mixing oscillator?
<20:28:38>"George - N2APB": Because the (necessary) Low Pass Filtering that happens later in the chain takes out the long series of odd harmonics.
<20:31:03>"George - N2APB": This is the same principle that says you could transmit using a cruddy squarewave-like signal as a direct-oscillator and then put the signal though a heafty LPF to knock out most of the odd harmonic energy.
<20:31:18>"Russ G3OTH": Sorry, meant to ask why is it recommended to use a square wave for the local oscillator rather than a sinusoidal wave, surely this would exagerate harmonic mixing product spurs, please explain.
<20:31:41>"George - N2APB": One would say sometimes: "Well, why can't we just use a VARIABLE square wave generator as a VFO with an LPF?
<20:32:28>"George - N2APB": Well, you would need a VARIABLE LPF to change frequency in tandem with the changing fundamental! This is because the harmonics change in frequency along with the changing fundamental.
<20:39:35>"Paul - wa0rse": Yes!
<20:49:16>"Mike WA8BXN": Its important to tune the notch low freq to high and not the other way around!
<20:51:51>"George - N2APB": Good point Mike. One reason for this would be that the fundamental frequency would be more evident when adjusting the notch from low-to-high, because there will likely be little-to-no unwanted "gudge" sprectra BELOW the fundamental. Thus, it will be very evident when the notch reaches the fundamental.
<21:00:03>"Russ G3OTH": why did you choose the parallel rather than the series resonant frq of xtal and put in series as per the LC ccts you used?
<21:04:25>"Rick K3IND": Want to comment on the NS SwitcherCAD software you mention in the online charts?
<21:10:05>"Joe N2CX": linear technonlogy lt spice link ltspice
<21:10:30>"Joe N2CX": LTspice : LTspice/SwitcherCAD III
SESSION NOTES ... Harmonics, Spurs and Parasitics
All transmitters produce signals other than the desired output.
Most common are harmonics or signals at multiples of desired (fundamental) frequency .
Spurs and parasitics are signals not related directly to the fundamental.
Can be due to unwanted oscillations or something caused by the basic signal generation process.
With SSB we can also get intermodulation distortion caused by non-linear amplification.
... And what it looks like in real life …
Real Life Spectrum
Transmitters must be designed and tested to meet FCC rules
FCC Part 97.703 Rules
HF - “ the mean power of any spurious emission from a station transmitter or external RF amplifier transmitting on a frequency below 30 MHz must be at least 43 dB below the mean power of the fundamental emission”
Above HF emissions must be attenuated by at least 60 dB.
Testing for these imperfections is often a challenge.
- Age old way is having a ham in the same town listen for off-frequency signals or poor keying or audio.
- Yet another way is to wait for a pink slip from the FCC.
Best way is to do careful measurements to check how much they are attenuated.
- ARRL Lab Test Procedures Manual has detailed methods for testing ham radio equipment
- They perform these tests on new ham gear to publish product reviews.
Good Test Equipment
Doing things by these methods involves a lab well equipped with good properly calibrated test equipment.
RF power meters
Calibrated dummy loads
RF summing devices
So … “How can you make quality measurements on a transmitter’s signal without all that test equipment.”
Rather than look at all the signals at once and compare their strength, first measure the power from all of them and subtract out the fundamental signal. The remainder is then how much unwanted power is left!
Knowing this one can then try filtering, reducing transmitted power or other means to decrease the unwanted signals relative to the wanted transmitter output.
Here’s a simple circuit that we designed as a group project at one of our Atlanticon QRP conferences several years ago. The idea was to implement the simple approach mentioned above: “Measure all the energy and then take out the fundamental, and as a result leaving a measure of all the unwanted energy being generated.”
And all this is done with straightforward analog components … “we don’t need no stinkin’ computer!”
So the idea is to use this circuit to measure the spectrum energy above the fundamental signal being generated, and then make whatever adjustments possible to reduce this somewhat relative number … reduce the drive, use better shielding, better grounding, better filtering, et al.
The Signal Quality Meter, or ‘SQM’, is a user-friendly device that produces a graphical display of relative quality of an RF signal applied to its input. Using a modern analog integrated circuit and other familiar RF components, the SQM examines an input RF signal and identifies how much unwanted harmonic or other off-frequency components are attenuated. It provides a simple and easy way to determine how “clean” the output of an oscillator, amplifier or properly attenuated transmitter really is.
An input signal in the HF range, as high as 4 Vpp down to 10 mVpp, is fed to the input. The signal is attenuated and passed through a notch filter to attenuate the fundamental signal. All other components (harmonics of the fundamental and undesirable higher frequency spurious signals) are allowed to pass on for measurements. A logarithmic detector then senses this remaining energy in the spectrum and feeds a bargraph display to show the cumulative amount of remaining off-frequency signals.
Input RF signals are fed to terminating resistor R1 and a level-adjust potentiometer R3. The termination resistor is set to a value that will present no more than a 1.5:1 input SWR regardless of the control settings. The level-adjusted signal then feeds a notch filter with several user-selectable configuration options.
