Vorläufig / Preliminary, work in progress
RF Calibrator Sine/Square Wave
This web page will describe the build and use of two 10 MHz Oscillators for -10 dBm Level Calibration purpose, for example an AD8307 power meter.
The circuit design was copied from QRPHomebuilder.pdf, Title: CMOS RF Calibrator page 617 from 945.
If you build a Radio Frequency (RF) network analyzer you have the difficulty to calibrate the RF level of the RF detector, usually an Analog Devices AD8307, or AD8302. For that you need a Calibrator. But where to begin with home made tools?
Fortunately there is a clever design from Bob Kopski K3NHI to allow so. It is basically a cheap commercial 10 MHz square wave CMOS clock oscillator. Then you just need a DC multimeter for calibration. It is no expensive measurement equipment necessary. His design was presented in the ham radio magazine QEX from Jan/Feb 2004. See Link #9, but this was the old version with -20 dBm amplitude.
The adjustment to 50 mV must be made with a 50 Ohm resistor, no DC blocking.
While at the square wave oscillator the AD8307 shows -10 dBm, a Spectrum Analyzer shows -13 dBm.
In the QEX May/June 2010 he wrote Tech notes to update the RF level from -20 dBm to -10 dBm, because of a problem with newer AD8307. Unfortunately this paper is not available in the web.
It is important to use the CMOS version (not TTL) of this clock oscillator, because you need the full voltage swing from rail to rail. Search on Ebay or electronic parts dealer for the words "oscillator 10Mhz cmos crystal" for ordering such a part.
If you just have a TTL crystal oscillator a hand, and want to make a test, a pull up resistor at the output of the oscillator should help to make it work.
If the CMOS oscillator is supplied with a regulated power supply, and the load is fixed, the output signal amplitude stays stable.
Another benefit of a square wave crystal oscillator are the stable frequency and the harmonics for checking a spectrum analyzer, or a Software Defined Radio (SDR).
On the right side you will see a picture of the Square wave Calibrator, with a self made 3D print housing.
With that calibrated square wave oscillator you can then calibrate an sine wave oscillator to the same amplitude level = -10 dBm.
The output impedance is very close to 50 Ohm.
When you click on a picture or diagram it will be shown in the original size for better reading.
Square Wave Calibrator
Please have a look at the schematic on the right side. In order to be independent from a power supply, a 9 V battery together with a 3 Terminal voltage regulator (78L05) will last for several hours of operation.
The principle of calibration is, that the integrator R5 + C3 (10K, 0.1uF) integrates the square wave. The resulting voltage is 1/2 Vpp (peak-peak) DC, if the duty cycle is 50%. In the data sheet of my CMOS crystal oscillator I have seen a specification in the range of 40 - 60%. Fortunately I could measure with my digital oscilloscope (Hantek DSO5072O) 49%.
Update 2017-10-23: While testing my new ActiveRFProbeBF998 I figured out, that the large ringing on the square wave signal can be lowered with inserting a ferrite bead, see in the schematic L1.
In order to bring theory to reality I did a LTSpice simulation.
The LTSpice upper circuit is the case with a pure 50.0 Ohm termination with DC coupling. I adjusted trimmer R4 (schematic) that I can measure at capacitor C3 (schematic) a DC voltage of 50 mV. That gives for R3 + R4 (schematic) a measured resistance of 1208 Ohm. For the CMOS oscillator output resistance I found a value of 100 Ohm, by taking the real peak to peak voltage of 4.72 V into account. For the output resistance I can calculate then: (1208 + 100) || 68 || 220 Ohm = 49.97 Ohm. You can see, that the simulated voltage of node n004 comes very close to 50 mV.
The LTSpice lower circuit is the case with a 50 Ohm (real 47.88 Ohm) termination with AC coupling (C2 = 100 nF). Because of the AC coupling the square wave amplitude sits on the average value of 50 mV. Therefore the simulated voltage of node n008 comes very close to 100 mV. The input resistance 47.88 Ohm of the AD8307 module come from 100 || 100 || 1100 Ohm (chip input resistance).
In you want to simulate yourself, please download the attached file Calibrator_LTSpice_squareW_AC-coupling.asc.
The 10 MHz signal measured at a 50.0 Ohm termination resistor (without capacitor in line) can be seen in the screen shot to the right.
In order to get a smooth curve, the signal was averaged by the DSO 64 times.
This gives a reading of -10 dBm at an AC coupled AD8307 logarithmic amplifier.
The AC RMS value (for “heat value” or Spectrum Analyzer) of the 100 mV square wave is simply 100/2 = 50 mV. The power on 50 Ω is (50 mV × 50 mV) / 50 = 50 mW.
