| CW Filtering |
Filters for SSB come standard in all HF transceivers, and they usually do a pretty good job. But SSB filters often do not do a good job for CW. Sometimes even if the transceiver has a built in CW filter, it doesn't do well. If your doing digital modes, like PSK-31, you might want to use a better filter. But some transceivers don't allow you to use CW filters when your in the SSB mode.
If you bought a new rig today, you might make sure it had the best in filters, SSB and CW. Even if you don't know CW. You might want to take the plung one day, maybe not. But it will help the resale value, should you decide to upgrade again.
But what if you are not getting a new rig and your current rig doesn't have a good CW filter. Of course, its always best to add your filters in the IF stages, before detection. But some extra filtering in the audio stages would definitely help.
So, this section is about narrow, audio level, active filters. These kind of filters are easy to construct, relatively cheap, and easy to design for your specific needs. I have several web pages on active filter design (Band-Pass, Band-Reject, Low-Pass, High-Pass). But the design is only for a single stage (second order) filter. These filters work fine, if that's what you need. To get higher orders of filtering that might be required in CW work, you can put two, or more stages, in series. But stringing out multiple stages, each tuned to the same frequency, can cause objectional ringing.
But there is another way to get good out-of-band rejection without all the ringing. Some time back, in 2002, Parker R. Cope W2G0M/7 wrote an article on Multiple CW Filters. It was a good article that covered two, three, and four section Staggard filters. A single stage filter is included as a introduction to the others. Staggard filters are simply multistage filters that have each stage tuned to a different center frequency. The advantages of this are detailed in the linked article and is well worth reading about. The article has some minor errors, and omissions.
I am using the examples from the linked article, I have changed the ordering and automated the design equations. This gives you a good idea of the steps needed to design these filters and gives you a chance to try your own values.
| The Equations |
First, let me restate the equations. The equations may be in a different order, and some variables may not exactly match the original article, but they are still correct. For the purpose of this discussion, it is assumed that "C1 = C2" and K (Gain) is set to "1". But that is not necessarily a limitation of the filter.
The capacitance for "C1" is approximated by C1 (µF) ≅ 10/F0. For a center frequency of 800 Hz that would be 10/800 = 0.0125 uF. The closest SVC (Standard Value Capacitor) would be 0.012 uF or 0.013 uF but a more common, and readily available capacitor might be 0.01 uF. Then, a minimum value for C2 is calculated such that C2 (µF) > (C1 (µF) × (K‑Q2)) / Q2. In the case of "C1 = 0.01 uF", "C2" should be "> 9,844 pF". Which, again, is very close to "0.01 uF". Making the two equal, simplifies things a little bit.
| Single Stage Band Pass Filter |
|
Cent. Freq., F0 Hz |
Stage Q |
||
| C1 |
C2 |
||
| Res. Tol. |
|||
This Single Stage Band-Pass Filter is really a introduction to the Two, Three, and Four Stage Band-Pass Filters. It provides the basic equations and selection criteria for a single stage. The center frequency on the Single Stage is chosen to be a fixed frequency. Whereas, the Two, Three, and Four Stage Band-Pass Filters use stagered frequencies that are offset from the filter center frequency.
You might notice, that the single stage filter, in my drawing and in W2G0M/7's description, employs as a default 0.01 µF capacitors. The multi-stage filters, on the other hand, employ larger value capacitors (0.1 µF). The reason for this is that larger value capacitors reduce the resistor values. And, of course, you can try this yourself by changing the capacitor values.
