Antennas · Volume 30
Multi-Radio Shared Antennas — Diplexers, Multiplexers & Band-Switching
One antenna for many radios — the three sharing strategies, the GPS exception, the TX-RX problem, and the uConsole 7-antenna case walked through end-to-end
Contents
1. About this volume
The use-case matrix in Vol 29 answered the per-radio antenna question: for each radio in the Hack Tools hub, here’s the right antenna. This volume answers the inverse question: for a single platform that hosts multiple radios across multiple bands, can the radios share antennas — and at what cost?
The motivating case is tjscientist’s fully-equipped Clockwork uConsole, which in its current build carries seven separate antennas — one each for cellular, 2.4 GHz Wi-Fi, 5 GHz Wi-Fi, Bluetooth, GPS, sub-GHz LoRa, and a wideband SDR feed. Seven antennas on a handheld is mechanically awkward, visually noisy, and operationally fragile (every connector is a failure point; every cable run is a potential common-mode current source). The same question applies to every multi-radio platform in the hub — the Banshee with its ESP32-C5 + ESP32-S3 + NRF24 + GPS + Ethernet stack, the Nyan Box with three NRF24L01+ radios plus an ESP32, any future build that hosts several radios simultaneously.
The chapter walks through the three structural strategies for sharing antennas (wideband / multiplexing / switching), names the parts you’d buy or build for each, exposes the efficiency tax that every sharing approach extracts, and closes with a concrete worked example: collapse the uConsole from seven antennas down to three (or two if you’re willing to compromise on GPS performance) using parts you can solder yourself in an afternoon or order from Mini-Circuits for under $200.
Where this volume sits in the series:
- Vol 18 (passive splitters) covers equal-power splitters and combiners; this volume covers frequency-selective combiners, a related but distinct topology.
- Vol 19 (active splitters & preamps) covers amplified distribution; sharing antennas usually doesn’t need amplification, but the masthead-LNA pattern from Vol 19 shows up when sharing a low-gain antenna among multiple receivers.
- Vol 29 (use-case matrix) gives the per-radio antenna recommendation; this volume is the “what if I want to use ONE antenna across all of them” follow-up.
- Vol 32 (antenna farms) covers the inverse problem: one radio, many antennas (stacking, phasing, switching matrices for contest stations). The two volumes share several switching-matrix parts and band-decoder topologies but solve opposite problems.
2. The problem — N radios, M bands, finite case real estate
A multi-radio platform faces three intertwined constraints:
- Each radio wants its own optimally-tuned antenna. A 2.4 GHz Wi-Fi radio works best with a 2.4 GHz half-wave antenna; a 5 GHz radio with a 5 GHz antenna; a cellular modem with a multi-band cellular antenna optimized for B1/B3/B7/B20 etc. Polarization, gain, pattern — each band has its own optimum.
- Physical space is limited. A handheld case has perimeter and a few dozen square centimetres of board real estate. Seven SMA / U.FL connectors plus seven antenna structures plus seven cables eats the entire mechanical budget.
- Each antenna is a regulatory and EMI exposure. Every transmit antenna needs its own FCC compliance review (Part 15 / Part 22 / Part 90 / Part 97 by band, Vol 31). Every antenna is a potential receiver of unintended signals — co-location of multiple TX/RX systems creates intermodulation products that don’t exist when each system is on its own platform.
The three constraints push in different directions. Constraint 1 says “give every radio its own antenna.” Constraints 2 and 3 say “the fewer antennas, the better.” Sharing is the compromise. The engineering question is: what shape of compromise costs the least in performance?
The answer depends on the band overlap of the radios. Three cases:
| Case | Example | Sharing strategy |
|---|---|---|
| All bands close together | 2.4 GHz Wi-Fi + Bluetooth + NRF24 + ZigBee — all in 2.400–2.485 GHz | A single 2.4 GHz omni or patch serves all four via a power splitter, almost no loss |
| Bands clearly separated | 433 MHz LoRa + 2.4 GHz Wi-Fi + 5 GHz Wi-Fi — three different octaves | A diplexer/triplexer (frequency multiplexer) lets one antenna feed three separate radios with > 30 dB port-to-port isolation |
| Bands wide and arbitrary | HF (3-30 MHz) SDR + VHF/UHF scanner + 2.4 GHz Wi-Fi + cellular + GPS | A wideband antenna (discone, Vol 12) or a band-switched matrix; sharing is more lossy |
The uConsole’s radio set spans all three cases simultaneously — which is why the consolidation is non-trivial.
