OEM autosampler checklist: building one into your instrument

Building an OEM autosampler into your own analyzer means treating it as a subsystem with eight interfaces: mechanical, fluidic, electrical, communication, software, compliance, service and supply. Most integration trouble comes from the interfaces nobody wrote down, such as the current drawn during a needle move, what the module does after a lost connection, or who is responsible for the CE marking of the finished instrument. This checklist covers what to specify, what to test and what to ask a module supplier before you commit a design to one.

Key points

• Write a one-page requirements sheet first (volumes, precision, pressure, sample count, cycle time, environment) and compare modules against it, not against each other's datasheets.
• Size the power supply for the peak current with all motors moving, not for the average draw.
• Treat the control protocol as a contract: every command needs a defined reply, error code and timeout, and the module needs a known safe state after power loss or a lost connection.
• The manufacturer who places the finished analyzer on the market is responsible for its conformity. A module's test reports help, but they do not transfer that responsibility.
• Get lifecycle commitments in writing: change notices, a last-time-buy window, a spare-parts period and firmware change control.

Start with a one-page requirements sheet

Before looking at modules, write down what your analyzer needs from the sampling subsystem:

The values are illustrative. Fill in your own and mark which are hard limits and which are wishes. For the pressure line, see HPLC pressure ratings and what your autosampler needs.

Mechanical: footprint, mounting, service access and vibration

Footprint and moving envelope

Ask for a STEP model and a dimensioned drawing that includes the moving envelope: needle arm travel, door swing, rack removal path and cable bend radii. A module that is 155 mm deep on the datasheet may need noticeably more once the rack pull-out and connectors are included. For full-size reference points, see how much bench space an HPLC system takes.

Mounting and orientation

Check the mounting points, screw sizes and torque, and whether the module may be tilted or must stand level; orientation affects leak paths, vial seating and bubbles in the syringe. Check also that a flexing enclosure cannot shift the needle's position.

Service access

List every wear part (syringe seal, rotor seal, needle, needle seat, tubing) and check that a technician can reach each one without removing the module. A short seal change at the bench becomes a long job if the module has to be unbolted, unplugged and re-aligned first.

Vibration, shock and transport

Portable and vehicle-mounted analyzers see vibration that a bench instrument never does. Ask what the module has been tested to, then test the integrated instrument yourself; the IEC 60068-2 series has the standard methods (part 2-6 for sinusoidal vibration, 2-64 for broadband random vibration, 2-27 for shock). Check that the motion system homes reliably after a knock and that vials cannot jump out of the rack.

Fluidics: connections, materials, volumes and leaks

Connections

Agree on fitting types at every interface (for example 10-32 coned ports for high-pressure lines and 1/4-28 flat-bottom fittings for low-pressure lines), on tubing ID, and on who supplies the tubing between module, column and detector. Mixed fitting standards are a classic source of dead volume and leaks.

Wetted materials

Ask for the complete wetted-materials list: seals, needle, rotor and stator, tubing, syringe barrel. Check it against every solvent, buffer, pH range and cleaning agent you use, and against analytes that adsorb to metal or polymer. Materials affect carryover as much as wash settings do; see autosampler carryover.

Delay and transfer volumes

Every millimetre of tubing between the sample and the injection point is volume that must be flushed. Tubing volume is V = π × (d/2)² × L.

Worked example: a 1 m sampling line of 0.50 mm ID holds π × 0.25² × 1000 = 196 µL. Flushing three line volumes before each sample costs about 590 µL per cycle. The same line in 0.25 mm ID holds 49 µL, but flow resistance scales with 1/d⁴, so it is 16 times higher: by the Hagen-Poiseuille equation about 1.7 bar per metre at 1 mL/min of water instead of 0.11 bar. That matters if the line is fed by a low-pressure syringe.

Waste and leak handling

Decide where waste goes, how its level is monitored and what happens when it is full. Ask how the module catches leaks (catch cup, sensor), route that path away from your electronics, and check closed enclosures for solvent vapour build-up.

Electrical: 24 V DC, power budget and EMC

Supply voltage

OEM lab modules usually run from a low-voltage DC bus, often 24 V. Confirm the allowed input range, reverse-polarity protection, and whether the module tolerates the dips your bus sees when pumps or heaters switch on.

Power budget: average versus peak

Ask for three numbers: idle power, average power over a typical cycle, and peak current with all motors moving at once (plus inrush at power-up). Size the supply for the peak.

Worked example with illustrative numbers: a module idles at 6 W, averages 15 W over a cycle and peaks at 2.5 A for 300 ms when the arm and syringe move together. At 24 V that peak is 60 W. If the same bus feeds a pump peaking at 1.5 A and a detector at 1.0 A, the worst case is 5.0 A, or 120 W. A 150 W supply leaves 25% margin; a 100 W supply chosen from the averages would brown out at the worst moment.

EMC

Stepper drivers and switching regulators are noise sources; photodiode, conductivity and electrochemical detectors are noise victims. Keep motor cables away from detector signal lines, ask the supplier for its emission and immunity data, and plan to test the finished instrument as a whole.

