Skip to Content

SiePortal

The integrated platform for your product selection, buying and support workflow - bringing together Industry Mall and Online Support.

SiePortalDeutsch

Search

Home
Support Support
{{(apiData.Support | filter:{'rank':1}:true)[0].title}} Support
- Menu temporary not available
- Menu temporary not available
- Menue temporary not available
- {{item.title}}

Products & Services
- Menu temporary not available
- Menue temporary not available
- {{item.title}}
Support Support
{{(apiData.Support | filter:{'rank':1}:true)[0].title}} Support
- Menu temporary not available
- Menue temporary not available
- {{item.title}}
MySiePortal MySiePortal
{{(apiData.MySiePortal | filter:{'rank':1}:true)[0].title}} MySiePortal
- Menu temporary not available
- Menue temporary not available
- {{item.title}}
Cart ({{apiData.CartCount}})

Navigation

BUMPLESS PID

Created by: MS10 at: 5/18/2011 7:38 AM (49 Replies)

Rating (12)

Thanks 4

Result pages:: |; 1 | 2 | 3; |

50 Entries

Entries per page: 10 | 20 | all

5/25/2011 9:31 AM	Rate (0)
hdhosseini Diamond Expert Joined: 1/28/2009 Last visit: 9/15/2025 Posts: 6875 Rating: (1368)	MS10 Just came out of a very long meeting and I'm glad that you told me about the ramp function asI just foundout that it is very crucial for ourPID.. Yes, I agree.This function handles any unwanted STEP or jump in your LMN value.I think , this helps the process to follow a ramp input instead of STEPs. MS10 Can I ask you another question on your ramp FB.. i know that 'high' and 'low' refer to the range of the 'input'.. but what about the 'step' and 'factor'? The high and low parameters are only for limiting LMN to low and high limits (e.g 0 and 100) or any seleted value.This part forms a limiter. When input inserted to FB67,the program check if the input and last saved value of input are equal or different.Then we it senses a positive or negative STEP.For example , If buffer is less than in so we have positive STEP and the gap between in and buffer should be filled in each OB35 cycle.(step* factor) is the value we add each time to buffer and compare it with in parameter.step depends on OB35 cycle time which considere 0.1 second and by using factor , you can increase rate of changes(increase clope). buffer:= (buffer+(step* factor)); MS10 also, your video shows that the software responds pretty quick when you modify values.. has this being speed up for the video? cos everytime i tried to monitor values, the code runs very slow.. As you can see, I active the manual mode of controller.So, as I changed the manual value of controller, the value transferred to FB67 and I just checked the response of LMN and analog out value.The FB67 is really a simple function and you can modify it to have a optimum response. I hope this helps regards
	Demo Channel on Youtube
Suggestion To thank Quote Answer

5/26/2011 12:19 AM	Rate (0)
MS10 Posts: 53 Rating: (0)	hdhosseini Yes, I agree.This function handles any unwanted STEP or jump in your LMN value.I think , this helps the process to follow a ramp input instead of STEPs. But i thought if we tune the PID loop properly, the LMN value should not have a sudden big jump? [pls correct me if i'm wrong.. bcos at the moment, without using FB67, I managed to put the gain and Ti value so that when my SP>PV, the CV will increase slowly.. however when PV>SP, the CV jump straight to 0.. hdhosseini As you can see, I active the manual mode of controller.So, as I changed the manual value of controller, the value transferred to FB67 and I just checked the response of LMN and analog out value.The FB67 is really a simple function and you can modify it to have a optimum response. Sorry for my poor explanation.. what I meant is everytime I tried to 'force' value using the VAT and monitor it.. the software seems to response very slowly... once i press 'right click' on the mouse and select 'modify value'.. the screen will freeze for a few seconds before I can monitor the value.. not sure if this is a common thing for everyone.. thanks for your explanation on FB67! will give it a go and update.
MS10 Posts: 53 Rating: (0)
Suggestion To thank Quote Answer

