Crystal Filter

Board 1 || VFO || TV & FM, BC Filters || RF Amplifier || Bandpass Filters || First Mixer & Amplifier || Crystal Filter || Board 2

4.0 & 3.547MHz Crystal Filters

Band Image Scheme || Crystal Filter/Crystal Oscillator Work in Unison
Switching Action || Adjusting Crystal Filter Bandwidth || 12 Volt Isolation and Shielding || Crystal Filter Inputs

The crystal filters are the control center of this receiver. All the important IR (infra-red) action occurs here.

Both crystal filters will hit two bands with a technique called band imaging. By using the difference and sum of a crystal filter frequency with a VFO frequency, two amateur bands can be received with each crystal frequency.

For an example of this technique, check out the 1988 ARRL Handbook for Radio Amateurs, Chapter 30. "A Band-Imaging CW Receiver for 10 and 18 MHz".

The upper IRED is pointing across the board to a phototransistor located between the two VFO amplifiers. This phototransistor turns on the 10.545 VFO relay. The lower IRED is pointed upward to hit the photo transistor at the crystal oscillator on the upper board (mounted upside down on the upper board underneath the crystal oscillator), which turns on the crystal oscillator to 3.547MHz.

Band Image Scheme

According to the article "A Band Imaging CW Receiver for 10 and 18 MHz", "A band imaging receiver appeared in every edition of this Handbook from 1953 through 1966, from "A Two-Band Four-Tube Superheterodyne" in 1953 to "The HB-65 Five-Band Receiver in 1966."

This receiver uses two band imaging schemes. The difference between the two band imaging schemes is 455kHz. This allows a 455kHz strip to be used giving excellent AGC and gain.

The first band imaging frequency uses the 3.547 MHz crystal filter to receive the 40 and 20 meter bands. One VFO at 10.455 MHz and up can cover both bands. However, the 40 meter band does not start until 10.545 MHz, which is inconvenient for quick band switching. For that reason, the VFO has two relays, [one set for 10.545 MHz (40 meters) and one set for 10.455 MHz (20 meters)] so both bands can be set at the beginning of the band.

The second band imaging frequency uses the 4.000 crystal filter to receive 30 and 17 meters. A 14 MHz VFO is band imaged with the 4.000 MHz crystal filter.

The 17 meter band starts at the VFO frequency of 14.068MHz to 14.168MHz and the 30 meter band starts at 14.100MHZ to 14.150MHz. Since both bands are covered completely by a 100KHz spread on the VFO, one VFO covering the 17 meter band does the job.

Crystal Oscillator
The BPX38, the photo transistor, is mounted underneath the board and cannot be seen in this photo. The crystal on top is the 4.000MHz crystal. The 4.000MHz crystal is turned on in the photo. The other LED below the one that is on, turns on the 3.547MHz crystal.

Crystal Filter/Crystal Oscillator Work in Unison

Two different frequencies are needed at the crystal oscillator, feeding the second mixer, to provide a 455kHz output with each crystal filter.

Diode switching is used to switch the crystals at the oscillator. LEDs are used instead of diodes. The LEDs provide an indicator of which crystal is turned on, providing diagnosis of proper operation of the switching circuit.

The crystal ground circuit first goes through a capacitor, and then to the LED which grounds the crystal. A switching circuit using IRFU220s switch the crystals and LEDs.

The capacitor serves two purposes. The first one is to trim the frequency of the crystals so that the BFO frequency is the same when switching between the crystal filters. A value of 200pf was found to get the frequencies close. An optional trim cap (unmarked holes on the PCB) was added in parallel to the 4.000 crystal capacitor, to help get them exact if the operator desires.

The second purpose of the capacitor is to block the DC voltage, which turn on the LEDs, from getting to the crystal. A typical diode switching circuit using 1N914s use 1ma or less. The LEDs use 10ma.

Switching Action

When the 3.457 MHz crystal filter is on, the IRED at the input activates a phototransistor at the VFO which turns on the 10.545 VFO relay. With the oscillator running at 4.000 MHz, and with the 3.547 MHz crystal filter, the second mixer puts out 455 KHz. The receiver is now ready for the 40 and 20 meter bands.

When the 4.000 MHz crystal filter is on the VFO relays are turned off. The IRED at the input turns on the phototransistor at the crystal oscillator, which turns on the 3.547 MHz crystal. The output of the mixer is 455 kHz. The receiver is ready for 30 and 17 meter operation.

Adjusting Crystal Filter Bandwidth

Wes Hayward's article, "Designing and Building Simple Crystal Filters," (QST, July 1987, Page 24) provided a revolution in building high performance homebrew receivers. After this article, any homebrewer could build crystal filters that performed exactly the way he preferred with no math involved. Hayward's design was installed in the receiver exactly as he suggested in his article.

Identical capacitors (100pf) were used in the filters and provide a compromise between SSB and CW bandwidth. Raising the value of the capacitors tightens the bandwidth and lowering the value widens the bandwidth. Since the impedance is difficult to change, his other formula, raising/lowering the impedance tightens/widens the bandwidth, can't be used as easily.

The 3.457 MHz crystals are very high "Q" crystals yielding a sharp bandwidth with the 100pf capacitors. For SSB reception it is recommended that the capacitors be changed to 39pf. This value is not critical.

The 4.000 MHz crystals have a lower "Q" value than the 3.547 MHz crystals, and SSB reception is fine with the 100pf capacitors. In fact, the bandwidth with the 100pf capacitors may be a little wide for reception of CW in a crowded band.

