Home›Academy›Specimen Papers
The standard, demonstrated

Specimen papers.

Ten items per level of the TCE (TELECIS Certified Engineer) examinations, with the answers — and the reasoning. Published so a candidate knows the standard before enrolling, and so anyone specifying against the credential can see exactly what it asks.

ACADEMY / SPECIMEN

These items are specimens of the TCE (TELECIS Certified Engineer) examinations: written to the live blueprints, reviewed and trialled like live items, and permanently excluded from live papers. They demonstrate format, depth, and — in the rationales — the way the examination thinks: requirement first, mechanism second, fashion never.

Work them closed book. A miss is not a verdict; the rationale names the domain, and the learning paths sequence the reading that repairs it.

A
TCE-Core

Level 01 —
ten items.

Ten items in the format and at the standard of the live paper — with the answer, and the reasoning. Domains reference the published TCE-Core blueprint.

ITEM 01 · SIGNAL FUNDAMENTALS

A source must deliver 4K60 to a display over an HDMI 2.0 (18 Gbps TMDS) path. Which signal format will not pass?

A — 4K60 · 4:2:0 · 8-bit
B — 4K30 · 4:4:4 · 8-bit
C — 4K60 · 4:4:4 · 8-bit
D — 4K60 · 4:4:4 · 10-bit

ANSWER — D  ·  4K60 4:4:4 at 8-bit sits just inside 18 Gbps (≈17.8 Gbps with encoding overhead). Adding depth pushes the rate past TMDS — 10-bit 4:4:4 requires HDMI 2.1 FRL signalling end to end. The lesson the domain examines: bandwidth is resolution × rate × chroma × depth together, never a resolution number alone.
ITEM 02 · SIGNAL FUNDAMENTALS

In a boardroom, spreadsheet gridlines and small text on the 4K display appear fringed and soft, while video plays cleanly. The most likely cause is:

A — HDCP falling back to 1.4
B — chroma subsampling to 4:2:0 somewhere in the path
C — EDID capping the format at 1080p
D — cable attenuation at the top TMDS rate

ANSWER — B  ·  Fine single-pixel detail is exactly what chroma subsampling discards — motion video hides it, text shows it. A 1080p cap would soften everything uniformly; attenuation produces sparkles or dropouts, not clean-but-fringed text. Verify the negotiated format at the sink, then walk the path for the device forcing 4:2:0.
ITEM 03 · EDID

A routed system feeds six 4K displays and one legacy 1080p projector. Sources must stay at 4K for the displays. The correct EDID strategy is:

A — pass through the projector's EDID — it is the safest common format
B — present a fixed internal 4K EDID at each input, and scale at the projector's output
C — clone the EDID of the largest display to all inputs and outputs
D — disable EDID and force the sources manually

ANSWER — B  ·  Lowest-common-denominator passthrough drags every source to 1080p for one legacy sink. A static internal EDID holds sources at 4K; the one output that needs 1080p gets it by scaling at the edge. This is the pattern the EDID domain examines: decide what the source should produce, then present exactly that.
ITEM 04 · EDID

After a power cut, a boardroom PC reboots before the matrix and comes up at 640×480 with no audio. The durable fix is:

A — replace the HDMI cable with an active optical cable
B — train users to reboot the PC after the matrix
C — persistent EDID emulation at the matrix input, so the source always sees a display
D — a longer HDMI handshake timeout at the display

ANSWER — C  ·  The PC read no EDID at boot and fell back to VGA-era defaults. Emulating EDID at the input means the source sees a valid display whether or not the route — or the far-end sink — is up. Order-of-boot instructions are not commissioning; they are a fault waiting for the next power cut.
ITEM 05 · TRANSPORT SELECTION

A single fixed 4K60 link must run 90 m from a lectern to a rack, in conduit, with no plans to re-route. The engineering answer is:

A — passive HDMI cable of premium construction
B — HDBaseT over shielded CAT6A, terminated to standard
C — an AV-over-IP encoder/decoder pair on the corporate LAN
D — DisplayPort with an MST hub at the rack

ANSWER — B  ·  Passive copper is finished long before 90 m at 4K rates. AV-over-IP buys routing flexibility this fixed point-to-point link will never use — at the price of network dependency. HDBaseT is engineered for exactly this: full bandwidth to 100 m over one shielded CAT6A run. Transport is selected by requirement, not by fashion.
ITEM 06 · TRANSPORT SELECTION

Which requirement, on its own, decides the choice for AV-over-IP rather than a matrix?

