Band Pass Filter Board

Updated 1 Nov 2024 MW

The bandpass filter board is in some regards might seem like a simple thing. Don’t be fooled. It is a central and extremely important part of any radio design. See it as a funnel, two directional for RX and TX. If there are any obstructions in this tunnel it will reflect significantly on the performance of both receiver and transmitter. When testing or finding one particular band, say, is not working well (low output power, or poor performance on receive), suspect the BPF first!, and re-test it. For this reason it is worth a lot of effort to get this board right, and indeed there have been quite a few prototypes and a lot of time spent in perfecting this board.

• Up to 16 BandPass filters:
  • 1800 & 630 Meters LF Bands in bays 1 & 2
  • 160 through 10 Meters HF Bands in bays 3 – 13
  • 6, 4 & 2Meters VHF Bands (Not all countries are allowed the 4M band) in bays 14, 15 & 16
  • 8 Mtrs or any other bands can be can be added.
• Each selected with 4 bit parallel TTL level logic (0000 – 1111 Binary) supplies to the board. This board is therefore suitable for non HamPi Radio projects.
• Switching can be done with 16 PIN diode switches The loss of a PIN diode is insignificant in the scheme of things and never loose a QSO of them.
• A LED indicator is there for each BPF bays to indicates which Band is selected. It is extremely helpful (and less frustrating) to be tuning the correct coils!
• 14dB or 18dB Attenuator (or any other, depending on the 3 resistors fitted)
• 18dB Pre-amplifier (remember one S unit is 6db below S9, and 10dB above S9 on HF)
• 2 x notch filters for tuning out Local Oscillator leakage. Very important on TX.
• 2 x Relays for selection of external LF and VHF modules.

Note in the picture above. There are three tuned circuit stages to each Band Pass Filter. Each stage has a transformer and capacitor that resonate together.

BPF Considerations:

Here are a few things I have learnt in designing and building filters.

• Aim to have all capacitors as low in value as possible, so as to achieve the best performance / least loss.
• If the loss on any band it too high (anything more than 3db loss should be a concern). The type of capacitor used can be a big issue. simply by changing capacitor type I have seen the difference between <3db loss and 10db loss or more.
• If the capacitance across the 10mm transformer is too Low in value, tuning DOWN to to the desired frequency will not be possible, and a higher value cap will be needed. You will find that the cores in the 10mm transformers are turned in clockwise very low. If the capacitance is too high, performance will suffer, shown by too much loss and there might be a triple peak and the pass-band will be too wide.
• The two Coupling Capacitors between the three transformers couple the stages and also set the bandwidth of the circuit – how wide the bandpass is. The aim is to have the lowest overall loss, while also having a wide enough – but not too wide bandwidth. The larger the capacitance the wider the bandwidth, but the loss lower overall loss. The lower the capacitance the sharper the notch and narrower bandwidth, but then loss will be higher. The aim again is to cover the particular band (Say 14.000 to 14.350mHz for 20 Meters) with a relatively flat top, and to see a sharp roll-off as the signal gets out of band, while filtering out as much out-of-band interference as possible (see below).
• With any 10mm type Ferrite transformer, as a rule of thumb, if the ferrite slug is close-ish to the top of the core but not protruding past the top this will be the best performance.
• In simple terms see the capacitors as slugging, or dampening the ferrite transformer. The higher the capacitance the more dampening. A bit like putting a finger on a piano string near the end. Too much pressure and the string cannot freely vibrate.
• See the image below. This is an example on 20 Metres and the loss here is 1.79dBm. Note that at 1mHz each side of the centre frequency the filter has achieved a 10dB attenuation, and at around 1.5mHz it is -20dBm attenuation.
• Tune all three transformers to achieve the shape below. The centre transformer only changes the frequency and not the amplitude of the filter. The two outer transformers adjust the amplitude mostly, and not much change in frequency.
• It is desirable to have a Spectrum Analyser with a tracking generator option to do this. The one below does a splendid job, but costs around NZ$2500. If you do not have one or you have no access to one (another HAM?, Radio Club?), a NANO VNA will also do the job, and are much cheaper. It can be done with a signal generator and a good oscilloscope too monitoring the RF and noting the loss, peaking the sign wave.
• For selecting parts and tuning use the band marked ’12’, and fit sockets for easy part substitution. This band is at present unused (11mtrs) and is covered by the 10m band 13. To enable this band fit a link to J12 marked ‘J12 Test 13’. See picture below.

Construction

1. The two Notch Filters on this BPF must be carefully tuned to the IF frequency to tune out any IF frequency leakage going into (RX) or out of (TX) the Radio Board. To do this introduce a large signal (0dbm or 10dbm) at the IF frequency (For example 7.9985MHz). Tune the two Notch Filters carefully for the minimum signal on a Spectrum Analyser. for 0dbm in this should tune down to about -65dbm (not far above the noise floor) or for a 10dbm input to around -59dbm.
2. All loss figures in the following table are measured on the board including the PIN Through circuitry, which accounts for a small loss in itself (0.4dB).
3. Band 1, 1900m. Will need a 4-800uH coil (red), found in old Medium Wave transistor radios. It was hard to get these manufactured.
4. HamPi 4356, 3245, 2224 & 1024 coils are custom wound 10mm transformers like the now obsolete TOKO type. If you want to self wind old TOKO type cans please contact me for turns details. The reason for going to the trouble of getting them custom wound was I couldn’t get good enough performance from ones available (Too much loss).
5. The 11mtr band is unused, and works on the 10mtrband coils. For broadband receive short circuit this BPF
6. 6, 4M bands are ‘OK, but need BW and loss improvement and to be done with open wound coils.
7. 2M band not done yet. This will need open would coils.
8. Note: Preamp is a supplied 12V directly from the CPU board, and the Through and Attenuator run on 5V sourced on this board – hence the different resistor values to maintain a similar current through the PIN diodes.
9. There is an added complexity when using a 10.7mHz IF (Or other IF frequency close to a HAM band) in that the skirt of the two notch filters on this board will affect the bandpass filter for 30 meters (10.1mHz). Since this is a low power band it won’t matter on TX too much as there will be enough drive, but in receive it could add extra loss tot he filter. Tune these two notch filters exactly on the LO frequency, 10.7, or 9mHz etc. This stops any stray IF signal affecting the transmitter and also stops any stray IF being seen at the antenna and radiated on receive (and upsetting your old AM/FL radios in the house).