3. The three sharing strategies — wideband, multiplexing, switching
The three structural ways to share an antenna among multiple radios:
THREE WAYS TO SHARE AN ANTENNA
Strategy 1: Wideband antenna
╔═════════╗
║Wideband ║
║antenna ║
║(discone)║
╚════╤════╝
│
▼
┌────────────┐
│ Pwr split │ Splits each radio gets a
│ (or just │ reduced-power copy of
│ T-joint) │ the same wideband signal
└─┬──┬──┬──┬─┘
│ │ │ │
R1 R2 R3 R4 each gets a 6 dB lossy split
Strategy 2: Frequency multiplexer (diplexer)
╔═════════╗
║Wideband ║
║antenna ║
╚════╤════╝
│
▼
┌─────────────┐
│ Diplexer │ Routes f<F to one port,
│ (low/high │ f>F to another, with
│ crossover) │ > 30 dB port-to-port
└──┬───────┬──┘ isolation
│ │
R1 (LF) R2 (HF) each radio sees its
full band, no split
loss in-band
Strategy 3: Band-switching matrix
╔═════════╗
║Wideband ║
║antenna ║
╚════╤════╝
│
▼
┌──────────────┐
│ SP4T switch │ Only one radio
│ + band │ connected at a time;
│ decoder │ the others are
└─┬──┬──┬──┬───┘ electrically isolated
│ │ │ │
R1 R2 R3 R4
The three are not mutually exclusive — a real multi-radio platform usually combines all three.
Strategy 1 (wideband + splitter) is the simplest. Loss is high (3 dB per 2-way split, 6 dB per 4-way), but the implementation is a few resistors or a Wilkinson combiner (Vol 18). Works for receive-only radios sharing a wideband antenna.
Strategy 2 (multiplexer) has the lowest in-band loss (~0.5-1 dB per port) but requires a frequency multiplexer matched to the specific bands. Off-the-shelf diplexers exist for common splits (low-HF/VHF, 2.4/5 GHz). Custom multiplexers are a PCB and a few hours.
Strategy 3 (switching) has the lowest loss when only one radio is active at a time (0.3-0.6 dB through a single switch element). The disadvantage: radios can’t transmit/receive simultaneously through the shared antenna. Requires a band-decoder or external control.
4. Strategy 1 — wideband antennas as multi-radio servers
A discone (Vol 12) or LPDA (Vol 13) covers many bands with a single antenna. Pair it with a passive splitter (Vol 18) and every radio gets the same signal, just lower in power.
Where this works:
- Multiple SDR receivers from one antenna: HackRF + RTL-SDR + scanner all sharing one discone. Receivers don’t care about a few dB of split loss since their noise floor is set by the antenna’s pickup, not the receiver’s NF.
- Multiple 2.4 GHz radios from one omni: 2.4 GHz Wi-Fi + BLE + NRF24 from a single 2.4 GHz omni via a 3-way Wilkinson. All radios are in the same 80 MHz band; loss budget is uniform per port.
- Receive-only EMI / spectrum-monitoring stations: one wideband antenna feeding multiple analyzers and SDRs.
Where this fails:
- Mixed TX + RX with co-location: a high-power TX leaking back through the splitter into a sensitive RX is a system-killer. Use Strategy 2 or 3 instead.
- Bands with strongly different antenna requirements (HF wire + 5 GHz patch): one wideband antenna won’t match both well; the compromise is too lossy on one or both ends.
Cross-link: Vol 19 describes the active distribution amplifier workaround that compensates for split loss when sharing one antenna across multiple receivers. A masthead-LNA + 4-way active splitter recovers the 6 dB of split loss and contributes a low-NF first stage to every downstream receiver.
5. Strategy 2 — frequency multiplexing: diplexers, triplexers, N-plexers
A diplexer is a 3-port device that splits one antenna port into two band-selective output ports. Signals below a crossover frequency F_c route to one port; signals above F_c route to the other. Port-to-port isolation is typically 30-60 dB at the design frequencies.
DIPLEXER FUNCTIONAL DIAGRAM
Antenna port (wideband)
│
▼
┌───────────────┐
│ Common port │
│ │ │
│ ┌─────┴─────┐ │
│ │ │ │
│ ▼ ▼ │
│┌──┐ ┌──┐│
││LP│ │HP││ L = low-pass filter
│└┬─┘ └─┬┘│ H = high-pass filter
│ │ │ │ C = crossover ~F_c
│ ▼ ▼ │
│L port H port
│f < F_c f > F_c
└───────────────┘
A diplexer is essentially two complementary filters sharing a common port. Each output port sees only the band assigned to it. The low-port and high-port radios can both transmit and receive simultaneously through the shared antenna without interfering with each other (modulo isolation limits).
