PID Converter
The same controller goes by a different name in every system: gain or proportional band, seconds or minutes, standard, series or parallel form. Choose the system the settings come from and the system they need to go to. The conversion appears right away, with the equation and the points of attention for both systems.
- This is the form most modern systems use, and the one meant everywhere on this site.
Source: ISA standard form, as in the textbooks.
- The I and D actions interact: with D switched on, the effective gain and integral time change too. The form of pneumatic and early electronic controllers.
Source: ExperTune, "Comparison of PID Control Algorithms".
- Ki and Kd are independent of Kp. Lowering only Kp here also changes the balance between P, I and D.
Source: ExperTune, "Comparison of PID Control Algorithms".
- P is a proportional band in percent: P = 100 / Kp. A smaller P therefore means a stronger controller.
- Range: P 0 to 1000 %, I 0.1 to 10,000 s, D 0 to 10,000 s. The control type (PID, PI-D, I-PD) only changes the response to setpoint changes, not the settings.
Source: Yokogawa / Control Station, PID Tuning Guide (csTuner), "Notes on Yokogawa PID Algorithms".
- T1 and T2 are in minutes. The PID block calculates in the ideal (non-interacting) form.
- Equation A, B or C determines whether P and D act on the error or on the measurement; the values of K, T1 and T2 stay the same.
Source: Honeywell, Experion Control Builder Components Theory, §13.8 PID Block.
- The classic TDC controller is interacting. When migrating to Experion, K, T1 and T2 therefore need to be converted as soon as T2 is not zero.
Source: Honeywell, Experion Control Builder Components Theory, §13.1 "Tuning Conversions from Interactive to Non-Interactive PID".
- The effective gain is Gain times the output range divided by the measuring range (NormMV and NormPV). With 0 to 100 % on both sides, Gain is the ordinary Kp.
- TI = 0 switches off the I action, TD = 0 the D action. DiffGain sets the D filter and is not converted.
Source: Siemens, PCS 7 Advanced Process Library V9.0 SP1, function manual (PID algorithm).
- PID_Compact works with the measurement in engineering units and the output in percent. Kp is then percent output per unit of measurement; divide Kp by (measuring range / 100) to get a normalised Kp.
- Setpoint weighting (b, c) and the D filter (a) are not converted.
Source: Siemens, TIA Portal help PID_Compact and STEP 7 CONT_C (FB 41).
- RESET is in seconds per repeat, not in repeats per minute.
- STRUCTURE (P and D on error or measurement, BETA and GAMMA) changes the setpoint response, not the conversion.
Source: Emerson, DeltaV Books Online (PID function block) and "Key Features of the DeltaV PID Function Block".
- The series form is interacting; if you switch FORM to Standard, convert the three values as soon as RATE is not zero.
Source: Emerson, DeltaV Books Online (PID function block, parameter FORM).
- The entered Gain is normalised: internally G = Gain times output range divided by measuring range. Different ranges, different effective gain.
- The D action works on the filtered measurement (Tfil); that filter is not converted.
Source: ABB, Compact Control Builder AC 800M 6.0, Binary and Analog Handling, §4 Controller Algorithms.
- TI and TD are in minutes. The PIDE calculates in velocity form, but the settings are equivalent to the standard form.
- The error is in percent of the measuring range, the output in percent; the gain is therefore normalised.
Source: Rockwell Automation, publication 1756-RM006, Enhanced PID (PIDE), "PIDE Algorithms".
- KI is in 1/min and KD in minutes. Between the two PIDE forms: Kp = Kc, KI = Kc / TI, KD = Kc times TD.
Source: Rockwell Automation, publication 1756-RM006, Enhanced PID (PIDE), "PIDE Algorithms".
- The classic PID instruction (ladder) with dependent gains: Ti and Td in minutes.
Source: Rockwell Automation, publication 1756-RM003, PID instruction.
