Frequency Multiplier,
Using a Step Recovery Diode
or Snap Diode
This Page was Started Dec. 18, 2004
Updated 1-25-05, add link for CAD layout of SRD board
Updated 2-25-11, Update this page.

    This page will act as a design guide for building a frequency multiplier using a diode that has a long Carrier Lifetime and short Transition Time. Some diodes that exhibit these qualities are Step Recovery Diodes, Varactor Diodes, and PIN Diodes. This page uses the terms SRD, Snap, and PIN Diodes randomly, but they all mean the same thing. I did not design this multiplier for maximum efficiency. Had I done so, the input and output would be specific (64 MHz X16 = 1024 MHz). The circuits and components would be extremely critical, and the layout topology would probably be impossible to duplicate. So, if you say "that doesn't look like any multiplier I have ever seen", you are right. However, this scheme will work with different frequencies and different multiplication factors other than just 64 MHz and X16.

     My original intention was to build a frequency multiplier for the Modularized Spectrum Analyzer. The MSA's 2nd Local Oscillator, which runs at 1024 MHz, is a PLL/VCO/Amplifier combination. There is absolutely nothing wrong with that scheme, but it can be replaced with this SRD multiplier scheme to improve Spectrum Analyzer phase noise results. The total SA phase noise improvement is only about 1 dB. Therefore, using a multiplier scheme for LO 2 has no benefit unless the SA's LO 1 is modified for low phase noise operation.

    At the end of this page is some technical information for Step Recovery Diodes.

Go back to the MSA Main Page

Frequency Multiplier Block Diagram to replace PLL2 and VCO 2:
SRD Multiplier Block Diagram
    This frequency multiplier scheme consists of a driver, impulse generator (Snap Diode), filter, and an amplifier. The filter is a coaxial cavity filter tuned to 1024 MHz and an insertion loss of 1.5 dB. The amplifier has approximately 20 dB of gain and is saturated with an output of about +13 dBm. In the following diagram, the driver and impulse generator are combined onto one pwb.

The Driver, the Step Recovery Diode Module
SRD Module
 
   This module can respond to a sine input, but it is expected to receive the square CMOS signal from a 64 MHz Master Clock. The first section of the 74AC04 (or ACT04) will convert the input.  The second stage will square up the signal. The third stage is a parallel driver, to assure maximum drive to the Snap Diode.  The Snap Diode acts as the impulse generator. Notice that the Snap Diode is not a Step Recovery Diode, but rather, a PIN diode (MPN 3401).  Although not as efficient as an SRD, this particular PIN diode performs well in this design. It is also extremely cheap and available (about 50 cents, compared to about $20 for an SRD).  The completed multiplier measures only 3 dB less efficient than when using the HP 5082-0113, an SRD.
    R3 is the self bias resistor for the PIN. Although its value is shown to be 50 ohms, it will be changed to optimize the circuit efficiency or accomodate different SRD's.  Notice, there is no load or resonator on the output of D1. A resistive or inductive load must be shunted at J3 for the diode to function. A 50 ohm load could be placed at J3 to create a "Comb Generator". But for this application, the hairpin input of the cavity filter will act as an inductive load for the SRD.

