Misdiagnosing DAQ ghosting wastes time replacing a working sensor. Compare the suspect channel alone with the same channel in the scan before changing the wiring. Then change one acquisition condition at a time and retain the raw records.
This checklist helps you test a settling-time explanation on a multiplexed analog input. It provides blank records and reasons to stop that diagnosis; it supplies no measured device results or universal timing setting. This is an independent technical publication, unaffiliated with NI or Emerson.
How DAQ ghosting works
In a multiplexed acquisition, successive channels share an analog conversion path. After switching channels, that path needs time to settle. If conversion happens too soon, the next reading can retain an influence from the previous input. NI’s digitization guide, pages 15–16 describes this as ghosting.
NI’s PCIe-6340 scanning documentation describes charge injection and explains why a high source impedance makes settling harder. It also identifies cable capacitance and input-range changes as relevant conditions. Those are reasons to record the complete signal path, rather than diagnose from similar-looking traces alone.
Record the hardware and baseline before testing
Find the channel-to-converter map in the actual device manual. A device can have multiple converters and still multiplex inputs within each group: the PXIe-4309 manual, page 2-8, documents both operating arrangements. Its grouping is a model-specific example, not a map for your unit.
Copy this setup record into your test notes. Leave an unavailable value blank or write Unknown / verify on the unit. A source’s output impedance and the DAQ’s input impedance belong in separate fields.
| Setup field | Evidence to retain | Your entry |
|---|---|---|
| Device, module, driver version | Identification and configuration export | |
| Converter architecture and channel grouping | Manual section or block diagram | |
| Suspect input and preceding scanned input | Actual channel order, including scan wraparound | |
| Terminal mode and input ranges | Configuration for each input | |
| Source output impedance | Sensor or conditioner specification, with conditions | |
| Cable and intervening circuit | Length, filters, dividers, buffers, connectors | |
| Requested and actual timing | Per-channel sample rate and conversion timing, if exposed | |
| Reference and acceptance band | Independent reference, application requirement, units |
Choose the error measure before the comparison: for example, deviation from a reference or the size of an unwanted variation on a nominally steady input. Record units and the analysis interval. Use the same measure throughout; do not call a trace acceptable just because it looks smoother.
Test the channel alone, then restore the scan
For this comparison, keep the physical connections and source operating conditions unchanged. Remove other inputs from the acquisition task rather than unplugging their wires. Preserve the suspect input’s range, terminal mode, scaling, and processing; record any timing the driver changes automatically.
- Save the original scan configuration and an unaveraged data record.
- Acquire the suspect channel alone and compare it with the chosen reference.
- Restore the original scan and repeat the observation.
- Repeat the pair of runs if the source changed during the comparison.
NI’s ghosting support article describes correct single-channel measurements that become contaminated during scanning. This is a useful diagnostic pattern, not proof of a particular failed component.
| Run | Enabled channels and actual timing | Reference and suspect reading, with units | Raw file |
|---|---|---|---|
| Original scan | |||
| Suspect channel alone | |||
| Original scan restored |
Stop this comparison if the single-channel result is already wrong. Resolve its reference, range, scaling, wiring, and source behavior first. For a floating voltage source, use the bias-return and common-mode worksheet. A single-channel fault does not exclude an additional scan-related error, but it prevents treating ghosting as the complete explanation.
Change timing, order, and voltage difference separately
The following is an editorial test sequence compiled from the cited manufacturer explanations. Start each comparison from the saved baseline, change only the named condition, and restore it afterward. If a setting cannot be controlled independently, record the coupled change and limit the conclusion.
Allow more time between conversions
Use a documented timing control for the device. NI’s interchannel-delay note distinguishes the conversion clock from the sample rate and describes adjusting AI Convert Rate on supported hardware. The scan must still fit within the sample interval. Record the actual accepted settings; lowering the requested sample rate alone does not establish what changed between adjacent conversions.
If you need to check the overall channel-rate budget, use the separate sample-rate worksheet.
Change which signal precedes the suspect input
Where supported, change the scan list without moving the cables. Keep timing and ranges fixed, and record the predecessor on every pass, including the last-to-first transition. Look for an error change associated with the new predecessor; a changed channel count or automatically changed conversion rate makes this a combined test.
Reduce the voltage step without changing the suspect source
With a controllable source and permitted input levels, bring the preceding signal closer to the suspect signal’s level. Keep the suspect source, range, and wiring unchanged. The digitization guide identifies large adjacent voltage changes as a settling burden; this comparison tests that dependency rather than applying a permanent correction.
| Variable changed | Observation consistent with settling | When the comparison cannot decide |
|---|---|---|
| Actual interchannel interval | Smaller error when more settling time is available | Interval is unknown, or filtering/range also changed |
| Scan predecessor | Error changes with the preceding signal | Actual scan order is unknown or timing also changed |
| Adjacent voltage difference | Smaller error with a smaller step | Suspect source or its reference changed too |
| Source path, in a later controlled test | Improvement with a documented low-impedance source or suitable buffer | Loading, offset, or grounding changed without a reference check |
An unchanged result weakens that particular test’s support for settling. It does not certify the hardware or exclude every other acquisition error.
Avoid fixes that erase the evidence
NI’s ghosting article says a grounded channel before the suspect input is not a universal cure: the voltage step can become larger. It also limits duplicate-channel readings to supported multiplexed X/M Series devices and excludes C Series modules from that feature. Extra conversions count toward the aggregate limit. Do not assume a dummy-channel recipe applies to your unit.
Use the manual’s source-impedance and settling conditions when considering a buffer or different signal conditioner. Do not choose a resistor or amplifier from the appearance of the photographs. The PCIe-6340 guidance concerns its own input path; it cannot establish a limit for an unidentified module.
Keep raw records before adding display averaging or smoothing. Otherwise the comparison no longer uses the same processing. If you change the source, confirm its level against the same reference before attributing an improvement to source impedance.
Keep a test record that supports the next decision
Use one row per comparison. These cells are deliberately empty: they are a place for your observations, not sample results.
| Record item | Your entry |
|---|---|
| Date, run ID, source operating state | |
| Baseline file and restored-baseline file | |
| Single variable changed; any unavoidable coupled changes | |
| Channel list, ranges, actual timing | |
| Reference, measured error, units, analysis interval | |
| Observation repeated after restoring baseline? | |
| Manual section permitting this configuration | |
| Conclusion and unresolved alternative explanation |
If the channel is acceptable alone, the fault returns with the scan, and controlled timing or predecessor changes affect it repeatably, retain those observations as evidence supporting a settling explanation. Check any proposed correction again under the intended operating configuration and acceptance band.
If those dependencies cannot be reproduced, keep the diagnosis open. Return to the signal reference and input configuration, then investigate the source and remaining signal path with the device’s troubleshooting documentation. The next useful deliverable is the configuration plus raw comparisons, rather than a label attached to a screenshot.
Method and sources
Compiled on 16 October 2026 from public NI documentation. The digitization guide supplies the settling explanation and adjacent-step principle. The PCIe-6340 section supplies source-path considerations; the PXIe-4309 manual shows why channel grouping must be checked. The support articles supply the scan symptom, timing controls, and dummy-channel limitations. Original PDFs were fetched during this research; support text was cross-checked with the saved research archive and current web retrieval. The blank worksheets and controlled-comparison sequence are editorial recommendations, not a manufacturer acceptance procedure. This publication did not test a DAQ or establish model-specific performance limits.
