Why the filter?
Several years ago I needed a "sharp" 160 Meter band-pass filter - one that would pass 160 Meters with little attenuation, but effectively block top-end signals on the AM broadcast band.
Nowadays, the AM (mediumwave) broadcast band goes up to 1700 kHz - but there are relatively few signals in the "new" portion (1610-1700) - but Murphy's law would dictate that one of those stations would be located near your QTH - but that wasn't the case here. The actual application was that this filter was to be used at the Northern Utah WebSDR where we have very strong (-20dBm, or about "50 over S9") signals coming across salt water (the Great Salt Lake) - one of which was from a transmitter on 1600 kHz.
If you happen to live near an AM broadcast transmitter and operate on 160 meters, you may well have faced challenges, yourself:
- A full-sized 160 meter antenna (dipole, vertical) will likely do a decent job of intercepting RF from any AM broadcast transmitter - particularly one near-ish the top end of the band.
- The filtering in most HF transceivers and receivers isn't particularly "sharp" and will likely do little to prevent a significant amount of RF energy from getting into the sensitive circuits.
- Both of these can cause overload of the RF amplifiers, mixers and/or non-linear responses in the PIN diodes typically used to select filtering.
- In a software-defined radio, a strong signal can also overload the same sorts of stages, but there's the additional problem of possibly overloading the A/D (analog-to-digital) converter - or at least, causing the receiver's gain to be reduced (to prevent A/D converter overload) to a point where it starts to become "deaf" to other, weaker signals. This "gain reduction" also has the effect of reducing the number of A/D converter bits that are used to represent weaker signals - something that can also increase distortion products.
The effects of all of these can a general degradation of the receiver, the most obvious being the generation of IMD (InterMoDulation) products, often manifesting themselves as spurious signal being produced in the receiver that may be harmonics of an AM transmitter and/or "mixes" of two or more AM broadcast signals. To a degree, these may be reduced by attenuating the input signal (e.g. the attenuator on the receiver) - and if these spurious signals' levels reduce in level more than the "real" signals on the band, that is a sure sign of overload within the receiver itself.
If the above are happening to you - and you care about 160 meters - a filter might help. 1
The challenge of such a filter
Achieving the combination of low loss at 1800-2000 kHz while providing a reasonable amount of attenuation at 1700 kHz - or even 1600 kHz - is a bit of a challenge as the percentage difference in frequency where you want signals to pass and and the frequency that you want to block is small, but this made the project even more attractive.
As I wasn't going to be transmitting through this filter, a bit of loss was acceptable and this filter ended up losing about 3dB (half of the power - or about 1/2 of an S-unit) through it. This may sound terrible, but if you have even a modest antenna for 160 meters, you will have way more signal+noise than is needed to deflect your S-meter significantly - even if you are lucky enough to be in an area with no man-made noise as the chart below indicates:
 |
| Figure 1: "Typical" noise floor for various radio environments. Because the above chart is based on a 500 Hz bandwidth, one would subtract 27dB from the power level to scale to a 1 Hz bandwidth. |
As you can see, compared to 40 meters, the noise floor - due to natural sources (the "quiet rural" graph) - is 5-10dB higher on 160 meters than on 40 meters, so it's likely - in most cases - that losing 3dB through a filter will go unnoticed: The ultimate test would be that of noting whether or not the S-meter read higher with the antenna connected (despite the loss) than without: If the former, you are probably hearing everything that there is to hear! 2
A practical receive-only filter
The schematic of the filter is here:
 |
| Figure 2: Schematic of the receive-only band-pass filter for 160 meters. Both "exact" and "standard" values for the capacitors and inductors are shown. Click on the image for a larger version. |
The filter isn't very complicated - using only ten components. As can be seen from the above diagram the "exact" component values were computed - but I also recomputed using "standard" values of capacitors and inductors and simulated both - first, the "ideal" values using the "Elsie" program configured to take into account the limited "Q" of real-world inductors and capacitors:
 |
| Figure 3: Plot using ideal values, from 1.5-2.5 MHz. Click on the image for a larger version. |
Now, using "standard" values for the capacitors - and some of the inductors:
 |
| Figure 4: Plot using "standard" values over the 1.5-2.5 MHz range. Click on the image for a larger version. |
The result is that the response is somewhat less flat - but only by about dB or so, most notably at the upper end. In both cases, the filter is attenuating signals at 1700 kHz by 15dB and at 1600 kHz by about 35dB. 15dB might not seem like much, but it represents a 32-fold decrease in signal level and this may well reduce IMD products to inaudibility. 3
Building the filter
In analyzing the schematic, you may note L1, L3 and L5 are paired with rather low-value capacitors (220, 150 and 220pF, for C1, C3 and C5, respectively for the "standard" value version) implying higher impedance. Conversely, inductors L2 and L4 are paired with high-value capacitors (14700pF for C2 and C4) implying low impedance and higher current.
