PLESTY Keithley 2410 SourceMeter

A PLESTY device module for the Keithley 2410 SourceMeter, a high-voltage source-measure unit (SMU). An SMU sources a voltage or a current and measures both at the same time. The 2410 sources up to 1100 V or 1.05 A and talks SCPI (the standard text command language of laboratory instruments) over VISA (the standard software layer for GPIB, serial and USB instruments) on GPIB or RS-232.

The source can put hazardous voltage on its terminals, so this module refuses before it acts. Read Safety rules before switching anything on.

This module replaces an older PLEXY driver for the same instrument. It is not backwards-compatible with it.

Supported hardware

Installation

git clone https://gitlab.com/plesty/hub/devices/keithley/plesty-k2410-source-meter.git
cd plesty-k2410-source-meter
uv sync

A VISA backend must be available to talk to a real instrument (for example NI-VISA, plus a GPIB adapter for a GPIB link). The mock address needs none.

Configuration

The address of the instrument is configuration, not code. Copy .env.example to .env and edit it:

• GPIB: GPIB0::<n>::INSTR. The factory GPIB address is 24; the front panel shows the current one under MENU > COMMUNICATION > GPIB.
• RS-232: ASRL<n>::INSTR. Set the instrument to 9600 baud, 8N1, CR terminator.
• mock: the word mock runs an in-memory simulator of the instrument, with no hardware.

Each value resolves as CLI argument, then environment variable.

Quick start

Run the device as a TCP server:

cp .env.example .env   # set K2410_ADDRESS
uv run python -m plesty.k2410_source_meter
uv run python -m plesty.k2410_source_meter --address ASRL3::INSTR --tcp-port 5555
uv run python -m plesty.k2410_source_meter --address mock   # simulator

Use it from Python (local, on the simulator):

from plesty.k2410_source_meter import Device

dev = Device(address="mock")  # a VISA address such as "GPIB0::24::INSTR" for the instrument
dev.init()
dev.connect()
reading = dev.source_and_measure(
    source_function="VOLT", level=5.0, compliance=1e-3, source_range=20.0
)
print(reading["voltage"], reading["current"])
dev.disconnect()

Preflight

Before the server binds its port, the module opens the configured address, reads the identification string (*IDN?) and requires MODEL 2410. When the configured address does not answer as a 2410 but another GPIB address does, the message names that address and tells you to set K2410_ADDRESS to it; the address is never changed for you. --skip-preflight serves anyway, for protocol work.

Safety rules

The 2410 answers a bad setting with an error-queue entry only after it has acted on the neighbouring ones, which is too late for a source that may carry 1100 V. So every rule below is checked in software before the first command is sent; a violation raises a refusal and sends nothing.

• Explicit compliance. The compliance is the limit on the quantity that is not sourced (amperes while sourcing voltage, volts while sourcing current). output_on is refused until configure_source set one explicitly since connect. There is no default compliance.
• voltage_limit defaults to 20 V. This is the instrument's hardware over-voltage protection. It is applied as 20 V on connect, so any higher voltage needs a deliberate write of voltage_limit first, while the output is off. A level above it is refused.
• output_on ramps from 0 V. It switches the output on at 0 and steps up to the configured level (steps of at most 10 V, a tenth of the range, with a short wait between steps). A warning is logged when voltage_limit is above 30 V.
• Operating boundary. Above 21 V the current compliance must be 21 mA or less; above 21 mA the voltage compliance must be 21 V or less.
• Compliance floor. A compliance may not be below 0.1 % of the measure range in force.
• Sourced quantity on the source range. The 2410 measures the quantity it sources on the source range and rejects a measure range for it (error 823). So voltage_measure_range and its _auto flag are refused while sourcing voltage, and the current ones while sourcing current; set voltage_range or current_range instead.
• No reading with the output off. measure never switches the output on.
• Stepped level changes. With the output on, set_level walks to the new level in steps.
• Refusals when unsafe. output_on is also refused while the instrument's error queue is not empty or, when interlock is on, while the interlock line is open.
• Shutdown on failure. safe_shutdown walks the level to zero in steps and then switches the output off. Every failure with the output on, and disconnect, do the same. A refusal changes nothing, so it does not shut down.

Parameters

Read with dev.query(name) and write with dev.write(name, value); dev.state returns all of them at once. Values are in volts, amperes and seconds.

The compliance written by configure_source is what allows output_on; the value applied on connect is a conservative baseline, not a choice. Change source_function, remote_sense, terminals and voltage_limit only while the output is off.

Functions

A compliance flag of True on a reading means the source did not reach its setpoint, so the reading is a limit and not a measurement.

Composite operations

Two methods of Device chain the functions above into a complete measurement. Both step the source down and switch the output off afterwards, also on failure.

• source_and_measure(source_function, level, compliance, source_range=None, keep_output_on=False) configures, switches on, takes one reading and switches off.
• run_iv_sweep(source_function, start, stop, points, compliance, spacing="LIN", delay=None, source_range=None) configures at the first level, switches on, sweeps and switches off.

The full generated reference (types, ranges, per-function arguments) is in the documentation under docs/; build it locally with CI= uv run plesty check.

Open items for the field test

Verified on a real 2410 over GPIB (2026-10-20, open terminals, at most 5 V):

• The instrument accepts :SYST:ERR:ALL? (read the whole error queue at once).
• :OUTP:INT:TRIP? answers 0 while the interlock is open and 1 while it is closed. The manual contradicts itself here; its SCPI reference table is right. With the interlock enabled and open, :OUTP ON is blocked with error 802.
• The five-element :READ? reading and its status bits; output_on stepping up from 0, set_level, safe_shutdown and a 6-point sweep; GPIB terminators.

Still to confirm, marked in the code:

• Whether NONE is the right setting for the 1100 V over-voltage protection.
• Whether :READ? behaves as expected with the output off (the driver refuses it anyway).
• The ramp step size (10 V) and wait (30 ms) on a real load, and compliance on a load.
• How the integration time is shared between the voltage, current and resistance functions.
• RS-232 line terminators.

License

LGPL-3.0-or-later. See LICENSE and COPYING.