Recommended order of checks
It is best to work from the physical layer up to the application, because a fault in a lower layer renders useless anything done above it.
1. Physical connection
- Correct cable for the interface: on serial connections, a straight-through or crossover cable depending on the equipment.
- Connectors properly seated; on parallel buses, screws tight.
- Length and number of devices within what the bus allows.
2. Interface configuration
- Serial: speed, data bits, parity, stop bits and flow control must be identical on both sides. If not, you receive garbled characters or nothing at all.
- Addressed bus: the instrument address must match the one used by the program and must not be repeated.
- Network: IP address, subnet mask, port and firewall rules blocking the connection.
3. Software and drivers
A missing or incompatible driver, or a wrongly named instrument resource, prevents the session from opening. Another program already holding the port open leaves it busy, and the second attempt fails.
4. Syntax and timing
- Commands with the wrong terminator: many instruments expect a specific end-of-line character.
- Queries sent before the previous one has finished, which produce out-of-order responses or full buffers.
- Timeout too short for slow operations, such as a measurement with long integration.
- Commands not supported by the model or by its current mode, for example in local mode.
What you see in practice
No response at all suggests the physical layer or the address; strange characters, serial parameters; incomplete or out-of-step responses, terminators and timing; intermittent failures, cables or electrical noise. Querying the instrument’s error queue after a failed command usually shows the exact reason.
What to check first
Try an identification query with a simple terminal, outside your program. If it responds, the problem is in the application; if not, it is in the connection or configuration.