Manual PID tuning oddities.

[schematix]

I would like to hear you explain how, when no proportional gain means output is 100% like a thermostat, how applying proportional gain can be anything other than a reduction in output.

Proof you don't understand anything I've said.

Good luck.

 

[BitterSipper]

Proof you don't understand anything I've said.

Good luck.

No I understood what you said, you kept talking about set values. You didn't understand that I was not talking about values despite my explicitly and repeatedly saying otherwise.

Proportional gain is a reduction in output

This statement is completely dependent upon the controller you have and the form of the PID controller that is used internally.

I challenge you again, to explain your claim knowing that I am not talking about specific values. This elusive "proof" I don't understand you when it can only be understood in one way.

 

[muhteeus]

I hope something good comes out of this thread

 

[Sideways]

For those of you who are not sure how the PID settins are expressed in the Mypin let me explain. Maybe this will help out anyone that has a mypin and are trying to get their process dialed in.

With the Mypins the proportional gain is expressed in degrees, not percent like some other controllers. This means that if integral and derivative are both turned off (zero) and proportional gain is set to 5 °F on a cold start your system will fire at 100% output until the process variable (PV) reaches within 5 °F of the setpoint value (SV). Once the temperature (PV) reaches within 5 °F of the SV the output will begin to taper off. For example, once the PV is within 2.5 °F of the setpoint the output will be at 50% and once the PV reaches setpoint the output should be 0%. I think this is where people with mypins are getting hung up. The controller is still outputing after the setpoitn has been reached/exceeded. This simply should not be happening. But I digress.

The problem with only using proportional gain to control at setpoint is that once the setpoint is reached the output is reduced to 0%. On really fast systems with constantly changing PVs that may be OK but on homebrew systems, we need some integral to maintain the setpoint.

Integral is typically expressed in seconds per reset (S/R) or resets per minute (R/M). Mypins express integral in seconds per reset. So if you set your integral to 60 on a mypin you will be capable of achieving 1 reset per 60 seconds. If you set it to 30 you will achieve 1 reset in 30 seconds, or 2 resets per minute. The more resets per minute, the faster your system will be able to react.

Don't worry about Derivative. It likely won't benefit your homebrew system enough to bother dialing in the proper setting. The proper P and I settings will get you where you need to be.

 

[TungstenBeer]

For those of you who are not sure how the PID settins are expressed in the Mypin let me explain. Maybe this will help out anyone that has a mypin and are trying to get their process dialed in.

With the Mypins the proportional gain is expressed in degrees, not percent like some other controllers. This means that if integral and derivative are both turned off (zero) and proportional gain is set to 5 °F on a cold start your system will fire at 100% output until the process variable (PV) reaches within 5 °F of the setpoint value (SV). Once the temperature (PV) reaches within 5 °F of the SV the output will begin to taper off. For example, once the PV is within 2.5 °F of the setpoint the output will be at 50% and once the PV reaches setpoint the output should be 0%. I think this is where people with mypins are getting hung up. The controller is still outputing after the setpoitn has been reached/exceeded. This simply should not be happening. But I digress.

The problem with only using proportional gain to control at setpoint is that once the setpoint is reached the output is reduced to 0%. On really fast systems with constantly changing PVs that may be OK but on homebrew systems, we need some integral to maintain the setpoint.

Integral is typically expressed in seconds per reset (S/R) or resets per minute (R/M). Mypins express integral in seconds per reset. So if you set your integral to 60 on a mypin you will be capable of achieving 1 reset per 60 seconds. If you set it to 30 you will achieve 1 reset in 30 seconds, or 2 resets per minute. The more resets per minute, the faster your system will be able to react.

Don't worry about Derivative. It likely won't benefit your homebrew system enough to bother dialing in the proper setting. The proper P and I settings will get you where you need to be.

This is the most helpful thing I've read so far. Thanks!

 

[BitterSipper]

With the Mypins the proportional gain is expressed in degrees, not percent like some other controllers. This means that if integral and derivative are both turned off (zero) and proportional gain is set to 5 °F on a cold start your system will fire at 100% output until the process variable (PV) reaches within 5 °F of the setpoint value (SV). Once the temperature (PV) reaches within 5 °F of the SV the output will begin to taper off.

With mine at least, this is not the case. The default of 3 begins ramping far earlier than 3 degrees.

[if] The controller is still outputing after the setpoitn has been reached/exceeded. This simply should not be happening. But I digress.

I agree, but mine still outputs over 20% if my memory serves me, after peaking the SV

The problem with only using proportional gain to control at setpoint is that once the setpoint is reached the output is reduced to 0%. On really fast systems with constantly changing PVs that may be OK but on homebrew systems, we need some integral to maintain the setpoint.

Agreed, I only turned I off in order to tune P, to no or minimal overshoot. before applying I.

 

[muhteeus]

By the way. I think the assumptions on P control are wrong. Hear me out.