5/26/2011 9:28 PM	Rate (1)
hdhosseini Diamond Expert Joined: 1/28/2009 Last visit: 9/15/2025 Posts: 6875 Rating: (1368)	MS10 But i thought if we tune the PID loop properly, the LMN value should not have a sudden big jump? pls correct me if i'm wrong.. Tunning a controller ensures us that the response of the algorithm to the SETPOINT change or process's disturbance has in range over(under)shoot, rise time and settling time.Sudden Jump in output of a PID controller is due to mode change (auto to manual or vice versa) or gain change in runtime. The topic is discussed in "Controlling with SIMATIC" published by SIEMENS . The reason explained and remedies provided(section 5.2 of book). MS10 bcos at the moment, without using FB67, I managed to put the gain and Ti value so that when my SP>PV, the CV will increase slowly.. however when PV>SP, the CV jump straight to 0.. This is explained in the following topic by dchartier and me. Please note that , You may select either of SP>PV or PV>SP. For example, in a tank which start from 0 to setpoint to be filled is like SP>PV while colling a process from ambient down to a setpoint is like PV>SP. If you select the action type wrongly , the controller may not function properly.If you have a process in which SP>PV so error is positive and PID functions properly.But if you wrongly chose PV>SP, you have always negative value and and PID may generates 0 in LMN. Determine reverse or direct action of PID hope to be helpful hdhosseini
	Demo Channel on Youtube
Suggestion To thank Quote Answer

5/28/2011 10:00 AM	Rate (0)
MS10 Posts: 53 Rating: (0)	Hi Hdhosseini, In my case, I need the CV (LM) value to increase when PV<SP and decrease when PV>SP.. when I keep the mode in Auto and change the SP, the value of CV will increase and decrease slowly.. However when I change the mode to manual, change the SP to be < then PV.. and put it back in auto, the CV always jump to the smallest value (0 in this case).. (as mentioned last time, this does not happen when PV<SP). In one of the PID block, I can't use your FB67 as there is a case where I want the CV to suddenly jump back to mid range once it reaches a certain value(with other code behind it).. so I have to make sure that the PID block is controlling/tuned well.. Pls let me knwo your comment.. thanks, MS
MS10 Posts: 53 Rating: (0)
Suggestion To thank Quote Answer

5/28/2011 11:35 AM	Rate (0)
hdhosseini Diamond Expert Joined: 1/28/2009 Last visit: 9/15/2025 Posts: 6875 Rating: (1368)	MS10 I need the CV (LM) value to increase when PV<SP and decrease when PV>SP.. when I keep the mode in Auto and change the SP, the value of CV will increase and decrease slowly.. Error in PID algorith is calculated by SP-PV.So, if you havePV>SP ,error generated by algorithm is always negative and it may not have proper result. You should select the negative Gain to be multiplied by negative error and lead to a positive value feed to algorithm. Optimum behavior The optimum behavior on a process change or setpoint change varies depending on the application. Two basic requirements are regulation (disturbance rejection – staying at a given setpoint) and command tracking (implementing setpoint changes) – these refer to how well the controlled variable tracks the desired value. Specific criteria for command tracking include rise time and settling time. Some processes must not allow an overshoot of the process variable beyond the setpoint if, for example, this would be unsafe. Other processes must minimize the energy expended in reaching a new setpoint. The speed of response or let say rise time is one of the characters of process and process together. Parameters of process are physical and can not be changed easily, so you have PID parameters to optimize the response of the loop. Please note that by using FB58 and the tuning capability of this function, you may obtain optimum P,I and D parameters. Increasing the Ti make the response slower but it may result smaller steady state error. Increasing Gain may result fast response while overshoot is inevitable. So , it is a trade-off to adjust the parameter to get the best response. Regards
	Demo Channel on Youtube
Suggestion To thank Quote Answer