The values printed on the schematic are the nearest SVC (Standard Value Capacitor) and nearest resistor values to the calculated values. Reducing the Resistor Tolerance will select resistor values that are closer to the calculated values.
| Two Stage Staggard Pair Band Pass Filter |
| Cent. Freq., F01 Hz |
Stage 1 Q |
Cent. Freq., F02 Hz |
Stage 2 Q |
| C11 |
C12 |
C21 |
C22 |
| Resistor Tolerance |
|||
The diagram below is for a Two Stage MFB Staggard Pair Band-Pass Filter. The diagram reflects the component values that are calculated based on the parameter entry area above the drawing. The filter is a MFB (Multiple Feed Back) type of filter, the same type used in the original artical. The initial values in the calculator are from the article linked above. The Gain of each stage, is fixed to 1.
A staggered pair uses just two op-amps. The response falls off at 12 dB per octave of bandwidth, but requires lower Qs and is more tolerant of component variation. The stages of a staggered pair are tuned to F0 +/- 0.35β0 each with a dissipation of 0.71δ. β0 is the 3dB Band Width and δ is β0/f0. A Flat Staggered pair centered at 800 Hz is made up of two stages, one tuned to 835 Hz and one tuned to 765 Hz, each with a Q of 11.26. Flat Staggered just means there is no ripple in the passband. The tuned frequencies and RCs are shown in the drawing.
Note that while the author specified F0 +/- 0.35β0 in the original article, the tables for the staggered pair uses a Q that produces a smaller band width than you might think. With smaller bandwidths, this may be considered a little Over Staggered. So, to prevent any possible ringing, you might want to slightly reduce the Q of each stage.
Over Staggering a pair, narrows the 12 dB bandwidth and increases the ripple in the passband. With about 1 dB ripple in the passband the octave bandwidth (200 Hz) will be down about 14 dB. Not a great improvement, but every little bit helps.
| Three Stage Staggard Band Pass Filter |
| Stage 1, Gain = 1.0 | Stage 2, Gain = 1.0 | Stage 2, Gain = 1.0 |
|---|
| F01 Hz |
Q1 |
F02 Hz |
Q2 |
F03 Hz |
Q3 |
| C11 |
C12 |
C21 |
C22 |
C31 |
C32 |
| Resistor Tolerance |
|||||
The Three Stage Staggard Band Pass Filter uses three Op-Amp stages. The first stage is centered at your design frequency, 800 Hz in this example. The other two stages are at F0 +/- 0.35β0 The Q's are not as high as those required by the quadruple and the filter is less sensitive to component variations.
| Four Stage Staggard Pair Band Pass Filter |
| Stage 1, Gain = 1.0 | Stage 2, Gain = 1.0 | ||
|---|---|---|---|
| F01 Hz |
Q1 |
F02 Hz |
Q2 |
| C11 |
C12 |
C21 |
C22 |
| Stage 3, Gain = 1.0 | Stage 4, Gain = 1.0 | ||
|---|---|---|---|
| F03 Hz |
Q3 |
F04 Hz |
Q4 |
| C31 |
C32 |
C41 |
C42 |
| Resistor Tolerance |
The Op-Amp used in this filter design is a TLC074, which is one of the Op-Amps recommended by the author. The TLC074 contains four Op_Amps and only requires a Single Voltage Power Supply. Shown on U1B, are two pins (V+ and V-).
| Power and Ground |
The power/ground connections for 1, 2, and 3 stage filters is shown on the drawings for those filters. I ran out of space to put it on the 4 stage filter schematic. However, the power/ground configuration on the 4 stage filter is the same as the other filters.
How you show the power connections depends on the package type. Some device packages contain multiple devices. e.g. the TL081, TL082, and TL084 contain 1, 2 and 4 individual Op-Amp devices in the package, respectively. And, while the devices are sepatate, a single power connection is all that exists. In most cases, with multiple devices, only one devices shows power connections.
The Op-Amps need to be operated from a split supply in order for the filters to handle bi-polar signals. This returns the non-Inverting inputs to a virtual Ground that is midway between +V and -V. Since there is no current into the non-inverting inputs, a bypassed resistor divider across a single supply can provide the relative plus and minus supplies. Depending on the Op-Amp, a voltage between 12 and 24 volts can be used.