- Note the difference with the PIDE: here Ki is in 1/s and Kd in seconds, in the PIDE in 1/min and minutes.
Source: Rockwell Automation, publication 1756-RM003, PID instruction.
- This is the form most modern systems use, and the one meant everywhere on this site.
Source: ISA standard form, as in the textbooks.
- The I and D actions interact: with D switched on, the effective gain and integral time change too. The form of pneumatic and early electronic controllers.
Source: ExperTune, "Comparison of PID Control Algorithms".
- Ki and Kd are independent of Kp. Lowering only Kp here also changes the balance between P, I and D.
Source: ExperTune, "Comparison of PID Control Algorithms".
- P is a proportional band in percent: P = 100 / Kp. A smaller P therefore means a stronger controller.
- Range: P 0 to 1000 %, I 0.1 to 10,000 s, D 0 to 10,000 s. The control type (PID, PI-D, I-PD) only changes the response to setpoint changes, not the settings.
Source: Yokogawa / Control Station, PID Tuning Guide (csTuner), "Notes on Yokogawa PID Algorithms".
- T1 and T2 are in minutes. The PID block calculates in the ideal (non-interacting) form.
- Equation A, B or C determines whether P and D act on the error or on the measurement; the values of K, T1 and T2 stay the same.
Source: Honeywell, Experion Control Builder Components Theory, §13.8 PID Block.
- The classic TDC controller is interacting. When migrating to Experion, K, T1 and T2 therefore need to be converted as soon as T2 is not zero.
Source: Honeywell, Experion Control Builder Components Theory, §13.1 "Tuning Conversions from Interactive to Non-Interactive PID".
- The effective gain is Gain times the output range divided by the measuring range (NormMV and NormPV). With 0 to 100 % on both sides, Gain is the ordinary Kp.
- TI = 0 switches off the I action, TD = 0 the D action. DiffGain sets the D filter and is not converted.
Source: Siemens, PCS 7 Advanced Process Library V9.0 SP1, function manual (PID algorithm).
- PID_Compact works with the measurement in engineering units and the output in percent. Kp is then percent output per unit of measurement; divide Kp by (measuring range / 100) to get a normalised Kp.
- Setpoint weighting (b, c) and the D filter (a) are not converted.
Source: Siemens, TIA Portal help PID_Compact and STEP 7 CONT_C (FB 41).
- RESET is in seconds per repeat, not in repeats per minute.
- STRUCTURE (P and D on error or measurement, BETA and GAMMA) changes the setpoint response, not the conversion.
Source: Emerson, DeltaV Books Online (PID function block) and "Key Features of the DeltaV PID Function Block".
- The series form is interacting; if you switch FORM to Standard, convert the three values as soon as RATE is not zero.
Source: Emerson, DeltaV Books Online (PID function block, parameter FORM).
- The entered Gain is normalised: internally G = Gain times output range divided by measuring range. Different ranges, different effective gain.
- The D action works on the filtered measurement (Tfil); that filter is not converted.
Source: ABB, Compact Control Builder AC 800M 6.0, Binary and Analog Handling, §4 Controller Algorithms.
- TI and TD are in minutes. The PIDE calculates in velocity form, but the settings are equivalent to the standard form.
- The error is in percent of the measuring range, the output in percent; the gain is therefore normalised.
Source: Rockwell Automation, publication 1756-RM006, Enhanced PID (PIDE), "PIDE Algorithms".
- KI is in 1/min and KD in minutes. Between the two PIDE forms: Kp = Kc, KI = Kc / TI, KD = Kc times TD.
Source: Rockwell Automation, publication 1756-RM006, Enhanced PID (PIDE), "PIDE Algorithms".
- The classic PID instruction (ladder) with dependent gains: Ti and Td in minutes.
Source: Rockwell Automation, publication 1756-RM003, PID instruction.
- Note the difference with the PIDE: here Ki is in 1/s and Kd in seconds, in the PIDE in 1/min and minutes.