A — the highest possible image quality
B — zero switching latency
C — endpoint count that will grow across floors and buildings
D — immunity to network outages

ANSWER — C  ·  Quality and latency are met by both architectures when tiers are chosen correctly, and network dependency argues against AV-over-IP, not for it. What a matrix cannot do is scale past its frame: AV-over-IP buys growth with switch ports. When the estate will grow, the network is the matrix.
ITEM 07 · HDCP

A UHD player outputs HDCP 2.3 Type 1 content through a splitter to a 2.3-capable display and an HDCP 1.4-only display. The result is:

A — both displays show the content
B — neither display shows the content
C — the 2.3 display shows it; the 1.4 display shows an HDCP error or black
D — both show it at reduced resolution

ANSWER — C  ·  Type 1 content rules forbid downconversion to HDCP 1.x — the 1.4 sink fails authentication while the compliant path plays. The domain examines exactly this: HDCP outcomes are decided per-branch by version and content type, and a mixed estate must be designed for the weakest sink it must actually serve.
ITEM 08 · HDCP

In a cascade of splitters, the displays nearest the source authenticate; the farthest tier shows black. The first parameter to check is:

A — cable category on the last hop
B — repeater depth and device count against the HDCP topology limits
C — EDID strategy at the source
D — mains earthing at the far rack

ANSWER — B  ·  HDCP limits the tree: repeater depth and total device count are finite, and the failure appears at the far edge of the cascade — exactly where the limits are exceeded. Counting the topology takes one minute and, in a design review, should have happened before anything was installed.
ITEM 09 · CABLING & PHYSICAL LAYER

An 85 m HDBaseT run passes through two patch panels and drops intermittently at 4K. The correct resolution is:

A — raise the transmitter's output drive
B — re-terminate as one continuous, directly-terminated run
C — add a mid-span CAT coupler of higher category
D — reduce the format to 1080p and close the ticket

ANSWER — B  ·  Every connection point spends part of the link budget, and at 85 m there is little budget left to spend. Direct termination recovers it. Reducing the format hides the defect in the infrastructure — the room was sold as 4K, and the physical layer domain examines whether the installed link honestly supports it.
ITEM 10 · COMMISSIONING METHOD

Cabling is verified and racks are powered. Per the TELECIS commissioning sequence, the next step is:

A — connect all sources and displays, then test the room as a whole
B — bring up the matrix alone and verify management access with no sources connected
C — load the control processor's program
D — set EDID on every input from the design document

ANSWER — B  ·  The sequence exists so that each step depends only on the steps before it. A matrix that is reachable and healthy with nothing connected is a known-good baseline; sources and displays are then added one at a time, each verified. A room tested 'as a whole' that fails has twenty suspects — a room built in sequence has one.
B
TCE-Advanced

Level 02 —
ten items.

Ten items at the standard of the written component. The practical is assessed in the lab, against the published rubric — no specimen can substitute for it. Domains reference the published TCE-Advanced blueprint.