5.1 Diplexer specifications you’ll see
A typical commercial diplexer datasheet:
| Spec | Typical value |
|---|---|
| Crossover frequency F_c | varies — e.g. 1 GHz for cellular/Wi-Fi split |
| Insertion loss in pass-band | 0.4-1.5 dB |
| Isolation port-to-port | 30-60 dB |
| Return loss at each port | 15-20 dB (SWR < 1.5:1) |
| Power handling | 100 mW to 500 W depending on design |
| Bandwidth per port | a few percent to multiple octaves |
The crossover transition is sharp — typically a few percent of F_c wide. Inside the transition band, both ports leak; outside, isolation is full-spec.
5.2 Triplexers and N-plexers
A triplexer is a 4-port device (common + 3 outputs) with two crossover frequencies. A quadplexer has three crossovers. In general an N-plexer can split into N bands.
Common commercial N-plexer crossover patterns:
- Cellular triplexer: 600-1000 / 1700-2200 / 2300-2700 MHz (covers most LTE bands grouped by sub-1G / 1.7-2.2G / 2.3-2.7G)
- Wi-Fi diplexer: 2.4-2.5 / 5.1-5.9 GHz
- Ham-radio triplexer: HF 1.8-30 / VHF 50-150 / UHF 400-450 MHz (Comet CFX-431, Diamond MX-72D-equivalent)
- GSM/UMTS/LTE diplexer: 0.7-1 / 1.7-2.7 GHz
If your bands don’t match a stock product, a custom diplexer is a PCB + a few SMD components (see §12 for the build).
5.3 Multiplexer topologies
Three common internal topologies:
- Lumped-element LC (low-frequency): discrete inductors and capacitors forming the two crossover filters. Used below
3 GHz where lumped components have manageable Q. Cheap ($5-50 commercial). - Distributed (microstrip stub): quarter-wave stubs on a microstrip PCB. Used 1-30 GHz. More expensive but cleaner roll-off at GHz frequencies.
- Cavity filter (very high power): machined cavities in aluminum or copper. Used in broadcast / cellular base-stations at multi-kW power levels. Out of amateur envelope.
For Hack Tools applications (≤ 1 W typical), lumped-LC up to 2 GHz and microstrip 2-6 GHz cover the entire envelope.
6. Strategy 3 — time multiplexing: band-switching matrices
A band-switching matrix uses an RF switch to route the antenna to exactly one radio at a time. Only the active radio sees the antenna; the others are electrically isolated. The switch is controlled by either a band-decoder (which reads the transmitting radio’s band output and selects the right port) or by direct GPIO control from the host system.
BAND-SWITCHING MATRIX
Antenna (wideband)
│
▼
┌───────────────┐
│ SP4T RF │ Single-pole, 4-throw
│ switch │ (SP4T) or SPNT switch
│ (PIN diode │ PIN-diode, MEMS, or
│ or MEMS) │ mechanical relay
└─┬──┬──┬──┬────┘
│ │ │ │
▼ ▼ ▼ ▼
R1 R2 R3 R4
Control:
┌─────────────┐
│ Band │ Reads current TX band
│ decoder │ from rig (BCD output)
│ or │ or from host MCU
│ MCU GPIO │
└─────────────┘
Trade-offs vs Strategy 2 (multiplexer):
| Property | Multiplexer (Strategy 2) | Switch (Strategy 3) |
|---|---|---|
| Simultaneous TX/RX on different radios | YES | NO (one at a time) |
| In-band loss | 0.4-1.5 dB | 0.3-0.6 dB |
| Out-of-band rejection | 30-60 dB (filter shape) | > 50 dB (full isolation) |
| Power handling | 1 W to 500 W | depends on switch type |
| Cost | $5-100 (off the shelf) | $20-200 (with control) |
| Control complexity | Zero (passive) | Band-decoder or GPIO |
The switch wins on in-band loss and isolation but loses on simultaneity. For a uConsole-class platform where most radios are receive-mostly and TX is intermittent, the switch is often the better choice. For a platform with simultaneous TX/RX (a cellular phone or simultaneous Wi-Fi + BLE), the multiplexer is mandatory.
7. The TX-RX problem — protecting receivers from co-located transmit
When multiple radios share an antenna, a transmitter on one radio can fry the front-end of a sensitive receiver on another radio. This is the dominant failure mode of sharing.
The math: a +30 dBm (1 W) Wi-Fi TX inside a 30 dB-isolation diplexer leaks +0 dBm (1 mW) into the adjacent RX port. A typical SDR or LNA front-end is rated for at most +10-15 dBm input before damage; +0 dBm is survivable but compresses heavily. A +37 dBm (5 W) UV-K5 TX with the same 30 dB isolation leaks +7 dBm — already in the compression zone.
Two cures:
- More isolation — use a multiplexer with 50-60 dB isolation, or stack a band-pass filter at each receive port for an additional 20-30 dB rejection at out-of-band frequencies.
- T/R sequencing — disconnect or detune the receive radio while the transmit radio is active. Mandatory above ~10 W TX power even with high-isolation multiplexers.
