COTS (Commercial Off-The-Shelf)
Why Programs Reach for COTS RF Hardware
The acronym dates to U.S. acquisition reform of the mid-1990s, when the Perry memorandum directed defense programs to buy commercial products and standards wherever practical instead of writing bespoke military specifications. In RF and microwave work the appeal is concrete: a catalog connectorized amplifier, attenuator, or synthesizer ships from stock at a fraction of the non-recurring engineering cost of a custom design, and its electrical performance is already characterized over a published commercial range. A program trades the last few decibels of optimization and the assurance of a full military qualification for schedule, price, and supply availability.
The cost difference is not subtle. A custom 18 GHz GaAs amplifier MMIC carries non-recurring engineering of tens to hundreds of thousands of dollars amortized over the build, while a functionally similar COTS module sells for tens to low hundreds of dollars per unit at quantity. The penalty appears later. Commercial parts are graded only to roughly 0 to +70 °C, are not screened for the latch-up and single-event effects that matter on orbit, and follow commercial product lifecycles of 5 to 7 years against platform lives of 20 to 30 years. Those three gaps, temperature, environment, and longevity, drive nearly every COTS-versus-custom decision in aerospace and defense RF design.
Upscreening and the Risk Equation
When a COTS part is otherwise ideal, engineers close the environmental gap through upscreening rather than redesign: added temperature cycling, burn-in, and electrical screening borrowed from MIL-STD-883 push a commercial part toward extended-temperature or hi-rel acceptance. Because the underlying die and package were never designed to those limits, screening lowers but cannot zero the residual risk, so a quantified failure-rate estimate accompanies any COTS insertion into a critical signal chain.
COTS Cost and Reliability Relations
Ctotal = Cunit + Cscreen + Cobsol + Crequal
Upscreen yield / lot acceptance (PDA gate):
PDA (%) = 100 × (nfail / nburn-in) ≤ 5% (lot rejected if exceeded)
Last-time-buy quantity (life-of-type):
QLTB ≈ (Yremaining × Uannual) + Qspares + Qattrition
Where Cunit = catalog price, Cscreen = upscreen and DPA cost, Cobsol = obsolescence mitigation, Crequal = redesign and requalification reserve; PDA = percent defective allowable; Yremaining = remaining program service years the part must cover after the last buy, Uannual = annual usage. A low Cunit can be dwarfed by Cobsol + Crequal over a 25-year program.
Component Grade Comparison
| Grade | Temp Range | Screening Basis | Relative Unit Cost | Lead Time | Typical Use |
|---|---|---|---|---|---|
| COTS (commercial) | 0 to +70 °C | Vendor datasheet only | 1× (baseline) | Stock to 2 wk | Lab, ground, low-stakes |
| COTS upscreened / MOTS | -40 to +85 °C | Added MIL-STD-883 methods | 3 to 8× | 8 to 16 wk | Avionics, rugged ground |
| Extended / industrial | -40 to +85 °C | Manufacturer flow | 2 to 5× | 4 to 12 wk | Outdoor base stations |
| Military (Class B) | -55 to +125 °C | Full MIL-PRF-38535 / 883 | 20 to 100× | 26 to 52 wk | Tactical, missile |
| Space (Class S) | -55 to +125 °C + rad | QML-V, DPA, rad-lot | 100 to 1000× | 40 to 78 wk | Satellite, deep space |
Frequently Asked Questions
What is the difference between COTS, MOTS, and NDI?
COTS parts are bought from a commercial catalog and used exactly as sold. MOTS (Modified Off-The-Shelf) starts as COTS but receives a defined change, a different connector, a screened temperature grade, or a conformal coat, which voids the original qualification and requires accepting the delta. NDI (Non-Developmental Item) is the broadest government category and covers COTS, MOTS, and reused military items developed under earlier funding. A catalog 0 to 18 GHz amplifier is COTS; that amplifier rebiased and burned-in to a custom profile is MOTS.
How do you upscreen a COTS RF component for high-reliability or space use?
Upscreening qualifies a commercial-grade part to a harsher environment through added testing, not redesign. A typical RF flow borrows MIL-STD-883 methods: external visual, temperature cycling (-55 to +125 °C, 10 to 200 cycles), constant acceleration, leak test on hermetic parts, 160 to 240 hours burn-in at +125 °C with a PDA gate, and final electrical hot, cold, and ambient. Because the die and package were never designed to those limits, screening reduces but never removes lot-variation and infant-mortality risk, so space programs add DPA and full lot traceability.
How is COTS obsolescence and DMSMS risk managed over a 20-year program?
Commercial RF semiconductors often live only 5 to 7 years against a 20 to 30 year platform life, creating Diminishing Manufacturing Sources and Material Shortages (DMSMS) exposure. Mitigations include last-time buys with bonded storage, life-of-type purchases sized to forecast attrition, qualified second sources and form-fit-function alternates, and obsolescence monitoring that flags end-of-life notices 12 to 18 months early. With no drop-in replacement, the choices narrow to aftermarket supply, die banking, or a redesign that may requalify the surrounding subsystem.