Solutions

Module or finished DTU? Cost, certification and time

Both routes use the same 470 MHz radio. The difference is 23 weeks, about EUR 40,000 of NRE and a break-even volume near 2,440 units.

Two designs arrive at the same 470–510 MHz link, the same 17 dBm, the same three data rates. One is an 18-pin LCC module measuring 40.5 × 25 × 4.4 mm; the other is an 84 × 83 × 25 mm box with RS485 and RS232 terminals. The radio inside them is not the variable. Everything around the radio is, and that is where the money, the calendar and the certification risk actually sit.

The choice is usually framed as "buy or build", which hides the real question: how many units will you ship, how soon, and who signs the declaration of conformity? Those three answers decide it. This article works through the sums with the parameters of two real products from this catalogue, marks every cost assumption as an assumption, and shows where the module route fails after the PCB comes back.

Start with the sum, not the spec sheet

A module is cheaper per unit and more expensive per project. That is not a slogan; it is a fixed cost divided by a volume. Write it as:

cost_module(N) = NRE + unit_module × N
cost_dtu(N)    = unit_dtu × N
break-even     = NRE / (unit_dtu - unit_module)

With the planning numbers used in this article — EUR 40,300 of non-recurring engineering, EUR 47.48 per module-based unit, EUR 64.00 per finished DTU — the break-even is 2,440 units. Below that, the box wins; above it, the module wins, and the gap widens by EUR 16.52 for every extra unit.

Total programme cost against volume for the module route and the finished DTU route, crossing at about 2,440 units

Those monetary figures are assumptions, not quotations: 380 engineering hours at EUR 60/h, EUR 12,000 of lab time, EUR 5,500 of tooling, and a EUR 46.10 bill of materials with a 3% rework adder for a 94% first-pass yield. Replace them with your own numbers. The structure of the answer rarely changes, because the ratio between a boxed DTU and a module plus its carrier board is the input that moves the answer most, and it is the first number to replace with your own quotes.

The point of writing it as a formula is that each term can be argued with separately. Sales will argue about volume. Engineering will argue about hours. Nobody argues about the EUR 12,000 lab bill until it arrives, which is why it belongs in the model on day one.

What each route already includes

The datasheets make the split visible. The TP1107-MP30-L gives you a radio, a supply range and a serial port at TTL level. The TP5600 gives you a product: a wide-range supply, protected field interfaces, a housing and a mounting method.

Item

TP1107-MP30-L module

TP5600 finished DTU

Form factor

40.5 × 25 × 4.4 mm, 18-pin LCC SMT

84 × 83 × 25 mm indoor housing, screw fixing

Supply

5.0–6.0 V; 650 mA in transmit, 20 mA receive, 6 µA sleep at 5 V

8–28 V DC, 12 V typical, ≤1 W

Field interface

AT UART at TTL level, wake-up input, status LED, SWD

RS485 ×1 and RS232 ×1, 1200–115200 baud, 8 data bits

Radio

470–510 MHz, FSK, 17 dBm, -117 dBm at 2.4 kbps

470–510 MHz, 17 dBm, 2.4/19.2/76.8 kbps

Antenna

LCC pad or IPEX

SMA female

ESD (published)

not stated on the module specification

air ±15 kV, contact ±8 kV

Temperature

-40 to +70 °C

-40 to +70 °C

Read the supply row twice. A 24 V cabinet bus and a 5–6 V module are not connected by a wire; they are connected by a DC-DC converter with input protection, and that converter has to survive whatever the cabinet does at switch-off. Read the interface row the same way: the DTU's RS485 front end includes the transceiver, the transient protection and the bias network that a module design has to add, lay out and test.

The module is the right starting point when the radio is one function of a board you were going to design anyway — a sensor node with its own microcontroller, a gateway with Ethernet and a power budget, a product whose differentiator is the application, not the enclosure. The DTU is the right starting point when the box is the product and the schedule is shorter than a redesign.

The certification question: what a module grant transfers

This is the part of the decision that gets the least attention and causes the most schedule damage.

In the United States, a transmitter module can be certified as a modular transmitter under 47 CFR § 15.212. To get that grant, the module must, among other things, have its own shielding, buffered modulation and data inputs, its own power-supply regulation, and an antenna arrangement that satisfies the unique-connector rules; it is tested stand-alone rather than inside a host. When those conditions are met, a host manufacturer may use the module's grant instead of certifying the same transmitter again — but only inside the grant's conditions.