ITEM 01 · MATRIX ARCHITECTURES

An operations wall must switch sources with no black frame, no resync, and no visible transition. The matrix class this requires is:

A — a standard switching matrix with fast EDID handling
B — a seamless matrix with per-output frame buffering and scaling
C — an AV-over-IP system with IGMP fast-leave
D — a mechanical HDMI patch field

ANSWER — B  ·  Seamless switching is bought with a frame buffer per output: the new source is acquired, aligned, and cut in without the sink ever losing lock. The cost — roughly a frame of added latency — is the trade the domain expects you to name, and to sign off as acceptable for the room.
ITEM 02 · MATRIX ARCHITECTURES

A 4K estate must also feed three legacy 1080p meeting-room displays. The clean solution is:

A — cap the whole estate at 1080p for uniformity
B — per-output scaling on the three legacy outputs, native 4K elsewhere
C — external down-scalers at each legacy display
D — replace the three displays before commissioning

ANSWER — B  ·  Per-output scaling localises the compromise to the outputs that need it, leaves the estate native, and removes three external boxes that would otherwise need power, mounting, and firmware of their own. Whole-estate capping punishes every room for the oldest display in the building.
ITEM 03 · AV-OVER-IP DESIGN

Which requirement forces the uncompressed / 10G-class tier, rather than JPEG-class 1G or H.26x?

A — distribution to 200 endpoints at lowest cost
B — pixel-accurate imagery with imperceptible latency for critical review
C — transport across a WAN to a second site
D — wireless presentation from laptops

ANSWER — B  ·  Scale-at-cost and WAN reach argue for the compressed tiers — that is what they are for. The 10G tier exists for the paths where compression itself is the risk: sub-frame latency and mathematically faithful pixels for surgical, radiological, or measurement-grade review. Tier selection is requirement-led, and this is the requirement that ends the discussion.
ITEM 04 · AV-OVER-IP DESIGN

Eight decoders on one access switch each subscribe to an 850 Mbps stream. The minimum uplink from that switch is:

A — 1 Gbps — multicast is shared, one copy serves all
B — 2 × 1 Gbps in a LACP bundle
C — 10 Gbps
D — the uplink is irrelevant; traffic stays local to the switch

ANSWER — C  ·  Multicast is shared only when subscribers want the same stream. Eight decoders on different sources pull eight distinct streams through the uplink — ≈6.8 Gbps — before protocol overhead. LACP does not help: one flow hashes to one member link. Endpoint planning is uplink arithmetic, done before the switch is ordered.
ITEM 05 · NETWORK ENGINEERING

Minutes after an AV-over-IP system starts, every port on the switch saturates and unrelated devices drop off. The classic cause is:

A — duplicate IP addresses on the encoders
B — IGMP snooping enabled with no querier — multicast flooding as broadcast
C — spanning tree blocking the uplink
D — jumbo frames enabled end to end

ANSWER — B  ·  Snooping prunes multicast only while a querier maintains group membership; with no querier the tables age out and every stream floods every port. The discipline is exact: snooping on, exactly one querier per AV VLAN, verified by inspecting group membership on the switch — not by waiting to see if the network survives.
ITEM 06 · NETWORK ENGINEERING

Corporate IT will host the AV system on existing switches. The minimum acceptable configuration is:

A — default VLAN with QoS trust enabled
B — a dedicated AV VLAN with IGMP snooping and one querier, and verified non-blocking capacity for the stream load
C — a separate physical network — nothing else is acceptable
D — port mirroring so AV traffic can be monitored

ANSWER — B  ·  A converged network is workable when the multicast domain is contained and the arithmetic closes — that is the defensible minimum, and it is what you write into the network requirements document the course teaches you to issue. Demanding physical separation everywhere is not engineering; it is the answer of someone who cannot state the requirements.
ITEM 07 · DANTE

A Dante path crosses four gigabit switch hops. The appropriate device latency setting is:

A — 0.15 ms — always use the minimum
B — 0.25 ms
C — 0.5 ms
D — 5 ms — always use the maximum

ANSWER — C  ·  Dante's guidance maps latency to hop count: 0.25 ms is comfortable to about three switches; four hops calls for 0.5 ms of margin against queuing under load. Minimum-latency bravado produces intermittent dropouts that commission clean and fail in the meeting; blanket maximums waste a budget other paths may need. Set it from topology, record it.
ITEM 08 · DANTE

The laptop running Dante Controller is closed and taken off site. The audio routes:

