DEV Community: Yang Fei

Why Your 10 W Attenuator Failed at 5 W

Yang Fei — Sat, 14 Nov 2026 00:48:58 +0000

A 10 W attenuator running at 5 W continuous should be the least interesting part on the bench. It still comes back. When it does, the report usually says "overpowered", and the test log says 5 W, and both are wrong.

The rating on the label is measured under conditions that almost never exist in a rack: 25 °C ambient, VSWR 1.0, bolted to a heat sink of a specified size, continuous, at the low end of the band. Move any one of those and the 10 W part is a 6 W part or worse. Here is where the margin goes.

An attenuator is a heater with connectors

A 10 dB attenuator turns 90 percent of the input into heat. A 20 dB pad turns 99 percent into heat. At 5 W input through a 10 dB pad, 4.5 W is being dissipated inside a package the size of a thimble. That is the whole design problem, and it is why the mechanical drawing matters as much as the electrical one.

Four things that eat the margin

Ambient temperature. Most parts derate linearly from full power at 25 °C to zero at the maximum operating temperature, often 125 °C or 150 °C. At 55 °C inside a closed rack you are already down 20 to 30 percent. At 85 °C, which is a normal interior temperature for equipment in a shelter, a 10 W part may be rated for 4 to 5 W.

VSWR at the load. Reflected power comes back into the pad and is dissipated there too. With a 2:1 mismatch on the output, assume 15 to 20 percent more dissipation than the matched calculation. This is the number that turns a 5 W test into a 6 W thermal load.

Mounting. A 10 W attenuator normally assumes it is bolted to metal with thermal compound. Hanging in free air on a bench, convection carries away a fraction of that, and the case temperature climbs until something gives. If the datasheet does not state the mounting condition for the rating, the rating is a marketing number.

Frequency. Thin-film resistor geometry and skin effect both change with frequency, and the thermal density along the film is not uniform. A part rated for full power at DC is often limited at 18 GHz or 40 GHz. Check the power-versus-frequency curve, not just the headline.

Peak is not average

For pulsed radar and for anything with a duty cycle, the average rating governs heating and the peak rating governs voltage breakdown across the resistor. A part rated 10 W average and 1 kW peak will not survive 1 kW average for a millisecond, and it will flash over at 1 kW peak if the pulse shape is sharper than the test pulse. Ask what pulse width and duty cycle the peak rating assumes.

What the failure looks like

The three signatures are distinguishable, and they tell you which limit you hit:

Open circuit after months of thermal cycling. The resistor film cracked from repeated expansion, not from one overload event. You were running close to the thermal limit the whole time.
Discoloured or reflowed centre pin, attenuation still correct. The connector, not the resistor, was the limit. This is common on small packages pushed to their rated power at high frequency.
Attenuation value drifted up by a few tenths of a dB. Film aging from sustained over-temperature. The part still works, the calibration no longer does.

How to size one properly

Compute dissipated power: input minus output, in linear terms, not dB.
Derate from the vendor curve for your actual ambient inside the enclosure.
Add 20 percent if the load VSWR can reach 2:1.
Apply 1.5 to 2x margin for a part that will sit in a rack for ten years.
Measure the case temperature after 30 minutes at the intended power. If you are within 25 °C of the maximum case temperature, you have no margin and should go up a size.

What to ask for before you buy

Derating curve, with the mounting condition explicitly stated
Thermal resistance, case to ambient
Maximum case temperature
Power rating versus frequency, not just at the band edge
Connector power limit, which is often lower than the resistor limit
Whether the part is intended for conduction cooling or convection

Where I sit

I work on these parts at DB Design / MeiXun in Wuxi, so read this as notes from the manufacturing side rather than a neutral survey. We build fixed and variable attenuators alongside coaxial loads and terminations and the coaxial switches that usually sit next to them in the same rack.

If an attenuator in your system keeps dying, the derating curve and the mounting are where the answer usually is, not the wattage on the label.

RF Switch Matrix Design: The Isolation Budget Nobody Calculates

Yang Fei — Fri, 13 Nov 2026 08:09:53 +0000

Every switch matrix spec sheet I get asked to quote lists port count and insertion loss. Almost none lists a leakage budget. That is the number that decides whether an 8x8 matrix actually behaves at 40 GHz, and it is the number that is hardest to fix after the chassis is built.

Here is how the three common architectures differ, and where each one quietly fails.

Three ways to build the same matrix

Fan-in tree. Each output port gets one SP(N) switch, and its throws land on the N input lines. One switch in the path, so loss is the lowest of the three and isolation is a single switch's off-isolation. It is non-blocking only for one-to-one connections: if input 1 is feeding output 3, nothing else can use input 1 until you move that switch. Cheapest, lowest loss, least flexible.

Full crossbar. A splitter on every input, a switch on every output. Any input can reach any number of outputs at once, which is what you want for a parallel ATE rack or a redundancy test. The cost is the splitter: a 1:8 resistive splitter burns 9 dB before the signal ever reaches a switch, and a Wilkinson version burns 9 dB too plus makes the unused branches reflective.

Two-stage cascade. Fewer switch throws, more routing options, at the price of two switches in series. Loss roughly doubles in dB, switching time doubles, and the isolation arithmetic gets interesting in a way that usually does not go the way the designer expects.

Isolation adds in dB until it stops adding

Two 60 dB switches in series look like 120 dB on paper. In a real chassis you will measure somewhere between 85 and 110 dB and then stop seeing improvement, because the leakage floor is no longer the switch. It is the path around the switch: connector shells bonded to the same panel, the gap between two SMA bulkheads 15 mm apart, a ribbon cable crossing under the cavity, the ground plane of the driver board.