Communication: protocol, errors, timeouts and firmware

Physical link

Common options are RS-232, RS-485, CAN and USB. A USB device that implements the CDC ACM class appears as a virtual COM port: on Windows 10 and 11 the in-box driver loads automatically, and Microsoft notes "you don't need to write your own INF file" (Microsoft). Ask which link and connector the module uses.

Protocol as a contract

Ask for the full command reference, not a summary. For each command it should define:

• the reply on success and every error reply, with codes;
• whether the command blocks until done or returns at once, and how to poll for completion;
• the longest time it can take, so the host can set a timeout;
• what happens if a new command arrives while one is running.

Error states and safe state

List the module's error states (needle crash, door open, leak, pressure fault, motor stall, homing failure) and check what it does in each. After power loss or a lost connection it should return to a known state and report why it stopped, never finish a half-done injection on its own.

Timeouts on the host

Every read needs a timeout and a decision for when it expires: retry, query status or abort. With pyserial, a read with a timeout set "may return fewer characters than requested" (pyserial API), so check every reply is complete:

import serial

port = serial.Serial("/dev/ttyACM0", 115200, timeout=2.0, write_timeout=1.0)

def ask(command: str) -> str:
    """Send one command and return one reply line, or raise on timeout."""
    port.reset_input_buffer()
    port.write((command + "\n").encode("ascii"))
    reply = port.read_until(b"\n")
    if not reply.endswith(b"\n"):
        raise TimeoutError(f"No complete reply to {command!r}")
    return reply.decode("ascii").strip()

Port name, baud rate, line ending and command strings come from the module's protocol document. More detail: controlling serial lab instruments from Python.

Firmware updates

Ask how firmware is updated, whether an interrupted update can leave the module unusable, whether you can stay on a validated version, and how the module reports its version so your software can check it at start-up.

Software integration

• Who owns timing. Your detector needs the moment of injection, either from a hardware signal (contact closure) or from the protocol reply that confirms the valve switch. Measure the jitter of whichever you use. The AS-48, for example, currently has no contact-closure or inject-start output, so its serial reply is the timing reference.
• Simulation and logging. A software fake of the protocol lets your team work before hardware arrives and test error paths; log every command, reply and error with a timestamp.
• Lab automation. If the analyzer must plug into scheduling software, look at SiLA 2 and OPC UA LADS; see lab automation integration standards.

Safety and compliance

Who is responsible

In the EU, "when a finished product incorporating another product is placed on the Union market, the manufacturer is responsible for the compliance of the complete product", and that manufacturer "can rely on the conformity assessment of the integrated product" when building its own documentation (Blue Guide 2022, section 2.1). The EMC guide says the same for EMC and recommends asking suppliers for the EMC characteristics and the method of incorporation of each component (EMC guide, 2025).

Standards and laws to check

This table is a starting point, not legal advice: scope decisions belong to your compliance engineer or test laboratory.

Reliability and service

• Ask for the rated life of each wear part in cycles and whether the module counts cycles and reports them.
• Ask what the power-up self-check covers, whether your software can read the results, and what tools each field repair needs.
• Set your maintenance interval from rated life and usage. Worked example: 40 injections a day on 250 days is 10,000 injections a year; a seal rated for 50,000 cycles under your conditions gives a yearly change a wide margin, while one rated for 15,000 leaves little.

Supply and lifecycle

Semiconductor suppliers formalise this with change notices and discontinuance notices. JEDEC J-STD-046 sets requirements for "timely customer notification of changes", and J-STD-048 (which superseded JESD48) aims to "ensure continuity of supply" when a product is discontinued (JEDEC). Analog Devices, for example, gives a 12-month last-time-buy period after a discontinuance notice (ADI). Ask your module supplier for the same in writing:

• notice period for any change to form, fit, function, materials or firmware;
• last-time-buy window and how much notice comes before it;
• spare-parts period after the last module ships;
• how the supplier manages obsolescence of its own key parts (motors, microcontroller, valve).

Worked example: you build 40 analyzers a year and expect to sell for 5 more years, plus 10% field spares. A last-time buy then needs 40 × 5 × 1.1 = 220 modules, which is why the notice period matters as much as the price.

Documentation to request

Questions to ask a module supplier

1. Which datasheet figures are measured on production units, and which are design targets?
2. Under what conditions were precision and carryover measured?
3. What is the peak current, and for how long?
4. What does the module do on power loss, on a lost connection and on each error?
5. Can we get the protocol document and a simulator before buying hardware?
6. How is firmware updated, and can we stay on a fixed version?
7. Which safety and EMC standards has the module been tested to, in which configuration, by whom?
8. What are the wear parts, their rated life, and can our technicians change them in the field?
9. How much notice do you give before a change or a discontinuation?
10. How long will you supply spare parts after the last module ships?
11. Can you supply a custom variant (tubing kit, connector, firmware option), and from what quantity?

The first question applies to us too: AS-48 values marked * on the specifications are design targets, and the module holds no certification yet.