FCom Fuji Crystal · Technical Guide
SAW bandpass filters provide compact frequency selection in RF and intermediate-frequency signal paths. Their usefulness depends on the required passband, nearby interference and available loss budget. Comparing them with LC, ceramic, BAW, cavity and digital filters requires more than choosing the smallest package or the steepest response.
Core answer: Consider SAW for compact, fixed-band analog filtering. Compare other technologies when tuning range, power handling, insertion loss or operating frequency becomes the main constraint. Digital filtering complements the analog front end, but cannot recover information already lost through clipping or in-band aliasing.
How SAW bandpass filters work
A surface acoustic wave filter uses electrodes on a piezoelectric substrate to convert electrical signals into acoustic motion and back. Its geometry and resonant or transducer structures establish the frequency response. Acoustic wavelengths enable compact filtering structures. BAW devices instead use acoustic vibration within the material’s thickness. [1]
A bandpass filter attenuates frequencies outside its intended band while passing the wanted signal with finite loss. It does not amplify that signal, and its response must be evaluated with the specified source, load and matching network.
SAW compared with other filter technologies
| Technology | Useful strengths | Key checks |
|---|---|---|
| SAW | Compact fixed-band RF/IF selectivity. | Matching, temperature shift, insertion loss and permitted input power. |
| BAW | Compact acoustic filtering; often attractive at higher RF frequencies. | Band-specific loss, rejection, thermal behavior and power rating. |
| Lumped LC | Flexible circuit values and response choices. | Component Q, tolerances, parasitics and board implementation. |
| Ceramic resonator | High-Q dielectric resonators can combine low loss with strong selectivity. | Size, bandwidth and temperature requirements. |
| Cavity | High unloaded Q and options for demanding power levels. | Mechanical volume, weight, integration and cost. |
| Digital FIR/IIR | Programmable response after sampling. | ADC headroom, aliasing, processing resources and latency. |
Here, “ceramic resonator” refers to dielectric-resonator RF filters. A ceramic package alone does not identify the filtering mechanism. LC, ceramic and cavity solutions occupy different design regions; ceramic filters should not be dismissed as universally higher-loss or less selective than SAW. [2]
Temperature behavior also depends on implementation. Conventional SAW, temperature-compensated SAW and BAW do not share one drift specification. Compare the complete response over the required temperature range, particularly near passband edges. [1]
Compare specifications under equivalent conditions
- Passband and bandwidth: Define both band edges and the allowed attenuation. A 1 dB bandwidth and a 3 dB bandwidth are different criteria; also check whether the reference is absolute loss or loss relative to the passband minimum.
- Insertion loss and ripple: Compare maximum loss across the useful band and variation within that band, rather than center-frequency loss alone.
- Stopband rejection: Specify attenuation at actual blocker frequencies or over defined ranges. One rejection number does not describe the entire stopband.
- Impedance: Compare devices using their stated terminations and matching conditions. Mismatch affects measured transmission.
- Group delay: Check its variation across the occupied signal bandwidth when waveform fidelity matters. A steep amplitude response does not guarantee flat delay. [3]
Power and temperature limits need separate review. Ask whether input-power ratings apply to continuous waves or pulses, and whether specified performance holds throughout the operating range. Do not interpret an absolute maximum rating as a recommended operating point.
Illustrative bandwidth calculation: A 20 MHz bandwidth at a 1,000 MHz center frequency gives a fractional bandwidth of 20 ÷ 1,000 × 100% = 2%. This describes relative bandwidth, not insertion loss, rejection or a particular FCom product.
Where digital and MEMS filtering fit
Digital filtering complements analog selection
Digital filters operate on sampled data. They can implement channel selection and change response through coefficients, but unwanted frequencies that have already aliased onto the desired band cannot generally be separated afterward. Analog filtering before sampling remains necessary where alias protection is required. [4]
Likewise, post-conversion processing cannot prevent ADC overload. Analog and digital stages should share the filtering task according to signal level, sampling rate and available processing capacity. Both practical analog filters and digital filters introduce delay; assess settling time and group-delay behavior rather than describing SAW as instantaneous. [5]
MEMS requires a specific architecture comparison
MEMS is not a single filter response or performance grade. For example, RF MEMS switches can select branches in a reconfigurable filter network. Analog Devices documents this arrangement in the ADGM1004 datasheet. Compare the whole network’s loss, isolation, switching time, control requirements and operating lifetime with a fixed SAW solution. The switch alone is not the bandpass filter. [6]
Choose by application and validate on the board
For a fixed-band receiver, begin with the wanted signal and strongest nearby interferers. For a tunable instrument, consider the required frequency coverage and switching strategy. For a transmitter, include power, mismatch and thermal conditions from the start.
- Write the frequency mask. State passband edges, allowable ripple, maximum loss and rejection at each interfering band.
- Define the interface. Record source/load impedance, balanced or unbalanced ports and the available matching components.
- Compare complete implementations. Include filter, matching network, switching, control circuitry and occupied PCB area.
- Measure candidates. Check transmission and return loss on the intended layout across temperature, then verify receiver sensitivity and blocker performance or the relevant transmitter requirements.
FCom’s FSF-5050 SAW filter family includes fixed-frequency options across 76.5–1000 MHz in a 5.0 × 5.0 mm SMD package. This is family coverage, not the tuning range of one device. Select an individual variant by its passband, loss and stopband specification.
Frequently asked questions
Is SAW always better than LC?
No. SAW can suit compact fixed-band designs; LC allows flexible circuit implementation. Compare actual bandwidth, loss, rejection and board constraints.
Can software replace an RF bandpass filter?
Software can perform filtering after conversion. It cannot protect earlier analog stages from overload or undo information loss caused by aliasing.
Do all SAW filters have 1–3 dB insertion loss?
No. There is no universal loss range. Use the selected part’s guaranteed limits and specified matching conditions.
Does the temperature rating guarantee an unchanged passband?
No. Operating temperature and frequency-response drift are separate considerations. Verify that the passband and rejection mask remain acceptable throughout the required range.
Discuss your filter requirements with FCom
Explore the FCom SAW filter range, or send your frequency mask, loss budget, impedance, input power, temperature range and package limits to FCom@fujicrystal.com. These requirements provide the basis for selecting and evaluating a suitable candidate.
Technical references
- Qorvo: BAW vs. SAW RF Filters
- Mini-Circuits: Classical RF Filters
- Mini-Circuits: Terms and Definitions—Filters
- Analog Devices: Anti-Aliasing Basics Using Switch-Capacitor Filters
- Analog Devices: Practical Filter Design Challenges and Considerations for Precision ADCs
- Analog Devices: ADGM1004 Datasheet—Reconfigurable RF Filter Application
