Troubleshooting
Detect a failed analog input channel and report it to SCADA
Short answer
Read the raw input word and the module's per-channel diagnostics before you touch the SCP scaling, then alarm channel health as its own tag that names the channel number. A failed channel then reaches SCADA as a fault with an identity, not as a plausible scaled level, so you know whether to carry a card, a transducer, or wire.
On this page
Read the raw input word and the module's per-channel diagnostics before you touch the SCP instruction. That pair tells you whether the channel, the wiring, or the transducer failed, and the channel status is what lets your logic name the bad channel instead of guessing. Alarm channel health as its own per-channel tag, with the channel number in the alarm text, so SCADA reports a fault with an identity rather than a plausible but dead level. Scaling is the last thing you change, not the first.
Why the scaled level hides the failure #
The SCP instruction maps a raw word into engineering units; it cannot tell whether the word is real. A broken loop at 0.1 mA still scales, into a negative level, and a frozen transmitter scales into a steady number that looks like a quiet wet well. Adjusting the endpoints to make the display reasonable is how a failed channel gets masked for weeks. If the scaled value feeds a PID, settle the raw signal versus scaled value setup in an RSLogix 500 PID as well, because that choice decides what the controller sees during a fault.
What the module already tells you #
- In Studio 5000 Logix Designer, open the analog module's properties and read the status and diagnostic information for each affected channel before rewiring anything. On a 1756 analog input such as an IF61, a bad channel shows up as a channel fault minor fault, and the diagnostic points at the channel.
- Siemens documents that an S7-1200 analog input module with a diagnostic error on any channel reports 32767 or -32768 on that channel even if diagnostics is not enabled. Those two values are never levels: intercept them ahead of the scaling and raise the channel alarm. In TIA Portal, switch the channel's diagnostics on so the error reaches your logic instead of only clamping the value.
- The documented cause tables for analog input messages list missing module load voltage L+ and a dead transmitter supply alongside broken field wiring as the first checks.
- Older cards, such as the 1771 analog inputs, offer only a module-level red FLT lamp. No per-channel bits exist there, so the program's own plausibility check has to name the channel.
If the word moves but not linearly, work through non-linear 4-20 mA readings on an S7-1200 analog input before condemning the card.
Sort a sensor fault from a channel fault #
Record the displayed level, the raw word, the channel status, and the SCP input while the fault is present, then place the symptom:
| What you observe | Where the fault likely is | First action |
|---|---|---|
| Raw word pinned, channel status shows a fault | Channel configuration, module, or field wiring | Follow the diagnostic to the named channel; measure loop current at the input terminals |
| Raw word pinned, no channel fault | Transducer, loop power, an open the module cannot see, or a steady process | Check the level independently; measure current at the terminals and at the transmitter |
| Raw word tracks the process, scaled value is wrong | SCP endpoints, data type, downstream logic | Fix the scaling now, the only case where scaling is the fault |
| Same field wires read correctly on a spare channel | Original channel configuration or hardware | Compare both channel configurations, then treat the original channel as failed |
Never swap channels as the first move: it changes the terminal, the channel configuration, and the program address at once, and the evidence is gone.
Build the per-channel fault alarm #
- Enable per-channel diagnostics in the module configuration so each channel's fault condition reaches controller data as its own point, one tag per channel rather than one for the card. Illustrative BOOLs like
LT101_CH2_Faultsay loop and channel in the name. - Add a plausibility check on the raw word. NAMUR NE 43 puts the usable signal band at 3.8 to 20.5 mA and reserves 3.6 mA and below, or 21 mA and above, for transmitter failure; alarm at the band edges. Convert the mA limits to raw counts using the module's documented endpoints for the configured range.
- On either alarm, stop scaling that word: drive the SCADA point to bad quality rather than forcing 0, and latch an alarm whose text names the channel.
- Debounce for 2 s before asserting and require 2 s of healthy signal to clear, and hold alarms off through module power-up so a settling card does not page anyone.
- Word the alarm for the operator. ANSI/ISA-18.2 alarm management in essence requires one alarm to state one specific condition that needs a response; 'AI card fault' fails that test on an eight-channel analog input card. Name the channel.
- Send it to SCADA the way you would a communication alarm for link loss: a distinct tag per monitored path, with its own severity and its own clear state.
Worked example: 0-10 ft wet-well transducer #
Illustrative numbers, real arithmetic. The transducer is 4-20 mA over 0.0 to 10.0 ft on channel 2 of the level card.
- Healthy: 12.0 mA at the terminals gives (12.0 - 4.0) / (20.0 - 4.0) x 10.0 ft = 8.0 / 16.0 x 10.0 ft = 5.0 ft. No alarm.
- Broken signal wire: the loop collapses to 0.1 mA, which scales to (0.1 - 4.0) / 16.0 x 10.0 ft = -2.4 ft and sits far below the 3.8 mA band edge. Decision: raise the channel fault alarm naming channel 2, mark the level bad quality, and send someone for the wire and transmitter. The mA reading clears the card.
- Card-side failure: the terminal measurement is a normal mid-scale current, but the channel status shows a fault and the raw word is pinned. Decision: the channel, not the field, has failed. Record the configuration, move to a spare channel, and change the program address and drawings together. No transducer work.
Verify the fix end to end #
- Channel status healthy, and the raw word responds to a known change: an independent level check or a signal source within the module's documented input range.
- One hand-derived point matches the SCP output.
- SCADA shows the level and the channel alarm as separate points, and both assert and clear in a forced test.
- After any channel move, configuration, program address, scaling, and drawings change together, or the next fault gets diagnosed against fiction.
One safety note: test with an approved method, a signal source into the terminals or a forced fault in a test program with pump control in hand, never by letting a wet well overflow or run dry.
Questions asked
The raw word is steady and diagnostics are clear. Is the channel healthy?
The channel probably is; the transmitter may not be. A failed transducer can freeze at one current while the loop stays intact, so no diagnostic fires. Check the level independently, and if the process is provably moving while the word is not, treat it as a transmitter failure and alarm on the disagreement. The card needs no attention.
What do I do with 32767 or -32768 on an S7-1200 analog channel?
Treat both as invalid data, never as a level. Siemens documents them as the values a channel reports during a diagnostic error even when diagnostics are not enabled, so intercept them before the scaling instruction, raise the channel alarm, and mark the SCADA point bad quality instead of letting operators read a full-scale level.
How do I test the alarm without a real failure?
Drive the channel with a signal source below 3.8 mA and above 20.5 mA within the module's documented input range, or force the fault bit in a test program. On a running station use your site's approved procedure and keep pump control in hand. The test passes when the alarm asserts, names the channel, and clears on recovery.
My card has no per-channel diagnostics. Can I still name the bad channel?
Yes, from the program. Alarm each channel on its own raw word against the 3.8 to 20.5 mA band converted to counts, and keep the card's module-level fault lamp as the card alarm. The per-channel plausibility alarm identifies the channel; pull the card only when the module fault is also set or several channels misbehave together.