Source: Rockwell Automation, publication 1756-RM003, PID instruction.
Why the same numbers do not do the same thing
Three forms
The standard form (ISA) calculates P, I and D independently and multiplies everything by one gain. In the series form I and D interact: with D switched on, the effective gain becomes larger and the effective integral time longer. In the parallel form I and D are independent of the gain, so a lower gain can actually make the controller more restless.
Units and proportional band
Honeywell and Rockwell work in minutes, Siemens, ABB, Emerson and Yokogawa in seconds. Yokogawa enters the gain as a proportional band: P = 100 divided by Kp, so a smaller P is a more aggressive controller. A factor of 60 or an inverted value is easily overlooked.
What does not move along
The converter converts the three settings, nothing more. The D filter, setpoint weighting, the structure (P and D on the error or on the measurement) and the scaling of measurement and output differ per system and can noticeably change the response. After a migration, check the behaviour on a real step and adjust where needed.
Overview per system
Form, parameters and units as the systems themselves show them. The equations are the transfer function without filter and without setpoint weighting.
| System | Form | Gain | I | D |
|---|---|---|---|---|
| General forms | ||||
| Standard form (ISA, non-interacting) | standard (non-interacting) | Kp | Ti [s] | Td [s] |
| Series form (interacting, classic) | series (interacting) | Kp | Ti [s] | Td [s] |
| Parallel form (independent gains) | parallel (independent) | Kp | Ki [1/s] | Kd [s] |
| Yokogawa | ||||
| CENTUM VP / CS3000 (PID block) | standard (non-interacting) | P (proportional band, %) | I [s] | D [s] |
| Honeywell | ||||
| Experion PKS (PID block, EQA/EQB/EQC) | standard (non-interacting) | K | T1 [min] | T2 [min] |
| TDC 3000 (interacting PID) | series (interacting) | K | T1 [min] | T2 [min] |
| Siemens | ||||
| PCS 7 APL (PIDConL, PIDConR, PIDStepL) | standard (non-interacting) | Gain | TI [s] | TD [s] |
| S7-1200/1500 PID_Compact and S7-300/400 CONT_C | standard (non-interacting) | Kp (GAIN) | Ti (TI) [s] | Td (TD) [s] |
| Emerson | ||||
| DeltaV PID (FORM = Standard) | standard (non-interacting) | GAIN | RESET [s] | RATE [s] |
| DeltaV PID (FORM = Series) | series (interacting) | GAIN | RESET [s] | RATE [s] |
| ABB | ||||
| 800xA / AC 800M (PidCC, PidAdvancedCC, PidSimpleReal) | standard (non-interacting) | Gain | Ti [s] | Td [s] |
| Rockwell Automation | ||||
| ControlLogix PIDE (Dependent gains) | standard (non-interacting) | Kc (PGain) | TI (IGain) [min] | TD (DGain) [min] |
| ControlLogix PIDE (Independent gains) | parallel (independent) | Kp (PGain) | KI (IGain) [1/min] | KD (DGain) [min] |
| Logix PID instruction (Dependent gains) | standard (non-interacting) | Kc | Ti [min] | Td [min] |
| Logix PID instruction (Independent gains) | parallel (independent) | Kp | Ki [1/s] | Kd [s] |
Further reading
- Article
PID controller explained: how it works, settings and process dynamics
How does a PID controller work? Learn what P, I and D do, how process dynamics determine the right settings and how to recognise control behaviour in trend data.
- Article
Process control: the basics of industrial control engineering
How does process control work in the process industry? Learn how feedback, feedforward and process dynamics come together in stable, well-performing control.
Converted is not the same as tuned.
With the right conversion, the controller does the same in the new system as in the old one. Whether that is also the best your process can do, the conversion does not tell you: many settings have been sitting for years at a value that once worked. In an introduction we discuss your control loops, and the free PID Performance Scan then shows where calm can be gained.