5/28/2011 12:40 PM	Rate (0)
MS10 Posts: 53 Rating: (0)	Hi Hdhosseini, I hope this doesn't mean that I have to use two different gain value??? (+) gain for SP>PV and (-) gain for PV>SP Cheers. MS
MS10 Posts: 53 Rating: (0)
Suggestion To thank Quote Answer

5/28/2011 1:35 PM	Rate (0)
hdhosseini Diamond Expert Joined: 1/28/2009 Last visit: 9/15/2025 Posts: 6875 Rating: (1368)	MS10 I hope this doesn't mean that I have to use two different gain value??? (+) gain for SP>PV and (-) gain for PV>SP Normal cooling, heating level or other processes require either of reverse-acting control or direct-acting control.It depends on the nature of the process and the desired operating point.For example, controlling level of tank which level varies from 0 to %100 and the operating point selected to be 50.So you need direct- acting control for this case(SP>PV) in most cases and positive gain is OK. To summerize, you select type of acting control and determine the gain.In direct- acting control , reducing LMN when error <0 is natural. If your process is sensitive to jumps in setpoins you can use FB67 in set point branch.So, (+) gain is OK for your case if SP>PV in most cases. Regards
	Demo Channel on Youtube
Suggestion To thank Quote Answer

5/28/2011 1:51 PM	Rate (0)
MS10 Posts: 53 Rating: (0)	Thank you for your reply.. This particular PID block is to control the temperature using a control valve.. so any cases may happen (SP>PV and PV>SP)... not only one out of the two cases.. I have tried tuning the gain using different (+) values but it didnt seem to help.. I can't test my code as I'm at home... but I shall try again and hopefully I manage to find the appropriate gain value to stop the valve to jump to 0 position when PV>SP... MS
MS10 Posts: 53 Rating: (0)
Suggestion To thank Quote Answer

5/28/2011 9:02 PM	Rate (0)
hdhosseini Diamond Expert Joined: 1/28/2009 Last visit: 9/15/2025 Posts: 6875 Rating: (1368)	MS10 This particular PID block is to control the temperature using a con- trol valve.. so any cases may happen (SP>PV and PV>SP)... not only one out of the two cases.. Please check the attachment,Sorry for drawing my own curve for step response of a process( I am afraid of copyright infringement). As shown in the most important part SP>PV and the control valve is opening. This is the time in whicj the overshoothappens PV>SP, so this is the nature of PID controller to make the error zero.The negative error will force the control valve to close. MS10 I have tried tuning the gain using different (+) values but it didnt seem to help.. I can't test my code as I'm at home... but I shall try again and hopefully I manage to find the appropriate gain value to stop the valve to jump to 0 position when PV>SP... I will check you codes again,please provide trends to judge better.Did you modify the codes or they are finalized? regards hdhosseini Attachment step response.JPG (590 Downloads)
	Demo Channel on Youtube
Suggestion To thank Quote Answer

5/29/2011 7:29 AM	Rate (0)
MS10 Posts: 53 Rating: (0)	Hi, Please find attached my modified code. the only difference is now the range of SP and PV_IN is 0-27648.. and I'm not using PV_PER anymore.. I have used gain = 0.01 and Ti = 20000ms (20s). When I'm in AUTO and I changed the SP to be either smaller or larger that the PV.. everything is OK.. the CV ramp up and down slowly.. but when I change to MAN.. then make the SP < PV.. then change back to AUTO, the CV goes to 0 and stays at 0... Thanks. MS
MS10 Posts: 53 Rating: (0)	Last edited by: MS10 at: 6/1/2011 9:02 AM
Suggestion To thank Quote Answer

5/29/2011 7:33 AM	Rate (0)
MS10 Posts: 53 Rating: (0)	Please advise how to do offline trending/simulation.. If not I will provide the snapshot/trending tomorrow. Cheers.
MS10 Posts: 53 Rating: (0)
Suggestion To thank Quote Answer