Other Transformers

1. T1 (preamp) BN43-2402 2t : 8t ct 0.2mm. This means two turns on the primary and 4 turned + 4 turns on the secondary. One turn is counted as the wire passing through one hole AND back through the other hole.
2. T2 (preamp) BN43-2402 10t ct : 2t 0.2mm. This means ten turns on the primary and 1 turned + 1 turns on the secondary. One turn is counted as the wire passing through one hole AND back through the other hole.
3. Other transformer ferrite options have been banded around on the internet for similar pre-amps including BN61-202 (0.5dB Less NF) and BN-73-202 . I have used various ones larger than the BN43-2402 and they work fine, generally using the type 43 material. When experimenting remember to test for coverage on all the bands you want the pre-amp to work on.
4. In general a pre-amp will not be needed below 10mHz on any reasonable antenna. On higher bands and less than ideal antennas a pre-amp can be of help when the band is not fully open, but it will add noise.

Attenuator Choice

1. If you are using the lower HF bands it can be best to use a higher attenuation where signal strengths are very high.
2. Attenuation levels can be chosen as follows:
   • -12db attenuation fit 82/91/82 R resistors (2 ‘S’ points)
   • -14db attenuation fit 75/120/75 R resistors
   • -18db attenuation fit 68/200/68 R resistors (3 ‘S’ points)
   • Note: Remember each step on the S Meter is 6db up to S9 and 10dB over S9

Adding the Hand Soldered parts

1. On Issue 3 and before boards: Solder 32 x 100nF 0805 SMD capacitors to the underside of the board. (This could not be done by the PCB manufacturer).
2. On Issue 3 and before boards: Find all the 48 Jumpers under the 10mm transformers on the underside of the board and do a solder bridge from the centre of the jumpers ‘to the outer pin of the transformers’. This leaves the transformer centre pins unconnected on the Jumpers. Now find the two remaining jumpers under the two Notch Filter 10mm Transformers and do the same bridge.
3. Carefully inspect the PCB solder in all parts not supplied already fitted. These will be:
   • If you are not going to be transmitting on the LF and the 2mtr band (This is a part of the project that is not developed yet, and is intended to be in a separate case) then do not fit relays K1 and K2, or coax connectors J2 or J3. Do fit J1 a J6 pin 0.1″ pin strip and fit a push on Pin Header Shorting Block Connector
   • Fit J12 is a similar 2 pin pin strip. This is used to select BPF 13 for experimentation without the need for any other HamPi boards. (More on this later)
   • Fit J13 (3 pin strip) and J15 (2 pin strip). These are used for testing also when no other HapPi boards are needed. J13 had 3 pins. Using a 2 pin Pin Header Shorting Block from the middle to one of the outside pins selects the Attenuator or the Through RF path. Using the Shorting Block on J15 selects the Preamp.
   • Coax connectors J2 (HF Input) and J9 (RF Output)
   • Inductors L1 to L12. These are through hole parts (THT) and were choses SMD inductors were found to not be suitable.
   • 2 x 16 pin DIL Sockets for the 2 x 74HC156 ICs
   • Q20 and Q21 are not normally fitted (to save fabrication cost). These are 2 x J310 devices. The footprint caters for both SMD or THT leaded.
   • It is suggested to not fit IFT38 and IFT51 just yet. These are to notch out any IF frequency leakage, which there will be some. They are very important and must be eventually fitted before going live on air. You don’t want to interfere with anyone listening on your IF frequency (say 8MHz)!
   • J11 16 pin IDC Header. This eventually connects the BPF board to the CUP board with a ribbon cable. Hint: If you happen to solder this the wrong way around! No need to unsolder it to correct this error. Use strong pliers and carefully pull off the plastic surround / body by pulling a little bit at a time each end, and simply turn it around ant push it back on. (Guess who has done this once!)
   • Fit J10. This supplies 12V / 13.8V to the board for standalone testing, with no other HamPi boards needed.

Testing and Alignment

1. Carefully inspect the board to ensure all the necessary parts have been correctly mounted.
2. Plug in the two DIL 74HC156 devices in the correct orientation. This device must be an HC part as only the HC will sink enough current to correctly switch on the PIN Diodes. (1N4148s)
3. Solder in the three Band Pass 10mm transformers for the band you want to assemble and try first. Say 20mtrs. Also add the 5 capacitors associated with the filter.
4. The board can now be powered on with 12V or 13.8V at J10.
5. If you have a Spectrum Analyser + Tracking Generator connect the TG output to the HF IP J2 and the analyser input to RF O/P J9
6. Under construction…
7. .
8. .
9. .
10. IF Notch Filters IFT38 / C75 / C130 and IFT51 C2 / C131. Set up procedure… MW