For the uConsole case (all radios are ≤ +20 dBm, most are receive-mostly), 30-40 dB isolation is sufficient and T/R sequencing isn’t needed. For higher-power platforms (UV-K5 at 5 W, or any HF SDR with an external amplifier), sequenced T/R is mandatory.
Cross-link to Vol 19 §9 for the T/R relay design used to protect masthead LNAs.
8. GPS — why it’s almost never sharable
GPS is the antenna that resists sharing more than any other in the Hack Tools envelope. Three reasons:
- Polarization: GPS L1 is right-hand circular (RHCP). Sharing with a linearly-polarized antenna costs 3 dB of polarization mismatch, and loses GPS multipath rejection (GPS receivers expect RHCP and use the LHCP rejection to suppress ground-bounce reflections — see Vol 2 §5.2).
- Integrated LNA: most GPS antennas are active — built-in LNA in the antenna unit, powered via bias-T over the coax. Sharing the antenna means sharing the bias-T DC, which conflicts with other radios that don’t expect DC on the coax.
- Sensitivity floor: GPS receivers operate at -127 dBm — among the most sensitive in the bench envelope. Any sharing loss (1-2 dB through a multiplexer, 3 dB through a polarization mismatch) directly degrades acquisition time and weak-signal performance.
Recommendation: keep GPS on its own dedicated antenna. The size cost (a small ceramic patch or active “puck” antenna) is small; the performance benefit is large. In the uConsole consolidation below, GPS is the one antenna that doesn’t get combined.
The exception is if the platform’s GPS-equivalent radio is non-critical — e.g., a NEO-M9N GPS module with no time-sensitive duty. Then sharing a low-band antenna (the GPS L1 at 1.575 GHz can route through a cellular triplexer’s mid-band port at modest loss) becomes a tolerable compromise. But for any GPS-dependent timing (NTP via GPS-disciplined oscillator, position-aware logging, time-stamped captures), keep GPS dedicated.
9. Worked example — consolidating the uConsole from 7 antennas to 3
The Clockwork uConsole in tjscientist’s fully-loaded build hosts seven radios across the following bands (typical layout — exact set varies by which expansion bays are populated):
| # | Radio | Band(s) | TX power | RX sensitivity | Polarization |
|---|---|---|---|---|---|
| 1 | LTE/5G cellular modem | 600-3500 MHz multiband | +23 dBm | -110 dBm | linear V |
| 2 | Wi-Fi 2.4 GHz (CM4 onboard) | 2400-2484 MHz | +20 dBm | -90 dBm | linear V |
| 3 | Wi-Fi 5 GHz (CM4 onboard) | 5180-5825 MHz | +20 dBm | -85 dBm | linear V |
| 4 | Bluetooth 5 (CM4 onboard) | 2400-2484 MHz (shares Wi-Fi 2.4 silicon) | +4 dBm | -95 dBm | linear V |
| 5 | GPS L1 receiver | 1575.42 MHz | RX-only | -127 dBm | RHCP |
| 6 | LoRa sub-GHz radio | 868/915 MHz (region-dependent) | +20 dBm | -130 dBm | linear V |
| 7 | Wideband SDR (HackRF / RTL-SDR via USB) | 1 MHz – 6 GHz | varies | varies | depends on antenna |
Now the consolidation question: how do we collapse this from 7 antennas to fewer, and what does it cost?
9.1 First-pass consolidation — 7 → 5 antennas
Two easy wins immediately:
- Wi-Fi 2.4 + Bluetooth are already on the same CM4 silicon and share the same antenna by design (CM4’s single 2.4 GHz front-end serves both). Most builds put Wi-Fi 5 on a second CM4 antenna port. So 7 antennas (cellular + Wi-Fi 2.4 + Wi-Fi 5 + BT + GPS + LoRa + SDR) is really 6 antennas (cellular + combined 2.4 + Wi-Fi 5 + GPS + LoRa + SDR) if you accept the standard CM4 dual-antenna config.
- LoRa 868/915 MHz + SDR sub-GHz can share via a sub-GHz omni antenna (vertical, e.g. an RTL-SDR-Blog-style telescoping whip resonant at 900 MHz) + a 2-way splitter (Vol 18). 3 dB split loss; the LoRa link budget tolerates it easily (LoRa’s chip-rate gain handles −10 dB SNR comfortably), and the SDR receive-only role doesn’t care.
Now: 5 antennas — cellular + (Wi-Fi 2.4 + BT) + Wi-Fi 5 + GPS + (LoRa + SDR sub-GHz).
