Why sampling deserves its own plan
Buyers often think of samples as a formality between quote and order. In smart glasses they are where the product is decided. A frame that looked right in a photo may pinch behind the ear; a speaker that sounded fine in a showroom may leak at the volume your customers use; a cell that met runtime on the bench may fall short with AI features on. Every one of those findings is cheap at the sample stage and expensive after batch production.
The sample path depends on the route. A catalog program needs one short round; a logo program needs one or two; a full custom program goes through engineering samples and EVT and DVT rounds before a golden sample is signed. The table below summarises the typical reference ranges, confirmed in writing per program.
The stages at a glance
| Stage | Route | Purpose | Typical timing |
|---|---|---|---|
| Catalog sample | Catalog | Check fit, finish, sound, runtime and app on the existing model | 5-10 working days |
| Logo sample | Logo and packaging | Approve logo method and position, color, device name, prompts and box | 10-15 working days |
| Engineering sample | Full custom | First build of the new design, often from soft tooling or early parts | Within 30-60 days overall, including 25-45 days of tooling |
| EVT sample | Full custom | Engineering validation: does the design meet the spec | Per program |
| DVT sample | Full custom | Design validation on production tooling; units for certification testing | Per program |
| Golden sample | All routes | The signed reference for batch production and inspection | After final approval |
Catalog and logo samples
For catalog and logo programs, the sample is the finished product or something very close to it. Use it to test what customers will notice. Wear it for a full day. Check the fit across a few head sizes; frame measurements such as lens width, bridge and temple length are covered in smart glasses sizing. Play music at the volume you expect customers to use and note runtime, comparing it with the typical category range of 4-13 hours for music, which varies with volume and features. Make calls outdoors and in a car. Pair it with the phones your customers use, try the app and check the firmware version.
For a logo program, add the brand checks: logo position clear of the touch strip and speaker port, logo durability after a rub test, the device name in the pairing list, voice prompts, box and manual. Sign off each item in writing; a short approval sheet with photos avoids arguments later.
Engineering, EVT and DVT samples
In a custom program, samples are a sequence of builds that each answer a different question.
- Engineering samples. The first physical builds of a new frame, temple or board. They may use soft tooling, 3D-printed or machined parts and hand assembly. Use them to check form, fit, cell and speaker space, button and touch position and basic electrical function. Expect them to look rough.
- EVT (engineering validation test). Builds that should meet the spec on paper. Test acoustics (frequency response and leak), call quality, runtime, charging, Bluetooth range and stability, hinge cycling and early environmental tests such as artificial sweat soak and IPX splash. Fix what fails and record each change.
- DVT (design validation test). Builds on production tooling with production parts. This is the build used for formal certification testing - FCC Part 15, CE RED with EN 300 328, EN 301 489, EN 62368-1 and EN 18031, battery tests to IEC 62133-2 and UN 38.3, and optical tests to ISO 12312-1 or ANSI Z80.3. Radio lab testing commonly takes 3-6 weeks per lab and scope.
Some programs add a small pilot run after DVT to check the assembly line itself. Whatever the naming, the rule is the same: do not send units for certification until the board, antenna, battery and enclosure are frozen, because a later change can trigger retesting. Engineering topics by module are on engineering.
The golden sample
The golden sample is the physical reference for the whole program. It is a finished pair, or a set of pairs, built on production tooling, with approved firmware, lenses, logo and packaging, signed and dated by the buyer and the partner plant. Usually at least two are kept: one with the plant, one with the buyer or the desk. Every batch and every pre-shipment inspection is compared with it.
If anything changes after signing - a new cell supplier, a revised firmware build, a different lens tint - a new golden sample should be signed. A written change notice rule in the program makes that automatic.
Common sample-stage mistakes
- Testing only at the desk. Runtime, call quality and fit behave differently outdoors, in wind, in a car and after a full day of wear.
- Approving by photo. Color, finish and logo contrast look different on screen; approve physical samples under daylight.
- Too few units. One sample cannot show variation. Ask for several units of the same build so differences in hinge feel or speaker level are visible.
- Undocumented changes between rounds. Ask for a short change list with each new sample: what changed, why and which test it affects.
- Lens decisions left to the end. A late switch to polarized or photochromic lenses changes the lens partner and the optical tests.
Sample costs, quantities and shipping
Sample charges vary by route and are confirmed per program; catalog samples are usually modest, while custom engineering samples carry tooling and engineering effort. Order enough units for your own testing, listing photos and any retailer or lab requests: a few per color for catalog work, more for EVT and DVT where some units are consumed in destructive tests. Samples with built-in cells ship by express under UN3481, PI 967 Section II, with a UN 38.3 test summary; allow 5-10 days in transit. Typical ranges for every route are on MOQ and lead time, and the next step after a signed golden sample is described in pre-shipment inspection. To plan a sample round, send the line and route through the quote form.