To provide a sharper bandwidth with the 4.000 MHz filter, parallel higher value capacitors with those on the board (soldered underneath to the existing capacitors) - 100pf to 200pf added to the ones on the board would optimize this filter for CW.

In the SWL design of this receiver, two crystals in each filter are removed to provide the wide bandwidth needed for AM detection.

Emails from ELR builders have indicated that filter performance could be improved. Wes Haywards' article was used for the simplicity of its implementation and understanding, but there is room for improvement.

One of the best articles on designing crystal ladder filters was published recently in QEX, "Crystal Parameter Measurement and Ladder Cyrstal-Filter Design", Sept/Oct 2003, by Randy Evans, KJ6PO. He may have the easiest program for determining capacitor values for a ladder filter. Input crystal values into a spreadsheet program available at the ARRL QEX download site. Spreadsheet program file is

This article shows promise of improving the filters used in the receiver. Also, computer audio spectrum analyzers show the response of crystal filters and can help in improving the design.

12 Volt Isolation and Shielding

All the 12 Volt lines were moved as far away from the crystal filter as possible. Since the 12 Volt line from the mixer/amplifier was subjected to the highest level of RF in the receiver, it was important to have the 12 Volt line isolated from the crystal filter.

Another isolation technique was to implement all the switching at the input of the crystal filters. The outputs are tied together with a 1:4 tapped transformer so that there is no 12 Volts near the output of the crystal filters.

Crystal filter shielding is important for a clean receiver, especially when using efficient outdoor antennas.

To further isolate the 12 Volts and stray signals from the crystal filter, a row of soldering pads to hold a shield, was installed between the switching circuit and the crystal filters.

To help crystal filter performance, shields can be installed between the first/second and second/third set of crystals. This provides maximum isolation for the output signal from the crystal filter for delivery to the second mixer. Soldering pads were placed at these locations to facilitate the mounting of the shields. The image below shows how the shields are installed.

The output of this board is the box labeled "Xtal Filter Out", with two soldering pads. The two pads are used to install a loop of wire for easy soldering and unsoldering during building and testing.

The "Xtal Filter Out" is attached to the "Xtal Filter In" inside the box of the Second Mixer on the second board.

Almost all the spurs in this receiver are the result of the VFO frequency getting into the second mixer. Always use shielded coax between the crystal filter output and the input of the second mixer to keep signal pickup to a minimum.

Crystal Filter Inputs

The inputs are switched by two LEDs in series, one an LED and the other an IRED which switches the VFO frequency or the crystal oscillator frequency. The two LEDs in series gave the best isolation of all the techniques tried and was easy to implement.

While experimenting with the two-LED switching circuit, it was found that LEDs have widely varying capacitance values that effects the loss through the LEDs. Also, the resistance used in the ground circuit of the LEDs makes a large difference. I found the resistance used and the capacitance of the LEDs have a relationship.

The first LED in the switch has the most effect on the loss of the switching circuit. A visible LED is used and the resistor value used is 1K, which makes for a bright indicator. The resistor value used with the IREDs is 470 ohm. Lower values increase isolation and lowers the impedance to the crystal filter.

For the lowest loss through the switching circuit, use a 2.2K resistor with the LEDs and a 1K resistor with the IREDs. The 1K resistor is probably the maximum value that can be used and still have enough IR energy to switch the phototransistors. The resistors are in the ground circuit of the LEDs.

The red LEDs supplied with the kit (the first LED in the switch) have about 5-10pf capacitance. LEDs with low capacitance values provide better isolation between the filters.

The super-bright LEDs have a much higher capacitance - approximately 100 to 200pf. A direct replacement of the stock LEDs (bright red) with the Super-brights provides the lowest loss of any of the combinations. You can lower the resistance values to 470 ohms and have the same loss as with the stock LEDs. This provides better isolation with the super brights. This will also increase the brightness of the super-bright LEDs.

It appeared that the lower the capacitance value of the LEDs, a higher resistance value was needed to keep losses at a minimum. Higher capacitance LEDs with 2.2K resistors have the least loss, but the isolation goes down slightly.


"Crystal Parameter Measurement and Ladder Cyrstal-Filter Design", QEX, Sept/Oct 2003, by Randy Evans, KJ6PO. Claims to have the easiest program for determining capacitor values for a ladder filter. Input crystal values into a spreadsheet program available at the ARRL QEX download site. Spreadsheet program file is

"Ladder Crystal Filter Design", by J. A. Hardcastle, November 1980, QST, Page 20.

"A Unified Approach to the Design of Crystal Ladder Filters", by Wes Hayward, W7ZOI, May 1982, QST, Page 21. This was the first article that hinted at the fact that regular computer crystals or color burst crystals could be used in a very effective homebrew crystal filter.

"Designing and Building Simple Crystal Filters," by Wes Hayward, W7ZOI, QST, July 1987, P 24. This article showed how using the same capacitance for each capacitor in a crystal ladder filter gave very good results, very inexpensively, and easily adjusted bandwidth.

Crystal ladder filters - This page contains JavaScript tools to design and simulate crystal ladder filters which are widely used in SSB transmitters and CW receivers. Could be used to fine tune the xtal filters used in the receiver. Excellent page!

Board 1 || VFO || TV & FM, BC Filters || RF Amplifier || Bandpass Filters || First Mixer & Amplifier || Crystal Filter || Board 2

Send E-Mail || Amateur Radio Receivers || Electroluminescent Receiver

Last Update: 9/5/06
Web Author: David White, WN5Y