 |
| Figure 5: As-built 160 meter band-pass filter on a piece of glass-epoxy board using "ME" squares. Click on the image for a larger version. |
The as-built filter is shown below:
The filter was built onto a piece of glass-epoxy copper-clad circuit board, the landings using "Me-Squares" from the QRP.me web store (link) glued to the substrate using cyanoacrylate ("super") glue. The in/out connections are in the lower left/lower right corners and the wound toroids are visible, glued to the board (to keep them from moving around - and to immobilize the windings to prevent mechanical de-tuning) using RTV ("silicone") adhesive. The three molded inductors are clearly visible as well. Precise alignment of this filter - mostly squeezing/spreading the turns on the toroids - is easily done with even the most inexpensive NanoVNA.
 |
| Figure 6: The filter installed in the chassis of one of the filter modules at the Northern Utah WebSDR. This filter was added on the port feeding the 160 meter receiver. Click on the image for a larger version. |
If you build this, DO NOT use disk ceramic capacitors for C2/C4 as 0.01uF and 0.0047uF are not likely to be temperature stable, low-loss NP0/C0G types - but rather they will probably be very temperature-unstable and lossy "Z5U", "Y5P", "X7R" or similar - and these types are completely unsuitable in this application.
As can be seen in Figure 6, the completed filter was installed in the filter module that feeds the 160 meter receiver at the Northern Utah WebSDR, but I could have easily put it in its own, shielded box.
Can a transmit-capable filter be constructed?
In theory, it should be possible to build a filter like this that would allow (survive!) being transmitted-through - but several things would need to be taken into account:
- Low-loss inductors. The loss through the inductors account for the majority of the losses here, mainly because - compared to capacitors - real-world inductors are terrible. For L1, L3 and L5, it may be that larger-gauge wire (16 AWG or so) on fairly large toroidal cores would suffice, but for L2 and L4 - which will be carrying quite a bit of current, - very heavy wire (perhaps 10 AWG) would be needed.
- Low-loss capacitors. Silver-mica capacitors would be used throughout. For C1, C3 and C5 some high-voltage units (1kV) should suffice, but for C2 and C4, paralleling a half-dozen or so 1kV silver-mica units to attain the desired capacitance - and to divide the current and losses - would be recommended.
Using the filter
As this filter has a rather high loss - which isn't really important for reception - that is not the case for transmitting: Running more than a watt or so through it would likely cause heating of the molded chokes - and even if lower-loss inductors were used, it would still have several dB loss and cause heating. If the molded chokes were replaced with toroidal inductors, it would have slightly lower loss, but likely be capable of handling QRP power levels (5 watts maximum).
What this means is that some means would be required to assure that this filter was used only used in receive: Some transceivers have connectors to allow the insertion of a filter in the receive signal path, but it may well be that RF relays will be required to switch the filter in/out.
Footnotes
- A filter will help ONLY if the spurious signals are being generated within the receiver itself. Low-level IMD (intermodulation) products can also be generated in strong RF fields by non-linear junctions in the vicinity - which can include metal fencing, rain gutters, rusty wires, corrosion, or even the "re-radiation" of IMD from another device that may have high-level RF energy on it.
- It is usually the case that modern receivers have way more gain than is necessary to receive signals at the noise floor - particularly at lower frequency where the natural noise floor is usually quite high. What this means is that you can "throw away" signal on the input of your receiver and not actually "miss" any signals at all. For example, if you have an attenuator on your receiver and without the attenuator the noise floor is S-6, but with the attenuator is still an S-3 - and the noise floor of your receiver without an antenna connected is S-0 - signals in the noise at S-3 with the attenuator switched in are not being lost: It is only how far above the signals are above the noise that affects their readability. As noted above, in the presence of very strong signals (as in the case of a very nearby AM broadcast transmitter) or even in the case of an event like Field Day where other transmitters are nearby - if you can still hear the "band noise" with the attenuator switched in, you may well be better off overall by using the attenuator - which can minimize the generation of IMD products within the receiver. For more information about this, see the earlier blog entry "Revisiting the 'Limited Attenuation High Pass' filter - again" (link).
- As a general rule, IMD products resulting from nonlinearities within an amplifier (or receiver) will diminish by about 3dB for every 1dB of total signal reduction. In our example where, at 1700 kHz, the band-pass filter reduced the undesired signal by 15dB, this could, in theory, reduce the IMD by 45dB or so - about 5-8 S-units, depending on the radio. If this IMD was "only" about S-9 to begin with, reducing it by 40+dB may make it inaudible - even though the filter knocked it down by only 15dB in the first place.
* * * * *
This page stolen from ka7oei.blogspot.com
[END]