The FCC's own module integration guidance is explicit that the host still owns everything the module grant does not cover: unintentional-radiator requirements for its digital circuitry, any other transmitters in the product, RF exposure in the final configuration, and the labelling that tells the world which module is inside ("Contains FCC ID: …"). Change the antenna type or gain beyond what the grant allows, or run the radio at a power the grant does not cover, and the host can no longer lean on that certification.

In Europe the situation is simpler and stricter. The Radio Equipment Directive 2014/53/EU has no modular-approval mechanism at all. The economic operator placing the finished radio equipment on the market is responsible for the whole assembly meeting the essential requirements, and issues the declaration of conformity and the CE mark. A module's test reports are useful evidence for the radio part; they are not a transferable approval for your enclosure, your power supply or your EMC behaviour.

Question

Module route

Finished DTU route

Who holds the radio approval

the module grantee

the DTU supplier

What you inherit

the transmitter certification, within its grant conditions

the supplier's declaration for that model

What you still owe

host EMC (e.g. Part 15 Subpart B), RED assessment of the assembly, labelling, RF exposure in your enclosure

integration checks against the supplier's declared conditions

Change the antenna or add a PA

new approval; the old grant no longer applies

not applicable

Typical extra lab spend

several thousand to low five figures, plus 4–8 weeks

near zero

So the honest accounting is this: the module route saves the transmitter certification, not the product certification. For a small LCC module, the host-side work — EMC of your board, the enclosure's effect on the antenna, the safety assessment — is the larger half of the bill.

Time is a line item too

A purchase order for a finished DTU can be placed before the firmware exists. The module route cannot ship until the design, the pre-scan, the fixes and the certification are done, and those steps are serial.

Critical path from kick-off to first shipped unit: 23 weeks for the module route and 6 weeks for the finished DTU

That 17-week gap is not idle time; it is delay you either absorb or pay for. Put a weekly price on it and the break-even moves:

break-even(C) = (NRE + 17 × C) / saving_per_unit
C = 0      → 2,440 units
C = 1,000  → 3,469 units
C = 3,000  → 5,528 units
Break-even volume rising with the assumed cost of one week of schedule delay

A project with liquidated damages, a seasonal installation window or a customer acceptance date in the contract often has a cost of delay far above EUR 1,000 a week. At that point the module route needs volumes in the thousands before it is the cheaper decision — and it still needs someone on the team who can own an RF layout and an EMC pre-scan.

There is a second timing effect that does not show up in the model: cash. NRE is spent in the first four months, before any revenue, while the DTU's cost is paid per unit as you ship. For a small company, the module route is partly a financing decision.

Five things that appear after the PCB arrives

1. The antenna does not fit the physics

At 470 MHz the wavelength is 63.8 cm, so a quarter wave is 15.96 cm. The TP5600's longest dimension is 84 mm, about 0.13 λ. No enclosure this size can hold a clean quarter-wave radiator; anything internal is electrically short and loaded, or the antenna is external through the SMA. A module with an IPEX connector inherits the same problem, plus a new one: the ground plane it sits on is now yours.

The -117 dBm sensitivity on the module sheet is a conducted figure, and the gap between a conducted and a reported signal reading is where most commissioning surprises come from. Measure it again with the antenna fitted and the board in its enclosure; a 2–3 dB loss is ordinary, and a metal DIN rail or a cast enclosure lid can cost more. If your link budget was built on the conducted number with no margin, the enclosure is where it breaks.

2. ESD arrives through the terminal block

The finished DTU publishes air ±15 kV and contact ±8 kV. The module sheet does not publish an ESD rating for its pins, because the module has no field terminals. On a meter cabinet, the RS485 pair is the path an installer's hand, a floating laptop or a nearby contactor uses to reach your electronics. TVS diodes, a common-mode choke and a layout that keeps the transient current away from the radio ground are part of the module route's real cost — and they are exactly the parts the DTU price already contains.

3. The 24 V bus is not 24 V

A module specified for 5.0–6.0 V in a cabinet with a 24 V bus needs a converter, reverse-polarity protection and transient clamping. The converter must also supply the 650 mA transmit peak at 5 V that the module sheet lists — 3.25 W of input for 50 mW of RF — without collapsing the rail during the burst. Size the converter for the peak, not the average, and check it at -40 °C if the cabinet is outdoors.

4. Heat is yours now

Both products are rated -40 to +70 °C, but the rating is measured at the product's own boundary. In a sealed enclosure, every watt stays inside. At a 10% transmit duty cycle the module's 3.25 W burst averages about 0.33 W, which is manageable; a streaming application at 30% duty puts about 1 W inside a box the size of a paperback. If the same box sits in the sun, the internal rise adds to the ambient, and the 70 °C limit is reached earlier than the datasheet suggests.

