General > General Technical Chat

How do I build a tuneable band-pass filter ?

(1/2) > >>

jeremy:
Hey all,

As of late, I have really wanted to get into RF. I have my microcontroller stuff down, but I really don't like that most of this RF stuff seems like magic to me. Thus, my new project is to build a lowish frequency spectrum analyser (say 500kHz - 50MHz) and I have been reading up on how spectrum analysers work. I am a little stuck on a few things so I am hoping I could get some advice from the RF magicians who visit this forum.

I have been looking through old service manuals and it is clear that YIGs are the way it is normally done. I don't want to use a YIG filter/oscillator for three reasons:

* they are expensive
* low frequency ones are hard to find
* I don't want to worry about about high frequency microwave coupling and noise; YIGs are intended for multi-gigahertz filters/oscillators
I have a few ideas how to build a VCO that can do what I want (although suggestions would be more than welcome!), but the thing I really have no idea about is how to build a electronically tuneable bandpass filter that can work over such a large range. If anyone could give me any ideas of things to research and/or proven designs I would greatly appreciate it!

One last thing to keep in mind: I am still a university student so I don't have a great deal of money to spare. But on the upside, I may be able to sneak into a lab and spend a few quality minutes with a VNA if I can build something to test!

joelby:
Spectrum analysers are usually superheterodyne radios with fixed filters. Different bandwidths are achieved by switching between different band pass filters, often with relays. A YIG oscillator is essentially a VCO with a very wide tuning range, not a filter.

There are a few spectrum analyser projects floating around that might give you some ideas. The most comprehensive is probably the Scotty Spectrum Analyzer. Here's a much simpler one that has a nice block diagram and explanation.

ejeffrey:
What kind of design are you looking at?  If you build a standard double-conversion spectrum analyzer you shouldn't need a tunable filter.  Your signal chain looks like:

Low-pass 50 MHz -> mixer -> high IF filter -> mixer  -> low IF filter -> detector -> video filter

The first mixer is where your swept VCO goes, and upconverts your signal to an IF that is above the maximum input frequency.  This provides the image frequency rejection.  The second mixer uses a fixed frequency LO to downconvert to the low IF for your RBW filter, then comes your power detector and video filter.

Search google for scotty spectrum analyzer which is a home-brew 1 GHz spectrum analyzer.

jeremy:
Thanks guys, that Scotty one looks pretty good. I have been looking at the design for the hp 8563e mostly, service manual is here. The functional diagram is on page 347, and on page 351 it has the following:


--- Quote ---A10 YIG-Tuned Filter/Mixer
The YIG-tuned filter/mixer (RYTHM) is a combination of an RF switch, a high band mixer, and a tracking preselector. The PIN diode switch directs the RF input to the appropriate mixer in the A10 RYTHM assembly or A8 low band mixer.

The tracking preselector is a YIG-tuned filter. It functions as a tunable bandpass filter for high band signals. Coarse frequency control originates from slope and offset DACs located on the A14 frequency control assembly. (Slope and offset DAC values are loaded into EEROM.)
Fine frequency control originates from a preselector peak DAC located on the A3 interface assembly. Values for the preselector peak DAC are interpolated approximately every 17 MHz based upon data taken during the frequency response (flatness) adjustment. The preselector's bandwidth varies from greater than 30 MHz, at 2.75 GHz, to greater than 60 MHz, at 26.5 GHz.
--- End quote ---

Why is there a tuneable bandpass in this design? Perhaps I am misunderstanding its purpose.

As an aside, is there actually a way to build a tuneable bandpass over this range ?

amspire:
What you have is a fixed bandpass filter. Because if is fixed, you can design and tune it to have exactly the shape you want - something that is extremely hard to do with tuneable filters.

You then mix the incoming signal with a low distortion variable frequency sinewave to transform the frequency of the incoming signal to the filter's frequency. Often two conversions are needed if you want to tune  from low frequencies to high frequencies. If you were just looking at one band, you could get away with one mixer.

In terms of doing it cheaply, the cheapest way I can think of is to get a TV tuner module that will probably be able to tune from less than 100MHz to over 800MHz. You mix the incoming signal with a fixed frequency sinewave at, say, 700MHz and then use the tuner to tune between 700 and 800MHz to get a near DC to 100MHz output. There are sources of affordable mixing modules such Mini-Circuits:

http://www.minicircuits.com/products/Mixers_what_is_new.shtml

The TV tuner will have an output bandwidth of probably a few MHz, so add an extra sharp tuner on the output. An AM band IF tuner would reduce your bandwidth to something like 20KHz.

You can use junk parts from TV's and VCR's but finding specs can be hard. There are a number of projects using selected tuners on the internet, or you can just chase after new tuner modules that do have good specs.

Richard.

Navigation

[0] Message Index

[#] Next page

There was an error while thanking
Thanking...
Go to full version
Powered by SMFPacks Advanced Attachments Uploader Mod