9.2 Second-pass consolidation — 5 → 3 antennas via diplexer/triplexer
The next stage uses frequency multiplexing:
- Cellular (600-3500 MHz) + Wi-Fi 5 GHz (5180-5825 MHz) can share a wideband antenna via a diplexer with crossover at ~4 GHz. The cellular port handles 0.6-4 GHz; the Wi-Fi 5 port handles 4-6 GHz. Cellular modems already include their own front-end filtering, so a 1-2 dB diplexer loss on the cellular path is acceptable. Wi-Fi 5 sees similar in-band loss. Antenna: a wideband 0.6-6 GHz omni (e.g. Taoglas Apex II), or a discone for receive-only context.
- Wi-Fi 2.4 (incl. BT) + LoRa + SDR sub-GHz could share through a triplexer with crossovers at ~1 GHz and ~2 GHz. The 2.4 GHz port handles Wi-Fi/BT; the ~1-2 GHz mid-port is empty (or hosts a future radio); the sub-GHz port handles LoRa + SDR via the splitter from §9.1. Antenna: a multi-band stub or a discone covering 800 MHz - 3 GHz.
- GPS stays separate for the polarization + sensitivity reasons in §8.
Now: 3 antennas — wideband-omni-for-cellular-plus-WiFi5 + sub-GHz-and-2.4-omni-with-triplexer + dedicated-GPS.
9.3 Third-pass consolidation — 3 → 2 antennas (aggressive)
If you’re willing to compromise GPS performance — accepting 1-2 dB cellular-band loss in exchange for one less antenna on the case — you can route GPS L1 (1.575 GHz) through the cellular triplexer’s mid-band port. The cellular antenna’s pattern at 1.5 GHz is roughly omnidirectional, and the GPS LNA inside the GPS module has enough gain (typically 25-30 dB) to overcome the diplexer’s 1-2 dB insertion loss.
This is plausible for non-critical GPS use cases (logging, casual navigation). For precision timing or weak-signal acquisition, keep GPS separate.
So: 2 antennas — wideband-cellular-with-GPS-merged + sub-GHz-2.4-triplex. Aggressive but viable.
9.4 The consolidation summary
| Build | # antennas | Top losses incurred | Suitable for |
|---|---|---|---|
| Stock (no consolidation) | 7 | None | Reference / no-compromise |
| Step 1 — share BT with Wi-Fi 2.4, combine LoRa+SDR | 5 | 3 dB on LoRa, 3 dB on SDR receive | Most pen-test workloads |
| Step 2 — diplex cellular + Wi-Fi 5, triplex sub-GHz | 3 | 1-2 dB cellular, 1-2 dB Wi-Fi 5 | Practical handheld use |
| Step 3 — merge GPS into cellular antenna | 2 | 1-2 dB GPS, slight pol-mismatch | Logging-only GPS use |
The right stopping point depends on the use case. For pen-test field work, Step 2 (3 antennas) is the sweet spot — significant case-real-estate recovery with manageable performance loss. For minimal-antennas-possible builds (drone, robot, embedded), Step 3 (2 antennas) is the engineering target.
9.5 Bill of materials for Step 2 (3-antenna consolidation)
Approximate components for the 3-antenna consolidation:
- 1× wideband 0.6-6 GHz cellular/Wi-Fi-5 antenna (Taoglas Apex II APEXII.07.0150C / Linx 0.7-2.7 GHz multi-band — $25-60)
- 1× wideband 0.8-3 GHz cellular/sub-GHz antenna or discone (Mini-Circuits / Pasternack ~$30-60, or DIY discone per Vol 12 §12)
- 1× GPS L1 active patch antenna (Beitian BN-180 or similar — $5-15)
- 1× cellular/Wi-Fi-5 diplexer with ~4 GHz crossover (RFMicroDevices ZAPD-2 or Mini-Circuits ZADC-23-3B-S+ — $20-40)
- 1× sub-GHz/2.4-GHz triplexer (Mini-Circuits ZTS-3-2-S+ family or custom PCB — $40-80)
- 1× 2-way Wilkinson splitter for LoRa + SDR sub-GHz share (Mini-Circuits ZN2PD-K1+ — $20-30)
- Cables, U.FL-to-SMA pigtails, mounting hardware — $20-40
Total: ~$160-285 for the multiplexer chain + antennas; bonus is reclaimed case real estate for an additional Pi HAT or expansion module.