    The actual measured output of the SRD module (as a comb generator) is approximately -23 dBm at 1024 MHz.  When the resistive load is replaced with the band pass filter, the filter's output power is -5 dBm at 1024 MHz. I realize that this sounds like voodoo reverse-conservation of energy, but the band pass filter acts as a reflector to all harmonics except the one at 1024.  The reflected harmonics are directed back to the SRD and their combined power will increase the efficiency of the SRD action.
    The Snap Diode can be mounted in either direction. I found that maximum power output was obtained with the diode's cathode on the output. This is not due to the effeciency of the diode. It attributed to the '04 driver, which has more current capability on its negative edge than its positive edge.
    Changing the bias resistor will change the length of time that reverse current is supplied to the diode. This does not change carrier lifetime of the SRD, but it does change the total circuit lifetime. This changes the circuit efficiency. Using various resistor values, the filter's output power can be maximized. The actual resistor value depends on many factors, but it should be between 200 ohms and 2000 ohms. After much "resistor tweeking", my final bias resistor value was 1.1 K ohms.
    Power could be further increased by using a specific length delay line between the SRD output and the input to the cavity filter. A critical phase delay would allow the reflections from the filter to affect the SRD efficiency.  This delay line technique could possibly result in a power output of +6 dBm.  However, this is below the requirements for the mixers in the MSA, so a post amplifier is needed.
    Without the delay line an output from the filter might be as high as +2 dBm, but a more conservative expectation is somewhere between -6 dBm and 0 dBm. Since an amplifier is required for either topology (with or without delay line), I chose not to include the delay line in the design. Its length is extremely critical, anyway. An amplifier can increase the total power to drive any mixer.

    Here are screen shots of the SRD Module Board Layout.  I used ExpressPCB software, which can be accessed by clicking, SRD Module PWB.  The size of the board is 1.4 x 1.35 inches (this does not conform to present SLIM requirements).  Layer 1 is the silkscreen, layer 2 is the top copper, layer 3 is the bottom copper (as seen from top) :
SRD PIC LAYERS ALL SRD LAYERS 2 AND 3 SRD LAYER 2
Click for :  all Layers,              Layers 2 and 3,                          Layer 2,                    
    The board is layed out so that either the DIP or SOIC package 74AC04 can be used. I used the SOIC for testing.

The Band Pass Filter
    The following dual coaxial cavity filter can be used as the band pass filter, to pass the 16 th harmonic of the SRD's 64 MHz input.  I won't repeat the build process on this page.  For pictures and construction techniques, go to the Cavity Filter page. This dual filter can be replaced with other narrow band filters with good results. I have done extensive multiplier testing using a single coaxial cavity filter instead of this two stage coaxial cavity filter. The 15th and 17th harmonics (960 MHz and 1088 MHz) are attenuated about -30 dB below 1024 MHz, which is adequite for use in the MSA.


1024 MHz Dual Cavity Filter, dimensions are inches.

The Amplifier
    Any amplifier with enough gain and output power at 1024 MHz could be used. Here is a proven amplifier design used in the MSA. It is a design for a dual RF Buffer Amplifier. It can be accessed by clicking SLIM-RFA-1.
SLIM-RFA-1 Schematic
 slim/skpwb_rfa.gif
    This amplifier design has no overall gain. To be used as an amplifier for the multiplier, perform the following modifications. Only one section is required for the multiplier, but the components for both sections are listed.
Change R22 and R32 to 0 ohms. Delete R20, R24, R30, and R34. The outputs will be +10 dBm. If a full +13 dBm output is wanted, change R27 and R37 to 0 ohms, and delete R26, R28, R36, and R38. Each stage has approximately 19 dB of gain and will saturate with an output of approximately +13 dBm, when the input is -7 dBm or greater. For a non-saturated output, the input resistive attenuators would be adjusted for proper output.
    To replace the PLO 2 in the MSA with Tracking Generator, both amplifier circuits are utilized, with each output at +10 dBm. Their inputs are driven by a power splitter. The splitter is fed from the output of the Multiplier's Band Pass Filter.

Technical information for Step Recovery Diodes
    The Step Recovery Diode is much like any other normal rectifier diode. It conducts current when biased in the forward direction, and exhibits a high resistance when biased in the reverse direction. What is not widely known is that all diodes can, and will conduct current when biased in the reverse direction. Step Recovery Diodes, PIN Diodes, and a few others are specifically manufactured to control these reverse current characteristics. The following documents are full of information. If you want to know more about these characteristics, please read:

Harmonic and Pulse Generation using the Step Recovery Diode
found at http://www.hpl.hp.com/hpjournal/pdfs/IssuePDFs/1964-12.pdf

Pulse and Waveform Generation with Step Recovery Diodes
found at http://www.hp.woodshot.com/hprfhelp/5_downld/lit/diodelit/an918.pdf