This is –13 dBm.
The ringing does come from the limited bandwidth of the DSO.
If you use the square wave synthesis the addition of 6 sine waves matches about the oscilloscope screen shot.
Sine Wave Calibrator
Please have a look at the schematic on the right side. The design is from Todd, VE7BPO.
In order to have a stable power supply voltage I like to use a fixed power supply, a unit like ebay.com ID 282489131402 12 V DC 1 A, with a round 5.5/2.5 mm plug. But what concerns me, is the missing shorting protection.
The plug I use is: 10* New DC-022 DC Power Outlet Diameter 5.5mm Inner pin 2.1mm ebay.com ID: 181357006386 price: 10 pc. $1.68
In the original schematic a FET type J310 is used.
Take care to use a ferrite bead over the FET drain leg, in order to avoid unwanted oscillations.
The capacitor C3 (10uF) is needed to bring the switching power supply ripple down from 200 mVpp to 20 mVpp.
The tuning is as follows:
- Terminate the output with a 50 Ohm resistor terminator.
- Connect a milli ampere meter between the +12 V input and the regulated power supply.
Adjust the FET source trimmer R6 so that the circuit draws around 2.7 mA — then connect the PSU directly.
Adjust the trimmer capacitor for the highest pk-pk voltage (and/or or best looking waveform) in a oscilloscope, or highest power in a AD8307 Power Meter. Please have a look at the screen shot on the right.
- Connect the sine wave oscillator to the network analyzer Yana (from Tom, K1TRB), or an AD8307 Power Meter. Then calibrate with trimmer R6 to a level of -10 dBm (about 200 mVpp).
Harmonic frequencies (spectrum)
Please have a look at the screen shot on the right. You can see, that the second harmonic (20 MHz) is about -39 dBc below the first harmonic (10 MHz).
In the original version a value of -39 dBc is achieved.
dBc, decibel relative to carrier, a measurement in RF engineering.
Amplitude Reference = 1 V RMS = 0 dB = 13 dBm @ 50 Ohm
Because of the few and mostly small parts, I used perforated board (stripes, Raster 0.1").
With an 3D-Printer I printed a little housing to just fit the perf. board and the plugs. The size is suited to the highest electronic part, the resistor trimmer (on ebay.com: Piher PT10LH01 Series Carbon Trimmer Pot, Linear 10mm Horizontal Adj, on ebay.de: Trimmer Trimmpoti PT6H 500 Ohm stehend PIHER).
Please see the photos below for the setup. The cut-outs in the perf. board are only needed, if you use the 3D printed housing.
Also below the perf. board photos for the Square-wave calibrator.
Below is an example screen shot of the housing 3D design with program FreeCAD (version 16). The size is 27 x 47 x 22 mm (width x length x height). The CAD files can be downloaded (*.FCStd = FreeCAD file, *.stl = Export file for slicing, e.g. program Cura). With the FreeCAD file you can make your own modifications at the housing. The .stl export file ist just for 3D printing.
On the right picture is the Sine wave calibrator with the attached switching PSU. Once calibrated, the PSU should stay permanently with the calibrator, because every PSU does have a different voltage.
RF level measuring, probe
If you are measuring Radio Frequency Level with an oscilloscope you usually have a measuring probe, see the picture on the right. There are several things to consider:
A 1:1 probe: Drawback: high input capacitance of about 70 to 120 pF.
A 1:10 probe: Drawback: needs waveform calibration to the oscilloscope capacitance. Input capacitance about 13 - 17 pF.
Ground lead length: How shorter how better, to avoid ringing on the signal. See the steel spring in the picture on the right, about 2 cm long.
If you need a lower input capacitance (about 1 pF) you can use an active probe.
RF level measuring, 50 Ohm
I found out by experiment, that a RF (Radio Frequency) measurement is not so easy, even at 50 Ohm impedance.
Because my Hantek DSO5072P does not have an internal 50 Ohm termination, I need to do it external, with an BNC T-connector. The -3 dB bandwidth is about 150 MHz.
Please see the photos on the below.
The ringing was much more (overshoot about 40%), than from the Gibbs phenomenon predicted (overshoot about 9%).
Next come my old HP1652B logic analyzer with digital oscilloscope and switch selectable 50 Ohm input termination, see the screen shot on the right. The -3 dB bandwidth is 100 MHz.
That looks more realistic on the top level, but not at the bottom level.
The overshoot was automatic measured with about 13%.