The practical consequence: if you need more than about 100 dB, do not buy it by cascading switches. Buy it with packaging. Separate cavities, feedthroughs instead of bulkhead connectors, absorber on the lid, physical distance between the input and output connectors, and a driver board that is not sitting under the RF path.

Also check isolation at the frequency you actually use. A switch quoted at 80 dB at 1 GHz is often 50 to 55 dB at 40 GHz, because the off-state capacitance of the contact and the coupling across the cavity both scale with frequency.

The number ATE engineers should be asking for

Absolute insertion loss is calibrated out on the first run. What does not calibrate out is the spread of insertion loss across paths, and how much that spread moves after a few hundred thousand cycles.

Ask for three things instead of one:

Insertion loss spread across all paths, worst pair, at the top of the band
Insertion loss repeatability on a single path over 100k operations
Return loss spread, same conditions

A matrix with 2.0 dB typical loss and 0.15 dB spread is more useful than one with 1.4 dB typical and 0.6 dB spread, because the second one forces a per-path calibration file and re-verification every quarter.

Terminate the ports you are not using

An unterminated reflective SP8T output reflects the leakage back into the cavity. In a crossbar the effect shows up as ripple in a path that has nothing to do with the unused port. Internally terminated switches cost more and add a little loss, but they make the matrix's behaviour independent of which ports happen to be idle, which is exactly what you need when the test sequence changes every product revision.

Latching relays, not failsafe, for large matrices

This is the failure mode that kills matrices in the field. A failsafe relay drops to a default position when the coil loses power. Put 128 of them in a chassis, lose the supply while the source is live, and you have hot-switched 128 contacts simultaneously into whatever load was there. Contacts pit, insertion loss drifts, and the failure looks random for months afterwards.

Use latching relays, drive them with a sequenced controller that never changes state while RF is present, and put a hardware interlock on the source. Also size the supply for the inrush, not the average: 128 latching relays at 28 V pull real current for a few milliseconds each, and a supply that sags during a re-route will leave relays in an undefined state.

Before you send the RFQ

Have a number for each of these:

Port count, and whether you need one-to-many or only one-to-one
Highest frequency, and isolation required at that frequency
Insertion loss spread target, not just typical loss
Switching time including driver and sequence
Lifetime in full re-routes, not individual switch operations
Control interface, and what happens on power loss
Connector series and the number of mating cycles the bench will do

Where I sit

I work on these parts at DB Design / MeiXun in Wuxi, so read this as notes from the manufacturing side rather than a neutral survey. We build electromechanical coaxial switches from SPDT to SP12T and DPDT rated to 2 million operations and 67 GHz, waveguide switches to 110 GHz, and PIN diode switches for the cases where the re-route has to happen inside a pulse.

If a matrix in your rack has started drifting, the leakage budget and the relay type are usually where the answer is.

Electromechanical vs. PIN Diode RF Switches: What Actually Fails First

Yang Fei — Fri, 13 Nov 2026 01:03:38 +0000

Every RF bench has a switch that dies young. In ATE racks and radar front ends, the switch is often the first part to fail, and it is almost never the part the designer blamed. The usual cause is a mismatch between the technology chosen and the way the switch is actually driven: hot switching, DC on the RF line, or a control interface that promises more than the contact can deliver.

This is not a datasheet comparison. It is the short list of things that decide which of the two switch families you should specify, and what will kill either one early.

Two technologies, one decision

An electromechanical (EM) coaxial switch moves a metal contact. Insertion loss is very low, isolation is excellent, and it works from DC upward, which matters if your signal carries a DC or low frequency component. The price is mechanical wear, a few milliseconds of switching time, and contact bounce during the transition.

A PIN diode switch has no moving parts. It switches in microseconds or less, survives millions of operations without wearing out, and is the only sane choice when you need to change path inside a pulse. The price is higher insertion loss, worse isolation at the top of the band, forward current to bias the diodes, and a device that is unforgiving of transients on its bias line.

Where each one wins

DC to 67 GHz with no bias network: EM coaxial.
Changing path inside a 10 us window: PIN diode.
Millions of operations over a decade: PIN diode.
Highest isolation and lowest loss above 40 GHz: EM coaxial, and waveguide above 50 GHz.
Hot switching into a mismatched load: neither is happy, but EM survives longer.

What kills them early

Three failure modes account for most field returns.

Hot switching. Breaking a live RF path at high power arcs the contact. Derate power handling by roughly half when the switch operates while the transmitter is on, and specify the hot switching rating separately from the cold one.
DC or low frequency on the line. A PIN diode switch cannot pass DC, and a bias tee in front of it will not turn it into a DC coupled path either. Check what your source actually puts on the centre conductor.
Control transients. An EM switch driven by a slow or bouncy supply can sit between positions and dissipate the full input power in the contact. A PIN diode driver that floats or overshoots will forward bias a diode that is supposed to be off.

A specifying checklist

Before you send an RFQ, have a number for each of these:

Isolation at the specific frequency, not the band minimum
Insertion loss and return loss over temperature
Switching time including the driver, not just the switch
Lifetime in operations, at the rated power
Power handling, cold and hot
Control interface: TTL, 12/28 V, latching or failsafe
Connector type and sex, and whether the body is the ground reference

Where I sit

I work on these parts at DB Design / MeiXun in Wuxi, so read the above as notes from the manufacturing side rather than a neutral survey. We build electromechanical coaxial switches from SPDT to SP12T and DPDT, rated for 2 million operations to 67 GHz, plus waveguide switches to 110 GHz and PIN diode switches for the cases where microseconds matter.

If a switch in your design keeps failing, the checklist above is usually where the answer is.