5/29/2011 7:54 AM	Rate (0)
hdhosseini Diamond Expert Joined: 1/28/2009 Last visit: 9/15/2025 Posts: 6875 Rating: (1368)	MS10 but when I change to MAN.. then make the SP < PV.. then change back to AUTO, the CV goes to 0 and stays at 0... Hi Please check the attachment,I think the trend roughly illustrates your problem.When you change from manual to auto, and the set less tha PV,algorithm close the valve immidately because of negative error.I am sure that algorithm will settle the process value to the setpoint as illustrated in the attachment.This is natue of PID controller.Please prepare some trends of your process.By the way Your running the PID algorithm with a very small gain. For getting optimum value for P,I and D you can use PID SELF TUNER of SIEMENS.Check here. Regards Attachment MAN_AUTO.JPG (644 Downloads)
	Demo Channel on Youtube
Suggestion To thank Quote Answer

5/29/2011 8:11 AM	Rate (0)
MS10 Posts: 53 Rating: (0)	Hi Hdhosseini, thanks for your reply. However i'm really running out of time here.. I'm sorry that I've got not much time leftto read the document on PID self tuner.. do we have to download this from somehwere? Yes, I know my gain is very small... this is because I've tried using a higher value (single digit and also one decimal place value).. both gain cause sudden jump to my CV to 100(max value).. not a steady small increase/ramp... only with a gain of 0.01 then my CV will ramp up and down steadily.. when it is in AUTO.. MS
MS10 Posts: 53 Rating: (0)
Suggestion To thank Quote Answer

5/29/2011 8:32 AM	Rate (0)
hdhosseini Diamond Expert Joined: 1/28/2009 Last visit: 9/15/2025 Posts: 6875 Rating: (1368)	MS10 Yes, I know my gain is very small... this is because I've tried using a higher value (single digit and also one decimal place value).. both gain cause sudden jump to my CV to 100(max value).. not a steady small increase/ramp... only with a gain of 0.01 then my CV will ramp up and down steadily.. when it is in AUTO.. I think I have got your problem, please check this picture.Note that bad combination may cause rapid action in LMN.The value of KOEFF=(GAIN/Ti) should be in a suitable range which is illustrated in the attachment.The combination form the slope of the ramp in step response of PI controller. To test , only to undrestand the cause of proble, select higer gain and also higher Ti .Just to check this is the problem. regards Attachment KOEFF.JPG (578 Downloads)
	Demo Channel on Youtube
Suggestion To thank Quote Answer

5/29/2011 9:08 AM	Rate (0)
MS10 Posts: 53 Rating: (0)	I will try tomorrow and see.. can i confirm that the larger the KOEFF, the more rapid the change in LMN? (brown line on your trend).. so we dont want this but we want below instead: the smaller the KOEFF, the slower the change (red line in your trend)... if yes, then increasing the gain and also the Ti will results in a larger KOEFF than what I already have now? (the gain is 0.01 and the Ti is 20000MS) thanks for your patient answering all my questions.. MS
MS10 Posts: 53 Rating: (0)
Suggestion To thank Quote Answer

5/29/2011 9:25 AM	Rate (0)
hdhosseini Diamond Expert Joined: 1/28/2009 Last visit: 9/15/2025 Posts: 6875 Rating: (1368)	MS10 thanks for your patient answering all my questions.. You are truly welcome,Also check: PID CONT_C FB41 hdhosseini Please check the attachment for step response of a PI controller which is exactly used by SIEMENS in FB41. This is extracted from here and manual introduced in last post. FB41 needs gain(KC in curve) and reset timeTi (T1 in curve). Please note that term 'repeats/minutes' which is widely used in DCS systems stands for result of (gain/Ti) or (KC/T1).With this factor (as shown in attachment), the integral mode tries to repeat proportional correction (h(t)) by the end of first integral time.The h(t) is added by gain if error is consatnt and each integration time passes. Note, the shorter the integration time more propertional correction is repeated (more repeats / minutes). Regards
	Last edited by: hdhosseini at: 5/29/2011 9:26 AM Demo Channel on Youtube
Suggestion To thank Quote Answer