Next, the 10 MHz square wave signal was measured again with DSO5072P, but this time with a direct connected (SMA) 50.0 Ohm resistor termination and the original Hantek probe PP-90 10:1 (bandwidth: 80 MHz). The maximum overshoot can be estimated with about 8%. You see, that those measurements are very difficult.
In the picture below right you see my homemade probe connection. On the SMA connector (with two 100 Ohm SMD resistors for 50 Ohm termination) at the center pin I soldered a little plug from a 1/10" wire connector that just fits the probe tip. For the ground connection I took a tinned copper wire (0.5 mm diameter = AWG24), soldered one end to the SMA connector ground, the other end I coiled around the probe ground shell (no soldering). Click on the picture to see it in full size.
In order to calculate values like overshoot an rise time in theory, I calculated those values with a LibreOffice Spreadsheet (can be downloaded).
The left diagram shows the square wave synthesis with 5 sine waves (10, 30, 50, 70, 90 MHz), the right diagram shows the square wave synthesis with 7 sine waves (10, 30, 50, 70, 90, 110, 130 MHz). The even harmonic are missing, if the duty cycle is about 50%.
You can see, that with increased bandwidth (oscillator and oscilloscope) the edge rise time (10% to 90% of the amplitude) will be shorter, from 4.4 ns to 3.3 ns.
SquareWave Oscillator, harmonics
The author of the square wave oscillator proposes to use the harmonics for the calibration of a spectrum analyzer.
I see also a use for testing a SDR receiver (Software Defined Radio). See the FFT spectrum screen shot (Fast Fourier Transformation) on the right side.
Keep in mind, because of the nearly 50% duty cycle of the square wave signal, that the even harmonics are very low in amplitude. In my case about 31 dBc below the carrier (10 MHz).
The formula for calculating the harmonic voltage is:
4 * Vpp Vpp harmonic = -------- n * Pi
where n = the wanted harmonic number ( 1 = carrier = 10 MHz).
For the conversion to dBm (50 Ohm) use: dBm = 10 * 20 * LOG(Vpp/(2 * 1000)) (2 = Vpp to Vp, 1000 = mV to V).
The frequencies from 10 to 110 MHz look like:
harmonic scope reading calculated 1 10 MHz -27.5 dB dV difference difference to theory 3 30 MHz -37.5 dB, 10 dBc, -9.5 dB, -0.5 dB 5 50 MHz -43.5 dB, 16 dBc, -14.0 dB, -2.0 dB 7 70 MHz -49.5 dB, 22 dBc, -16.9 dB, -5.1 dB 9 90 MHz -52.7 dB, 25.2 dBc, -19.1 dB, -6.1 dB 11 110 MHz -56.3 dB, 28.8 dBc, -20.8 dB, -8.0 dB 2 20 MHz -58.7 dB, 31.3 dBc, other even harmonics are similar
The difference to the theory is probably caused by limited bandwidth, or the FFT calculation of the oscilloscope is not correct.
The table was calculated with an OpenOffice spreadsheet (can be downloaded).
Because it was not so easy to build the Sine Wave Calibrator, there was an idea from Tom K1TRB to use a Square Wave Oscillator and low pass filter out the harmonics, in order to get a pure sine wave at the output. See the schematic on the right. The schematic can also be downloaded as pdf file and Eagle 7.5 CAD file.
The low pass filter (type Butterworth) was designed with the Windows (R) software Elsie (see at Link #5) and optimized for E12 series values for the inductors and capacitors, see below.
Because of problems with the DC termination, capacitor C4 (0.1 uF) was inserted.
Because of the high ringing amplitude of the square wave signal a Ferrite bead L1 was inserted at the CMOS oscillator output, see the DSO screenshots below. The ringing could be reduced by 1.6 Vpp.
measuring at the CMOS oscillator output
measuring after Ferrite bead L1, Hantek DSO5072 (BW = 150 MHz)
measuring Frequency Spectrum at RF-Out
measuring after Ferrite bead L1, HP1652B (BW = 100 MHz)
The screenshot above shows the spectrum of the sine wave at the output. The third harmonic is about 44 dB below carrier
The HP1652B shows less ringing, compared to the DSO5072, but nearly the same peak-to-peak voltage (7.04 V / 6.94 V)
The level adjustment can be done indirect only.
Next connect the Square-Sine Calibrator to Yana, and adjust trimmer R4 for a -10 dBm reading at the Yana PWR display. The reading at the DC-Cal port was in my case about 583 mV. With a good scope you should measure a sine wave amplitude at RF-out of about 200 mVpp (= -10 dBm @ 50 Ohm).
Contact Email, please enter your Email address
The entered Email address will not be published, or given away.
List of pages in this category:
-- RudolfReuter 2017-09-01 21:12:20