10. The efficiency tax of sharing — what you actually pay
For each sharing strategy, the in-band efficiency loss decomposes as follows:
| Stage | Loss source | Typical (dB) |
|---|---|---|
| Wideband antenna mismatch | Antenna SWR at off-design frequencies | 0-2 dB |
| Diplexer insertion | Filter pass-band loss | 0.4-1.5 dB |
| Splitter / combiner insertion | Wilkinson 2-way | 3.0-3.2 dB |
| Splitter / combiner insertion | Wilkinson 4-way | 6.0-6.3 dB |
| Switch insertion | Single-element SP4T PIN diode | 0.3-0.6 dB |
| Connector / cable | Per coax termination | 0.05-0.5 dB each |
| Polarization mismatch | Wrong pol | 0-3 dB (V/H) to 20-30 dB (orthogonal) |
A typical Step 2 consolidation for the uConsole accumulates:
- Cellular: 1-2 dB diplexer + 0.5 dB pigtail = ~2 dB total
- Wi-Fi 5: 1-2 dB diplexer + 0.5 dB pigtail = ~2 dB total
- Wi-Fi 2.4 + BT: 1.5 dB triplexer + 0.5 dB pigtail = ~2 dB total
- LoRa: 1.5 dB triplexer + 3 dB splitter + 0.5 dB pigtail = ~5 dB total
- SDR sub-GHz: 1.5 dB triplexer + 3 dB splitter + 0.5 dB pigtail = ~5 dB total
- GPS: dedicated antenna, no sharing loss
For receive-mostly radios (Wi-Fi, BT, LoRa, SDR), the 2-5 dB losses don’t matter operationally — atmospheric and local noise floor dominate the link budget. For TX, the same losses reduce effective radiated power by the same dB — a 5 dB loss on LoRa drops effective TX from +20 to +15 dBm, which still meets most LoRa link budgets.
The losses become problematic only when:
- The base TX power is already marginal (LoRa at +14 dBm with 5 dB sharing loss = +9 dBm effective — possibly insufficient for the deployment range)
- Receivers are near their noise-floor limit (deep-space SDR work, GPS weak-signal acquisition)
- Regulatory ERP limits are tight and the sharing loss eats into the budget the operator wants for antenna gain
For typical pen-test / SDR / cellular-handheld use, Step 2’s losses are acceptable.
11. PIN-diode, MEMS, and mechanical RF switches
For Strategy 3 (band-switching), the switch element determines the switch’s electrical and mechanical characteristics. Three common types:
11.1 PIN-diode switches
- Mechanism: a PIN diode with appropriate bias acts as a low-resistance conductor (forward bias, low ohms) or a high-resistance isolator (reverse bias, high impedance). Control via a few mA of DC bias.
- Bandwidth: DC to ~10 GHz typical, some parts to 40 GHz.
- Switching speed: ~100 ns to 1 μs typical.
- Insertion loss: 0.3-0.8 dB per element.
- Isolation: 25-50 dB.
- Examples: Skyworks SKY13351 (DC-3 GHz SP4T), Skyworks SKY13384 (40 MHz-12 GHz SP4T), Mini-Circuits ZASW-SP-T.
- Drawback: requires DC bias network, which adds blocking caps and chokes.
11.2 MEMS RF switches
- Mechanism: a micromachined metallic beam moves electrically (electrostatic or magnetic) to make or break RF contact.
- Bandwidth: DC to ~40 GHz.
- Switching speed: 10-50 μs typical.
- Insertion loss: 0.2-0.5 dB.
- Isolation: 30-60 dB.
- Examples: Analog Devices ADGM1304 (DC-40 GHz 4SP4T), Menlo Micro MM5120.
- Drawback: smaller power handling than PIN-diode (typically -20 to +20 dBm); more expensive ($10-50 vs $1-5).
11.3 Mechanical relay RF switches
- Mechanism: mechanical contacts close/open via electromagnetic or piezo actuation.
- Bandwidth: DC to 18 GHz (specialist coax relays).
- Switching speed: 1-20 ms typical.
- Insertion loss: 0.1-0.3 dB (the lowest of any switch type).
- Isolation: 60-90 dB (the highest of any switch type).
- Power handling: 100 W to several kW (the highest of any switch type).
- Examples: Tohtsu CX-520D, Sentec RF SP4T mechanical, Coto / OEG relays.
- Drawback: bulky, slow, expensive ($50-500), mechanical wear-out.
For the uConsole and similar handheld platforms, PIN-diode is the right choice — small, fast, no mechanical wear, low cost. MEMS is overkill unless the platform is high-end. Mechanical relays are out of scope for handhelds.
12. DIY build — a 2.4 / 5 GHz diplexer on FR-4
A complete recipe for the Wi-Fi diplexer that splits 2.4-2.5 GHz (low port) and 5.1-5.9 GHz (high port) — the most common Hack Tools sharing need.