Harmonic Generation using Step Recovery Diodes
found at http://www.hpmemory.org/an/pdf/an_920.pdf

Low Cost Frequency Multipliers Using PIN Diodes
found at http://cp.literature.agilent.com/litweb/pdf/5966-4998E.pdf


Results of SRD Testing
In the following photos, the SRD module is used for a test bed, but using an SRD with with its anode connected to the output. A 50 ohm resistive load is on its output. The left photo is the output voltage waveform (with a 3.33 MHz square wave input signal to the SRD Module).
Pic of SRD output  Pic of SRD output 2
    Disregard the small positive going glitches.  These are line reflections with my test set-up.  The right photo shows the same output, but with a second trace showing the load voltage when the SRD diode is bypassed with a short.
    During the negative portion of the square wave (4-7 cm), the SRD acts somewhat like a normal diode. However, there is a noticeable negative overshoot (4-4.2 cm).  This is due to the charging of the high junction capacitance of the SRD. By 4.5 cm, the junction capacitance has been fully charged and the SRD is behaving like a normal diode.  In the right photo (at 5 cm), you can see that there is about .7 volts difference between the applied negative square wave and the SRD output.  This is the normal forward bias voltage drop of a diode. In the left photo (at 5 cm), the voltage is about -.3 volts. This indicates that the 50 ohm load resistor and series SRD are conducting 6 ma.

    When the square wave input transitions from negative to positive (at 7 cm), the SRD conducts, even though it is biased in the reversed direction.  The conduction will continue until the Carrier Lifetime is completed. This occurs at 7.2 cm (+1.5 volts). The peak series current of the 50 ohm load resistor and series SRD is = 1.5v/50 ohms = 30 ma. The lifetime for this particular circuit is about 10 nsec.  If this SRD had a longer lifetime, the positive transition would reach its full potential at +1.7 volts, allowing a peak series current of 34 ma.

    When the lifetime is completed (at 7.2 cm), the diode abruptly changes to a high resistance. This is shown as the voltage transition from +1.5 volts to about 0 volts. This abrupt change is not instantaneous. It does take a finite time for the transition. This is called Transition Time. It is the "Step" in Step Recovery Diode and "Snap" in Snap Action Diode. I cannot measure this particular SRD's Transition Time, but it is likely on the order of 100 psec or less. Transition Time is the most important characteristic of an SRD when designing a frequency multiplier. The "rule of thumb" is that the Transition Time should be less than 1/4 the period of the selected harmonic output. In the case of a 1024 MHz multiplier, this transition time should be less than .244 nsec (244 psec).

    If the module's 50 ohm load resistor is replaced with an inductive load, the previous pictures showing the SRD voltage output, can be thought of as current through the inductive load.  Basic electrical theory says that if you create current flow through an inductor, and then immediately remove the current, the magnetic lines of flux collapse and cause a voltage spike across the inductor.  This is exactly what we are doing by installing the cavity filter.  The hairpin input of the cavity is the inductor.  Even through it is a short, straight piece of wire, it has Inductive Reactance at 1024 MHz.

    The carrier lifetime is created by minority carriers conducting in the SRD in the normal forward direction.  When the diode is immediately reversed biased (the square wave transitions from neg to pos), the minority carriers will continue to conduct current, even though they are conducting in the reverse direction.  When they have lost their energy to conduct (at the end of the Lifetime), the diode will quickly cease conducting (1.2 and 7.2 cm). This is not exactly an immediate action.  The time it takes to cease conducting is called the Transition Time.  The faster this action occurs, the steeper the falling edge of the signal.  This action of the transition is what will create the many harmonics of the input signal.  A very fast transition time will create many more higher frequency harmonics.  I cannot measure this particular SRD's Transition Time, but it is likely on the order of 100 psec or less.

    This final picture is the same SRD device, but, instead of driving it with a square wave source, it is driven with a 13 MHz sine wave.  The action is exactly the same as the square source.
Pic of SRD sine input