Jul/Aug 2004 15

22525 SW BaselineHillsboro, OR 97123rob@pythonemproject.com

Crystal Filters with VariableBandwidth and Constant

Center Frequency

By Robert Lytle, N3FT

This new filter was inspired bythe Ten-Tec Jones filter (LeeJones, WB4JTR, US patent

#5051711). The Jones filter achievesvariable bandwidth but with a shiftin the center frequency of the filter.Depending on the application, thismay be good or bad.

For a simple receiver, the Jones fil-ter could be used to advantage by set-ting the frequency shift so that thewide and narrow bandwidths are oneither side of the BFO frequency. Thenarrow side would be for CW and thewide side for SSB1. Fig 1 shows theJones filter. As can be seen fromthe figure, the shunt varactors are1Notes appear on page 17.

The author presents a new approach to variable-bandwidth crystal filters allowing the center frequency

to remain constant while bandwidth is changed.

Figure 1—A typical Jones filter with voltage-controlled matching networks, plus a simplediagram illustrating the capacitive pulling effect on a crystal.

pulling the frequency of the crystals,much as in a VXO. Fig 2 shows theresponse of a Jones filter at minimumand maximum usable bandwidth. Re-fer to Fig 6 for the test setup. Noticethe shift in center frequency. For ap-plications needing a fixed center fre-quency (eg, spectrum analyzers andmore sophisticated receivers), the fre-quency shift of the Jones filter may notbe a good match for the application.

The key to developing the new fil-ter was the realization that certainvaractors had novel properties whenconnected in series. To get to this un-derstanding, we need to discuss the Cvs. Vr curves for varactors. For anyvaractor one can express the capaci-tance vs. reverse bias by the equation:

)V(V

CC

r

o(1)

Gamma (!) is a function of the dop-ing profile of the diode. V" is the turn-on voltage of the diode. Co is a constantdepending on the diode geometry.

It is particularly interesting if thegamma of two series-connected diodesis close to 1.0. In that case, one canshow algebraically that the capaci-

Figure 2—The measured insertion loss vs.frequency (vs. Vcontrol) of a three-crystalJones filter. Top—maximum bandwidth,bottom—mimimum bandwidth.

Figure 3—Diagrams to illustrate the independence of Vcontrol on the series capacitancepresented to a crystal.

Figure 4—The schematic diagram of the new filter. Notice the voltage-controlledmatching networks. These were borrowed from the Jones filter and work very well.

tance of two series connected varactorsis no longer dependent on the voltageon the middle node. Referring toFig 3, simply substitute the varactorequation into the equation for series-connected capacitors, adding Cshuntto the second capacitor (i.e. C3 in se-ries with C1+C2). C1 is the shunt cap,which normally would shift the filtercenter frequency. Now only the volt-age at point B sets the capacitance atB. The equation is:

)2V(VC

Cr

ofixed (2)

Where Vr=Vfixed. The Vcontrolterms cancel out.

What this means is that by usingvaractors with a gamma of nearly 1.0,a network can be constructed whichallows for the bandwidth adjustment ofa filter without upsetting the center fre-quency of the filter. Many varactorshave been found to have gammas closeto 1.0 over at least part of their ranges.

For example, the MVAM125,MMBV109L and MMBV609L may beusable. Some of these are smaller-valueSMT types, so they would have to beparalleled. The varactors chosen for thisproject are particularly good in this re-gard. Although the retailer (Hosfelt2) ismarking them as matched triplets ofMV1662s, the diodes are much toohyperabrupt to be of that type.Hyperabrupt refers to the diode dopingprofile. Abrupt junctions tend to have asmaller ratio of maximum to minimumcapacitance whereas the hyperabruptdiodes have a large ratio.

Fig 4 shows the entire test filter.The potentiometers on each end areadjusted for flattest passband re-sponse. Each 6 µH inductor is simplyfour turns on a BN61-202 balun core.A smaller core will work just as wellwith a few more turns added. Thecrystals used in this project are all5.200 MHz, taken from carrier filterson a surplus ATT FDM unit. The fil-ter cans were opened and the crystals

extracted. Crystals for other frequen-cies can be used, with varactors of anappropriate Co .

Fig 5 shows the passband responseof the new filter at minimum and maxi-mum usable bandwidth. See Fig 6 forthe test setup. Notice that the centerfrequency remains fairly constant. Thesecond trace on each plot is the filterinput return loss (add 10 dB).

In conclusion, a new type of vari-able bandwidth crystal filter has beendeveloped that maintains a constantcenter frequency and may have usesin test equipment and radio receivers.

Notes1Conversation with W7ZOI.

Figure 5—The measured insertion loss ofthe new filter as a function of Vcontrol.Top—maximum bandwidth, bottom—mimimum bandwidth.

Figure 6—The test setup used tomeasure the insertion loss of thefilters.

2Hosfelt Electronics, 2700 Sunset Blvd,Steubenville, OH 43952, 800-524-6464,p/n MV1662/S. At last count they had over5000 of these left. There are three diodesper package.

Rob Lytle was originally licensed asa ham as WN3YXB at the age of 12. Atthe age of 16 he became an Extra Classham with call sign N3FT. A few yearslater, he graduated with a BSEE fromthe University of California at Berke-ley. Most recently he had been employedas an Applications Engineer at TriQuintSemiconductor in Hillsboro, OR. Robstill enjoys the hobby with emphasis onconstruction projects, antenna projects,and working 6 and 160-m.

Photo: IOWA TOURISM OFFICE

Des Moines, Iowa is the place to be September 10-12 for the ARRL /TAPR DigitalCommunications conference at the Holiday Inn Des Moines—Airport and ConferenceCenter. There is something for everyone at the conference, including forums on softwaredefined radio (SDR), digital voice, digital satellite communications, Global Position System(GPS), precision timing, Automatic Position Reporting System® (APRS), high-speedmultimedia and much more.

APRS is a registered trademark of APRS Engineering LLC.