12.1 Topology
A 3rd-order lumped-LC diplexer:
- Low-pass arm: C1-L1-C2 series-L-shunt-C filter, cutoff 3.5 GHz
- High-pass arm: L2-C3-L3 series-C-shunt-L filter, cutoff 3.5 GHz
- Common port: T-junction connecting both arms to one SMA jack
12.2 Bill of materials
| Part | Designator | Value | Package | Approx $ |
|---|---|---|---|---|
| Inductor 1 | L1 | 1.8 nH (or 1.6 nH) | 0402 | 0.05 |
| Inductor 2 | L2 | 1.0 nH | 0402 | 0.05 |
| Inductor 3 | L3 | 0.7 nH | 0402 | 0.05 |
| Capacitor 1 | C1 | 0.5 pF | 0402 NP0 | 0.05 |
| Capacitor 2 | C2 | 0.6 pF | 0402 NP0 | 0.05 |
| Capacitor 3 | C3 | 0.3 pF | 0402 NP0 | 0.05 |
| FR-4 PCB | — | 2-layer, 0.8 mm | 50×30 mm | 5 (from JLCPCB) |
| SMA connectors | 3× | edge-mount female | — | 6 |
Total parts: ~$12 for one diplexer. Can build ~10 from a single $5 PCB order (panelized).
12.3 Layout notes
- Common port on one short edge; low and high ports on opposite long edges.
- 50 Ω microstrip (1.5 mm trace on 0.8 mm FR-4) connecting each port to the corresponding LC filter.
- Filter components grouped tightly (<5 mm total path) to minimize parasitic inductance.
- Ground via stitching every 2 mm around the perimeter and under each component pad.
12.4 Tuning
- Sweep S11 at each port with a NanoVNA (Vol 24) into 50 Ω terminators on the other ports.
- Expect ~0.5 dB insertion loss in-band, ~30 dB rejection 2 GHz outside the band.
- If insertion loss is high (>1 dB), check component values — 0402 part tolerances at GHz frequencies are sensitive; consider higher-Q parts (Coilcraft 0402DC inductors, Murata GJM series caps).
- If isolation is low (<20 dB), check ground stitching and shielding around the filter.
13. DIY build — a 4-port band-switching matrix with band-decoder
A complete recipe for an SP4T PIN-diode band switch that routes one antenna to one of four radios, controlled by an Arduino reading the active band from a transceiver’s BCD output.
13.1 Topology
- SP4T switch IC: Skyworks SKY13351-378LF (DC-3 GHz, +33 dBm power, 0.4 dB IL, 28 dB isolation).
- Control: 2 GPIO pins from Arduino (encoded as 2-bit BCD for 4-state select).
- Power: 3.3 V via LDO from 5 V USB or onboard regulator.
- Antenna port: SMA-female (common).
- Radio ports: 4× SMA-female (or U.FL on a smaller PCB).
13.2 Bill of materials
| Part | Value | Package | Approx $ |
|---|---|---|---|
| SKY13351-378LF | DC-3 GHz SP4T | QFN 12-lead | 3-5 |
| Arduino Nano | — | — | 5 |
| 3.3 V LDO | LM1117-3.3 or AP2112 | SOT-223 / SOT-25 | 0.5 |
| 0.1 μF caps | bypass | 0402 | 0.05 each ×6 |
| 10 kΩ resistors | pull-down | 0402 | 0.05 each ×2 |
| SMA connectors | 5× | edge-mount | 10 |
| FR-4 PCB | 2-layer | 60×40 mm | 5 (from JLCPCB) |
| Enclosure | Hammond 1591B | aluminum | 8 |
Total: ~$35-45 per matrix.
13.3 Arduino sketch (excerpt)
// SP4T band-switching matrix control
const int CTL1 = 7;
const int CTL2 = 8;
const int BAND_INPUT = A0; // 0-5V from rig's band decoder output
void setup() {
pinMode(CTL1, OUTPUT);
pinMode(CTL2, OUTPUT);
pinMode(BAND_INPUT, INPUT);
}
void loop() {
int band_v = analogRead(BAND_INPUT);
int sel = 0;
if (band_v < 256) sel = 0; // 0-1.25V: port 1 (sub-GHz)
else if (band_v < 512) sel = 1; // 1.25-2.5V: port 2 (2.4 GHz)
else if (band_v < 768) sel = 2; // 2.5-3.75V: port 3 (5 GHz)
else sel = 3; // 3.75-5V: port 4 (cellular)
digitalWrite(CTL1, sel & 1);
digitalWrite(CTL2, (sel >> 1) & 1);
delay(10);
}13.4 Validation
Sweep S21 through each port with NanoVNA; expect 0.4-0.6 dB through the active port and >25 dB isolation between active and idle ports. Verify the switch logic with a multimeter on CTL1/CTL2 and ensure all four states route the antenna correctly.
14. Commercial buys — Mini-Circuits, Marki, Pasternack, AVX, K&L
Sorted by price tier (USD, mid-2026):
14.1 Diplexers
- Budget (~$20-50): Mini-Circuits ZADC-23-3B-S+ (2.3 GHz crossover, $25), RFMicroDevices ZAPD-2 (Wi-Fi 2.4/5 GHz, $30), generic Aliexpress diplexers ($10-15).
