Band Pass Filtering For The G6LBQ Irwell HF Transceiver Part 1

 

HF Transceiver - Band-Pass Filter Module

"High Performance RF Band Pass Filtering for the G6LBQ Irwell HF Transceiver"

In this blog post, I lay the foundations for a high-performance RF Band Pass Filter  suitable for the Irwell HF Transceiver as well as SDR receivers and other transceiver builds.

For my own transceiver project, the filter module needs to meet the following criteria:

• Have a sufficient number of filters to enable general coverage reception of 0.5MHz to 30MHz and provide uniform high performance for any given frequency within this range.
  
  
• Be usable on the transmit path with a rf power handling rated at 1 watt.
  
  
• Provide good suppression to frequencies that fall below or above the desired frequency pass band.
  
  
• Provide low insertion loss within the wanted pass band & good return loss.
  
  

In order to gain a better understanding of the design criteria, it is worth outlining the basic concept of the band-pass filter but feel free to skip this section if you are already conversant with the basics.

The job of the band-pass filter is to block unwanted signals in the frequency spectrum and only allow frequencies to pass within a predefined bandwidth referred to as the "pass band".

The following image illustrates a typical band-pass filter and provides a useful visual reference:

Typical Band Pass Filter

In the above image, the filter response is drawn with a red line, which visually has the appearance of a mountain with a flat top. 

The flat-top area is technically the filter's pass band and the closer this is to the 0dB line, the better the filter will perform. Any gap that is present between the flat top and the 0dB line is deemed as loss or attenuation and something that needs to be kept to a minimum.

In the illustration image, the pass band is defined by a blue line. Notice that it extends beyond the flat top of the mountain and encroaches onto the mountain sides. This is called the -3dB point or the frequency at which the measured power level is 3dB below the maximum value, it is a common standard used for RF measurements.

It should be noted that ripple can occur on the mountain top of a poorly designed filter resulting in unwanted distortion and losses so it is important to keep ripple to an absolute minimum.

The filter stop band is the frequency range measured from the -3dB point down to the -60dB point or base of our mountain. In an ideal world, the mountain slopes would be a sheer drop or 90 degrees to the flat top, resulting in a "brick wall" filter. Fundamentally, this would only allow signals through in the wanted pass band. 

Signals that fall within the stop band may be deemed as unwanted neighbours, they do not fall within the wanted pass band and only the brick wall will keep them out! Unfortunately, this is not practically achievable in a band-pass filter so a compromise must be made to keep the stop band narrow and the neighbours at bay. We do this by adding more poles (stages) to the filter, which, in turn makes the mountain slope steeper. As a consequence the build size of the filter increases and the RF losses become greater.

It should be noted that roofing and intermediate stage filters in a receiveralso provide protectionagainst keeping the neighbours out hi hi!

To conclude, I hope my basic and simplistic approach helps those with lesser knowledge gain a better understanding of the principal operation of the BPF. It is worth researching the different types of filters in order to gain greater knowledge but that is beyond the scope of my blog or intentions.

So with the basic principles and design criteria out of the way, it is time to develop a set of filters for my Irwell HF Transceiver project.

I am using a free software application called RFSim99 to model the filters. It is an old but very functional piece of software that is ideal for simulating RF circuits such as filters. 

The software ceased working on Microsoft operating systems after Windows XP but it will run fine on a virtual PC emulator such as the free Oracle VirtualBox software.  

Alternatively fellow ham AD5GG has made available a version of RFSim99 that will run on Windows 7,8 & 10 and it can be downloaded from the following web link:

There is a known quirk when running the version from AD5GG. Sometimes some of the menu buttons fail to appear but you can get around this issue by hovering your mouse cursor over the buttons, which makes them reappear and functional.

I have opted for 5 pole filters and will be creating 10 in total. Once the filters were simulated in RFSim99 and optimized I built them manhatten style on a piece of scrap copper clad board for testing.

Here is an image of my test PCB ready to be populated with components:

And here is the same PCB populated with components to form a filter covering 3.0MHz to 4MHz:

 This is the filter modelled in RFSim99:

Click image to enlarge to full size!

And here is a plot of the filter hooked up to my Spectrum Analyser:

Click image to enlarge to full size!

You can see from the image captured on my spectrum analyser that the filter is excellent. The mountain top is flat and slightly below the 0dB line indicating negligible insertion loss. The pass band is within the desired frequency range needed and at 10MHz the  signal is down to almost -65dB.

There is a lot of work to do on the remaining filters and altogether I will have to wind 50 toroid cores, which are all of the T50 variety. I think it's fair to say that my fingers are going to be somewhat sore after winding those but as they say no gain without pain!

I will provide an update on the filters soon and take on the challenge of producing a suitable PCB for the project.

Until next time...

73's From Andy G6LBQ

Its all about the Radio Ga Ga...

 

 For part 2 of the BPF module click here.