# Tuning a SAW filter?

Does anyone know if it's possible to shift the center frequency of a SAW filter? I'm only talking about 0.1% shift, say from 433MHz to 437MHz. Can it be done or is the center frequency baked into the design during manufacture?

• Quartz crystal can increase or decrease freq, by small amount only (not a few MHz), by soaking in chemical solutions that deposit or etch the quartz mass. Wonder if similar apply to SAW as both based on some form of 'mechanical vibration'. This patent paper describes tuning by mechanical deformation. Not sure if used in practice. – EEd Aug 18 '16 at 12:38
• @EEd, why not post that as an answer? – user157 Aug 18 '16 at 15:21
• I did actual work on quartz but not SAW. So, not 100% sure how a 'related process' can be practiced in SAW filter. – EEd Aug 18 '16 at 23:01
• I was hoping I could pull the frequency by a few tenths of a percent by adjusting the matching network or some such, a full chemical etch sounds like it'd require quite a bit of prep work and some... unpleasant compounds – Sam Aug 18 '16 at 23:12
• @Sam what is the actual reason you want to tune the SAW? – Marcus Müller Aug 19 '16 at 20:22

As ~430 MHz SAW filters generally have a 3dB bandwidth of 0.2–2 % $f_\text{center}$, tuning them by some 0.1% does seem like a relatively big change.

You could try to build a small "oven" for your filter to operate in, e.g. by placing a resistor atop of your filter, using that to heat up the filter, and encasing everything in spray foam to make the thing thermically more stable. However, I doubt that will let you get the adjustments you want.

Frankly, when tuning an SAW, you're probably adjusting the wrong part: SAW filters are used because they are stable. The most common usage in technology is as IF (intermediate frequency) filter. Now, IF signals get generated by converting an RF signal to that IF; you do that by mixing the RF with a local oscillator at the difference frequency $f_{LO}=f_{RF}-f_{IF}$.

Typically, these oscillators are adjustable to allow for tuning a receiver to different frequencies – the fact that the SAW stays fixed is an advantage here, not a problem, because you'd normally just tune the LO!

Adjusting oscillators is also pretty easy; there's a lot of types of oscillators, but let's identify the more relevant ones here:

• L/C oscillators, Colpitts etc.: You can vary a capacitance or an inductivity; that's most often done mechanically, eg. by turning a screw
• Adjustable crystal oscillators: These come with an "adjust voltage" pin; you change the voltage applied to that pin, and that changes the frequency. This is often done with control loops, which are used to compare the frequency that the oscillator generates to a frequency normal. For example, assume you have a 900 MHz oscillator that has a factory accurace of 50ppm; it might be $\pm$45 Hz off. But you have a known-to-be-very-good 10 MHz oscillator. You can simply count how many "900 + error" MHz oscillations you see while the 10 MHz oscillator does one – that would be a very simple digital PLL.
• Integer-N/Fractional-Q synthesizers: In essence: you do the above, but the signal you're generating is derived by frequency multiplying your input frequency normal with an adjustable factor, and then using a bit of algebra to derive a wide range of possible tones on demand. Note that these things aren't continously adjustable, they can just give you fine-grained grid of frequencies. This grid is typically finer than the bandwidth of a 400 MHz SAW filter, so that's not really a problem – "we're off by around 200 kHz" translates to "ok, so let's change the the LO frequency by something in the vicinity of 200 kHz; not like a 30 Hz offset would make any difference to a filter of 1MHz bandwidth"
• Yeah, I was afraid of that. I hadn't thought of ovenizing it though, I don't know what the thermal expansion of a SAW filter is but it's something worth trying – Sam Aug 19 '16 at 22:29