5/30/2011 2:23 AM	Rate (0)
MS10 Posts: 53 Rating: (0)	Hi Hdhosseini, I've increased the gain and also and Ti and it doesnt seem to help with my PID block.. the LM just goes up to the maximum value or jump to the minimum value. I have just noticed another thing.. When SP>PV.. and i put it in AUTO.. the LMN will increase steadily.. then I put it in MANUAL (LMN is the same as the last AUTO value)... but when I put it back to AUTO.. the LMN value seems to always jump back to either theLMN_P or the sInvAlt.. I have tried this with different setpoint and see the same pattern in all cases.. let's say in AUTO the LMN increase up to 100 (max.) then I change it to MANUAL (LMN stays at 100).. then I change it back to AUTO.----> the LMN always jumps back to the LMN_P or sInvAlt then start increasing from there again.. Please help.. I started to get quite desperate here cos I've only got today to fix this problem.. This may have some relation with the other error that I have(LMN goes to 0 when PV>SP).. Thanks... Please see the attached screenshot of the DB.. note that the value of the LMN on the screenshot is not exactly = to LMN_P as it depends on when I press the 'screenshot' button on my keyboard.. but I am sure it always jump back to there values.. Attachment pid screenshot.pdf (578 Downloads)
MS10 Posts: 53 Rating: (0)
Suggestion To thank Quote Answer

5/30/2011 6:11 AM	Rate (0)
hdhosseini Diamond Expert Joined: 1/28/2009 Last visit: 9/15/2025 Posts: 6875 Rating: (1368)	FB41 MANUAL Manual Value It is possible to switch over between a manual and an automatic mode. In the manual mode, the manipulated variable is corrected to a manually selected value. The integrator (INT) is set internally to LMN - LMN_P - DISV and the derivative unit (DIF) to 0 and matched internally. This means that a switchover to the automatic mode does not cause any sudden change in the manipulated value. The problem is internally rectified.Attach your final project to trace the main cause. regards
	Demo Channel on Youtube
Suggestion To thank Quote Answer

5/30/2011 6:33 AM	Rate (0)
MS10 Posts: 53 Rating: (0)	Hi, I have checked the other PID loop that I have on the code and they are all behaving similarly... Once we move from MAN back to AUTO ---> LMN always move to LMN_P (and i found out that LMN_P = error * gain) what is this LMN_P?? Attached is the latest code.. it has not changed since the weekend.. I know now why my gain value is very small... is it because my PV and SP is set to 0-27648? thanks. MS Attachment Pid_test (2).zip (565 Downloads)
MS10 Posts: 53 Rating: (0)
Suggestion To thank Quote Answer

5/30/2011 7:16 AM	Rate (0)
hdhosseini Diamond Expert Joined: 1/28/2009 Last visit: 9/15/2025 Posts: 6875 Rating: (1368)	MS10 (and i found out that LMN_P = error * gain) what is this LMN_P?? It is P part of PID controller's output MS10 I know now why my gain value is very small... is it because my PV and SP is set to 0-27648? When use peripherl inputs, you should be careful of correct scaling.You should set SP in physical range . Important: The value PV_FAC =1 for PV of 0 to 100.If you have different low and high value for Temperature transmitter , you can change PV_FAC and PV_OFF.Check PV output variable to be sure that the scaling is done correctly. I think scaling is your problem.Ask the instrument technician about the high and low value of transmitter.It is crucial.(Check the attachment) regards Attachment pid.jpg (561 Downloads)
	Demo Channel on Youtube
Suggestion To thank Quote Answer

|

Share this page:

Share this page on...

Related topics by keywords:

Follow us on