A — fail after the lease interval expires
B — continue — subscriptions persist in the devices themselves
C — continue only if a backup controller is elected
D — fall back to analogue

ANSWER — B  ·  Dante Controller is a configuration tool, not a session controller: subscriptions are stored in the endpoints and re-establish themselves after power cycles. This is why naming discipline matters — the estate must remain legible to the next engineer who opens Controller a year later.
ITEM 09 · FLEET FIRMWARE

A firmware update must be applied across a working estate of 40 devices. The correct practice is:

A — update everything simultaneously overnight to minimise version skew
B — pilot one non-critical unit, verify against a checklist, then stage the rollout with a recorded rollback path
C — update only the devices currently exhibiting faults
D — defer all updates until a fault forces the issue

ANSWER — B  ·  Simultaneous fleet updates convert one bad build into an estate outage. The staged pattern — pilot, verify, roll out by criticality with rollback ready and versions recorded — is how uptime survives maintenance. Both extremes offered as distractors are common in the field, and both are estate-management by luck.
ITEM 10 · DIAGNOSTIC METHOD

One display in an AV-over-IP estate is black; the rest are healthy. In layer order, the first check is:

A — the decoder's stream subscription in the management software
B — HDCP negotiation at the display
C — physical link — port state and link lights at the decoder and its switch port
D — the EDID presented to the source

ANSWER — C  ·  Layer order exists so that no conclusion rests on an unverified layer below it. Subscription state is meaningless on a port with no link; HDCP cannot negotiate across a dead PHY. Thirty seconds at layer one prevents an hour of software archaeology — this is the method the practical examines under fault injection.
C
TCE-Specialist

Level 03 —
ten items.

Ten items drawn across the four endorsements — Medical, Broadcast, Command & Control, Hospitality. The live assessment adds a review of submitted installation history; these items illustrate the written standard only.

ITEM 01 · MEDICAL

For the path carrying diagnostic imagery to a review display, compression is avoided because:

A — compressed streams cannot carry HDCP
B — codec artefacts on smooth grayscale gradients can be indistinguishable from — or mask — clinical detail
C — compression adds unacceptable cost
D — networks are prohibited in clinical spaces

ANSWER — B  ·  On diagnostic gradients, ringing and banding are not cosmetic: they can read as artefact-versus-pathology ambiguity. The diagnostic chain is specified for pixel integrity end to end — uncompressed or mathematically lossless transport, with the display chain's calibration preserved. Convenience tiers belong on the conference feed, not the reading path.
ITEM 02 · MEDICAL

In an operating theatre design, the routing hardware belongs:

A — inside the sterile field, closest to the surgeon
B — in the technical zone outside the theatre, with control surfaces at the nurse station and nothing serviceable inside the sterile field
C — in the ceiling void above the table
D — in a rack in the corridor, shared with the adjacent theatre

ANSWER — B  ·  Anything inside the field must be sealed, cleanable, and ideally untouched during a procedure; service access mid-list is not a plan. Locating switching in the technical zone keeps maintenance outside clinical workflow, and shared-corridor racks couple two theatres' risk. The endorsement examines designs that respect the room's clinical reality, not just its signal flow.
ITEM 03 · MEDICAL

Acceptable added latency for a live surgical camera feed to the in-theatre display is on the order of:

A — under two frames — low tens of milliseconds
B — a quarter of a second
C — half a second, if image quality is higher
D — latency is irrelevant on a local link

ANSWER — A  ·  The surgeon's hands are closed-loop on that image: perceptible lag degrades instrument control. Low tens of milliseconds — uncompressed or line-based processing — is the working budget; a quarter-second is disorienting and indefensible. Latency is a stated, measured number in the medical specification, never an accident of tier choice.
ITEM 04 · BROADCAST

Production sources are locked to house reference (black burst / tri-level) so that:

A — their colourimetry matches
B — switches and mixes between sources are clean, with no timing jump at the cut point
C — cable runs can be longer
D — embedded audio stays in sync