The simplest configuration for the SQM uses a high-Q crystal that resonant at the desired signal frequency. When a user-provided crystal is plugged into socket X1, the resultant notch filter will block most of the energy at the fundamental frequency and pass harmonics and spurious signals with little attenuation. The high Q of the crystal resonator also provides a narrow notch that is usable for measurement of close-in spurious signal attenuation. This configuration is useful for single frequency measurements where the source is stable and fixed at a specific frequency.
However, instead of using the SQM at a fixed frequency with a crystal, the SQM may alternately be configured with a tunable notch filter by using variable capacitor C8 and inductors L1 and L2. L1 is used for 3–10 MHz operation and L2 allows tuning up to 30 MHz. C1 and the inductor chosen form a series-resonant circuit which has an effective resistance of less than 2 ohms at its resonant frequency. With 200 ohm R2 in series with the input signal, it forms a voltage divider that shunts the fundamental signal to ground at the fundamental frequency. This configuration provides strong attenuation of the fundamental and little attenuation of off-frequency harmonics and spurious signals, and is useful in cases when a VFO output signal is being measured.
FET Q1 serves as a high impedance buffer to prevent circuit loading by the detector circuit IC2. The detector is an Analog Devices logarithmic detector that produces a DC output corresponding to the log of the RF signal applied at its input. For example an input of -40 dBm (~6 mVp-p) produces an output of about 1.6 Volts DC.
Increasing the input to -__ dBm (__V) raises the output to about 2.08V. This is a slope of about 24 mV per dB of input.
The 5V regulator IC1 provides clean 5V DC supply voltage to the log detector chip.
This DC output of the log detector is filtered by a simple RC circuit then fed to IC2, an LM3914 LED display driver IC. The LM3914 uses an internal series of voltage dividers, voltage comparators and an precision voltage reference to switch on its bargraph style display in response to the input DC signal. With the chosen values of R7 and R8 this means that the lowest output L1 turns on with a 120 mV input, L2 also then turns on turn at 240 mV and so on until all outputs are turned on with an input of 2.50 V. U4 then drives a series of LEDs to provide a visual display. The combination of the log detector chip and the LED driver chip produce a bargraph display over a 50 dB range in ten 5 dB steps. The bargraph display can be either a 10-segment display component (DZ1) or 10 discrete LEDs.
Here's some SPICE analysis stuff to show SQM notch depth using the free National Semi SwitcherCAD software. (Great tool!)
Series Resonant Crystal Configuration
The LC configuration below assumes an inductor with a Q of 100 which is conservative. It actually looks pretty good …
Series Resonant L-C Configuration
Input: RF Signals between 3 and 30 MHz
Input Impedance: 50 ohms nominal, SWR less than 1.5:1
Input RF range: __V (xxdBm) to __V (yy dBm)
Note: damage can occur with D inputs or RF greater than zzV.
Transmitter spurious measurements require attenuation to prevent smoke, fire and consternation.
Notch Tuning range: 3-30 MHz in two ranges
Notch depth: Adequate to display 30-40 dB off-frequency spurious signals
Display: 10 segment display with 5 dB nominal steps
Power: 9V at less than (TBD) mA. Alkaline battery recommended.
USING THE SQM
Jumper Settings on JP1 pinheader:
Pos 1 used for tunable notch mode)
Pos 2 used for 3-10 MHz operation
Pos 3 used for 10-30 MHz operation
1) Place shunts on position 1 of the JP1 pinheader. The other two positions of JP1 should remain open (no shunt applied).
2) Initially adjust potentiometer R3 to its minimum setting (fully CCW, as viewed from the front/plastic side of the pot.)
3) Apply the RF signal to be measured.
4) Press and hold power switch S1. Carefully rotate pot R3 clockwise until all 10 segments of the display are lighted. Rotate the control first so that only 9 are lighted then slowly adjust until the 10th segment just turns on. This is the “full scale” reference setting. Release S1.
5) Place another shunt on JP1 at position 2 (for 3-10 MHz operation) at position 3 (for 10-30 MHz operation). (NOTE: Keep the shunt on position 1.)
6) Again depress S1 and tune C8 to reduce the number of display segments that are lighted. Spurious signal attenuation can now be observed. It is indicated by noting the number of display segments that are now dark. Each segment corresponds to approximately 5 dB attenuation. For example, two dark shows 10 dB reduction, 3 dark is 15 dB, etc.
A Spectrum Analyzer for the Radio Amateur, Part 1, Aug 1998 - QST (Pg. 35)
ARRL Members only web site: <http://p1k.arrl.org/pubs_archive/94603>
A Spectrum Analyzer for the Radio Amateur, Part 2, Sep 1998 - QST (Pg. 37)
ARRL Members only web site: <http://p1k.arrl.org/pubs_archive/94665>
Additions and corrections: Go to <http://w7zoi.net/tech.html> and search for “Spectrum Analyzer Information”
Back to Teamspeak Home