- Mid (~$50-200): Mini-Circuits ZX10R-14-S+ (custom-spec diplexers, $50-150), Marki D-0205S (DC-20 GHz wideband, $200).
- Premium (~$300-1000): Pasternack PE9501 (premium Wi-Fi diplexer with 50 dB iso), K&L Microwave custom diplexers ($500-2000 depending on bands).
14.2 Triplexers / Quadplexers
- Budget (~$50-150): Mini-Circuits ZTS-3-2-S+ (3-way 0.3-2.5 GHz, $75), Comet CFX-431A (HF/VHF/UHF triplexer, $130).
- Mid (~$200-500): Mini-Circuits custom quadplexers, Diamond MX-72D (HF/VHF/UHF, $200), Anaren custom multiplexers.
- Premium (~$700-3000): K&L Microwave, Mercury Systems custom-spec’d cellular triplexers for OEM applications.
14.3 PIN-diode SP4T switches
- Budget (~$10-40): Pasternack PE71S2007 (DC-3 GHz, $25), Mini-Circuits ZASW-SP4T-50DR-S+ ($45).
- Mid (~$50-200): HMC241 (Hittite, DC-3.5 GHz, $35 each), Mini-Circuits ZAS-SP6T (6-port, $120).
- Premium (~$300-1500): complete switch modules with control logic, integrated bias networks, ruggedized for outdoor use.
14.4 Wideband multi-radio antennas (the “one antenna for all” approach)
- Budget (~$20-60): Taoglas Apex II APEXII.07.0150C (multi-band cellular/Wi-Fi, $35), Linx ANT-W63WS-SMA (0.7-6 GHz, $25), Pulse W3911 (similar, $30).
- Mid (~$60-200): Taoglas Maximus FXP series (wideband flexible PCB, $80-150), Antenna Company multi-band omnis ($100-200).
- Premium (~$300-1500): Aaronia HyperLOG (calibrated EMC-grade wideband), specialized cellular/Wi-Fi/GPS combo antennas with integrated LNA and filtering.
14.5 What to avoid
- Cheap eBay “all-band” antennas with no published S-parameters — usually wideband in name only, with deep nulls in actual operating bands.
- Diplexers without explicit isolation specs at the bands you care about.
- Switches without published power-handling and 3rd-order intercept (IP3) — needed to check IMD performance under multi-signal loading.
15. Common gotchas and myths
- “More isolation is always better” — false. Higher isolation costs filter complexity (more components, more loss). 30-40 dB is plenty for most amateur sharing scenarios where TX powers are <1 W.
- “A diplexer is bidirectional” — true, but only if both ports are matched at the relevant frequencies. A diplexer optimized for 2.4 + 5 GHz behaves like a 2-port filter at other frequencies, with unpredictable response. Use the diplexer only in its designed bands.
- “Sharing antennas always degrades performance” — partial truth. For receive-only sharing among multiple SDRs, the noise floor is set by the antenna, not the receiver; 3 dB of split loss doesn’t degrade SNR. For TX sharing, every dB of loss is a dB out of the link budget.
- “GPS can share with anything below 1.5 GHz” — risky. GPS L1 is 1.575 GHz; the sub-GHz triplexer port that “almost” covers it has typically -10 to -20 dB rejection at 1.5 GHz. The GPS LNA can compensate, but the polarization mismatch + sensitivity floor mean weak-signal acquisition suffers.
- “PIN-diode switches break at higher TX power” — true. Most PIN-diode SP4T switches handle +20-30 dBm comfortably; some specialized parts go to +40 dBm (10 W). Above that, mechanical or specialty PIN-diode designs are needed. Most uConsole-class platforms stay well below the +30 dBm limit.
- “I can put a generic 50 Ω resistor across unused ports” — sort of. A 50 Ω termination on an unused port stabilizes the filter response but doesn’t prevent leakage from the active port. Use proper terminators (Mini-Circuits ANNE-50+, $3 each) on every unused switch port.
16. Resources
- Mini-Circuits app notes: AN-10-005 (power splitters), AN-30-038 (RF switches), AN-30-024 (PIN-diode switching)
- Skyworks SKY13351 datasheet — canonical PIN-diode SP4T reference
- Analog Devices ADGM1304 datasheet — MEMS RF switch reference
- Pozar, Microwave Engineering (4th ed.), Ch. 8 (filter design — the diplexer crossover math)
- Marki Microwave application notes on broadband multiplexer design
- K&L Microwave filter catalog (cellular / GPS / WLAN multiplexer reference designs)
- Cross-link: Vol 18 (passive splitters), Vol 19 (active distribution & preamps), Vol 29 (per-radio use-case matrix), Vol 32 (antenna farms — multi-antenna systems)