ANSWER — B  ·  Reference timing aligns sources at the switcher so a cut lands invisibly; unlocked sources force frame synchronisers and their latency into the chain. Colour and audio have their own disciplines. The endorsement expects the timing plan — who is locked, to what, and where sync is added — to be drawn before equipment is chosen.
ITEM 05 · BROADCAST

A production gallery runs an SDI backbone; the monitor wall is HDMI. Conversion belongs:

A — at the source, so the backbone is HDMI throughout
B — at the display end — SDI to the wall, converted per monitor
C — anywhere — the formats are electrically equivalent
D — nowhere: SDI monitors are mandatory in a gallery

ANSWER — B  ·  SDI earns the backbone: locked, robust, long-reach, routed cleanly. Converting early surrenders those properties everywhere downstream. Per-monitor conversion at the wall keeps the plant SDI and treats HDMI as what it is here — a display-edge interface. The pattern generalises: hold the professional transport as deep into the chain as possible.
ITEM 06 · COMMAND & CONTROL

A 24/7 operations wall is fed by two processors. The design requirement that matters most is:

A — both processors from the same production batch
B — source inputs and wall regions distributed so no single failure — processor, PSU, or feed — darkens the whole wall
C — a spare remote control in the operations room
D — matched cable lengths to both processors

ANSWER — B  ·  Continuous-duty design is failure-mode design: interleave wall regions across processors, duplicate critical sources, N+1 the power, and rehearse the degraded state. A wall that loses half its regions but keeps every critical feed visible is a design; a wall that goes dark together is an incident report.
ITEM 07 · COMMAND & CONTROL

For operator positions switching between many computers, KVM-over-IP is specified with:

A — best-effort latency on the corporate LAN
B — sub-frame video latency and USB switched deterministically with the video, on a qualified network
C — video-only switching, with keyboards attached per computer
D — VNC software on the operators' PCs

ANSWER — B  ·  An operator's confidence collapses when keys land on the wrong machine or the cursor lags the hand. Deterministic USB-follows-video switching and sub-frame latency are the specification; the network carrying it is engineered and qualified like any other mission system. Software remote-desktop is a different tool for a different job.
ITEM 08 · HOSPITALITY

Four hundred displays across a resort argue for AV-over-IP chiefly because:

A — picture quality exceeds every alternative
B — central management — monitoring, sources, and firmware for the whole estate from one seat, and growth by adding switch ports
C — it eliminates displays' own firmware
D — guests expect visible technology

ANSWER — B  ·  At estate scale the nightly question is operational: what is on, what failed, what version runs where. Centralised AV-over-IP answers it from one seat and prices growth at a switch port. Quality parity is achievable in several architectures — manageability at four hundred endpoints is not.
ITEM 09 · HOSPITALITY

Redundancy budget across a venue estate is allocated:

A — uniformly — every path protected equally
B — by revenue and reputation criticality — dual-pathed sports bar and ballroom, single-pathed corridor signage with spares strategy
C — only to the longest cable runs
D — nowhere — hospitality tolerates failure

ANSWER — B  ·  Uniform redundancy doubles cost to protect corridors nobody pays for; zero redundancy puts the match feed at risk on a Saturday night. The endorsement examines the allocation argument: rank paths by cost-of-failure, protect the top tier properly, and cover the rest with spares and swap procedure.
ITEM 10 · REGULATED ENVIRONMENTS

Across regulated verticals, the commissioning record and as-built documentation matter because:

A — they are marketing collateral for the integrator
B — they are the auditable evidence — what was installed, configured, verified, and signed, and the baseline every later change is controlled against
C — they are required only when a fault occurs
D — they replace operator training

ANSWER — B  ·  Clinical, government, and broadcast estates live under change control: the signed record is what proves the system's verified state and anchors every subsequent modification. This is why the TELECIS method ends in a signed document, and why renewal by installation evidence is submitted on exactly that record.
Ready

Sit it for real.

If the specimens read like a fair day's work, you are ready to prepare properly. Core is complimentary.