5. A power increase is a new approval

The catalogue already shows the packaged answer to range: the TP5500 outdoor DTU delivers 26 ± 1 dBm from an IP65 housing. That extra 9 dB is not a setting on the 17 dBm module. Adding an amplifier after a certified module takes the product outside the grant's power and antenna conditions, which means a new certification — the most expensive way to buy decibels.

A worked example: 1,200 four-wire meter points

Take a retrofit with 1,200 RS485 meters, one gateway per building, and a requirement to read every meter every 15 minutes. The radio side is fixed: 2.4 kbps, 64-byte responses, 512 bits per response, about 0.21 s of airtime per meter before framing and acknowledgements. Polling 1,200 meters serially consumes roughly 256 s of channel time per 15-minute window — 28% of the window, before retries. Neither route changes that number; the packaging decision buys cost and time, not capacity.

At 1,200 units the totals are:

Volume

Module route

DTU route

Difference

1,200 units

EUR 97,280 (EUR 81.1/unit)

EUR 76,800 (EUR 64.0/unit)

DTU cheaper by EUR 20,480

2,440 units

EUR 156,151

EUR 156,160

break-even

4,000 units

EUR 230,232 (EUR 57.6/unit)

EUR 256,000 (EUR 64.0/unit)

module cheaper by EUR 25,768

At 1,200 units the module route costs 27% more and arrives 17 weeks later. Unless the enclosure must be integrated into a machine or the volumes are expected to grow past 2,400 within the product's life, the box is the better decision. This is the unglamorous answer most projects should reach, and it is why so many industrial retrofits use a DIN-rail DTU.

At 4,000 units the same arithmetic reverses: the module route saves EUR 25,768 and gives you control of the enclosure, the connector and the labelling. That is the volume band where designing the board starts to pay.

What to verify on the bench before you commit

Whichever route you take, these checks belong in the plan before tooling is cut.

  1. Conducted versus radiated sensitivity. Measure packet error rate at the module's rated sensitivity with a conducted connection, then repeat with the antenna fitted inside the final enclosure. Budget for 2–3 dB; investigate anything above 4 dB.
  2. VSWR across 470–510 MHz. A return loss worse than about 10 dB (VSWR 2:1) at the band edges means the antenna is detuned by the enclosure or the ground plane.
  3. ESD on the field terminals. Contact discharge at ±8 kV on the RS485 lines, with the unit powered and communicating. A finished DTU publishes this figure; a module design has to demonstrate it.
  4. Supply transients. Switch a 24 V load on the same bus and watch the 5 V rail during a transmit burst. The module's 650 mA peak is the load that exposes a weak converter.
  5. Thermal soak at maximum duty cycle. Run the transmitter at the highest duty the application allows, at the rated ambient, until the case temperature stabilises; then check the radio still meets its sensitivity figure.
  6. Pre-scan before tooling. An EMC pre-scan on the first prototype is cheap; a failed certification after the enclosure tool is cut is not.
  7. Production test. A fixture that measures transmit power, current and receiver sensitivity on every unit is part of the module route's cost, and it is the only way the yield assumption in the model stays true.

FAQ

Is a certified module automatically allowed in my product?

No. The grant covers the transmitter under specific conditions — antenna type and gain, output power, supply, shielding. If your design stays inside those conditions and you follow the integration instructions, you can reference the module's certification; anything outside them needs a new filing.

Can I add an external amplifier to a 17 dBm module?

You can build it, but you cannot keep using the module's certification. The output power and antenna conditions change, so the transmitter needs a new approval. If 26 dBm is the requirement, a packaged product such as the TP5500 is usually cheaper than certifying your own amplifier chain.

Does a finished DTU remove all certification work?

No. It removes the radio certification for the module inside it. You still own the conformity of the machine or installation you build around it, any other directives that apply, and — if you sell it under your own brand — the obligations that come with becoming the manufacturer of record.

Which route makes sense at 500 units a year?

At 500 units the module route's NRE is spread over too few units to pay back: the total is roughly EUR 64,000 against EUR 32,000 for the DTU. Choose the module only if the board carries other functions you need anyway, or if the enclosure is a hard requirement that no off-the-shelf unit meets.

The decision is not about which product is better. It is about which fixed cost you are willing to carry, and whether the calendar can absorb 17 weeks in exchange for EUR 16.52 a unit. Write both numbers down before the first schematic is drawn, because after that the answer is already chosen.

More in Solutions