Some applications are what is know as "integrating" such as filling a tank with water or level control. Most are not. Temperature control should be non-integrating, in that there is a residual output required to maintain temperature. I.E. the steady state is >0% output. This is what integral is for. To bring back down that proportional overshoot.

From the horses mouth. Proportional band is in % not degrees.

 

[Sideways]

[
From the horses mouth. Proportional band is in % not degrees.[/QUOTE]

I stand corrected on how the Mypins express proportional gain. It is not in degrees as I had mistakenly remembered. It is in %. Looking at the manual linked to above the input range for a PT100 on Mypins is -199 to 600 °C (-328 to 1112 °F). So the input range is 1,440 °F. That means that a proportional band setting of 1% is equal to 11.4 °F. The same rules would apply as I stated earlier. The controller will output 100% until it reaches within 11.4 °F of the setpoint and then will taper off. Sorry for the misinformation. I should have fact checked before posting.

 

[rudylyon57]

[
From the horses mouth. Proportional band is in % not degrees.

I stand corrected on how the Mypins express proportional gain. It is not in degrees as I had mistakenly remembered. It is in %. Looking at the manual linked to above the input range for a PT100 on Mypins is -199 to 600 °C (-328 to 1112 °F). So the input range is 1,440 °F. That means that a proportional band setting of 1% is equal to 11.4 °F. The same rules would apply as I stated earlier. The controller will output 100% until it reaches within 11.4 °F of the setpoint and then will taper off. Sorry for the misinformation. I should have fact checked before posting.[/QUOTE]

In the interest of developing some harmony on this post...I think you are both correct about proportional parameters. After all a % of a degree is still a degree!

 

[muhteeus]

I stand corrected on how the Mypins express proportional gain. It is not in degrees as I had mistakenly remembered. It is in %. Looking at the manual linked to above the input range for a PT100 on Mypins is -199 to 600 °C (-328 to 1112 °F). So the input range is 1,440 °F. That means that a proportional band setting of 1% is equal to 11.4 °F. The same rules would apply as I stated earlier. The controller will output 100% until it reaches within 11.4 °F of the setpoint and then will taper off. Sorry for the misinformation. I should have fact checked before posting.

In the interest of developing some harmony on this post...I think you are both correct about proportional parameters. After all a % of a degree is still a degree![/QUOTE]

PID runs off the ERROR. P only: Out % = Kp * (SP-PV). So yes, sortakinda, Kp is unitless gain, and the error is in degrees.

 

[schematix]

PID runs off the ERROR. P only: Out % = Kp * (PV-SP).

Convention is that error is defined as e = SP - PV. Otherwise you'd have negative output unless your gain was also negative.

 

[muhteeus]

Convention is that error is defined as e = SP - PV. Otherwise you'd have negative output unless your gain was also negative.

I apologize. This is correct for heating.

 

[schematix]

I apologize. This is correct for heating.

It is correct for error in general. I will give you that if you error is positive that you may want to do cooling, but now we're getting into a much more complex situation of double acting PID loops.

A heater can't actively cool. You need another device - e.g. a fan or coolant loop to do the cooling if you overshoot. That system has a different response than the heater does so you need another loop. Gets even more fun when you're tryign to cool something that is above the boiling point of your coolant. Been there done that.

 

[augiedoggy]

It is correct for error in general. I will give you that if you error is positive that you may want to do cooling, but now we're getting into a much more complex situation of double acting PID loops.

A heater can't actively cool. You need another device - e.g. a fan or coolant loop to do the cooling if you overshoot. That system has a different response than the heater does so you need another loop. Gets even more fun when you're tryign to cool something that is above the boiling point of your coolant. Been there done that.

like a fermetation chamber

 

[schematix]

like a fermetation chamber

Yep... but most of those at a home brew level are handled with thermostat style control - simple on/off with hysteresis and dead band - Not PID.

 

[BitterSipper]

It is correct for error in general. I will give you that if you error is positive that you may want to do cooling, but now we're getting into a much more complex situation of double acting PID loops.

There are applications of PID control for cooling, so you can't say it's correct in general, but it's definitely the context of this thread. He also said;

This is what integral is for. To bring back down that proportional overshoot.

When if anything (in heating), I is responsible for the overshoot in the first place, but if you translate it as speaking from a cooling perspective to heating, he's saying the job of I is to bring back up the droop [and keep it steady], which is fairly accurate.

So I'm going with either confused or uses PID for a cooling something.

 

[augiedoggy]

Yep... but most of those at a home brew level are handled with thermostat style control - simple on/off with hysteresis and dead band - Not PID.

Yes true, I am aware of this... I have just flashed my first stc 1000+ to replace my stc 1000s and knockoffs...

I cant believe we are on day two of debating and no one has yet bothered to try my suggestion to change the I to 1 and see how it performs... It seems its more fun to argue about the process than to attempt to correct the problem ... Now you have me concerned and I have had no problems (that I know of)

 

[BitterSipper]

I cant believe we are on day two of debating and no one has yet bothered to try my suggestion to change the I to 1 and see how it performs... It seems its more fun to argue about the process than to attempt to correct the problem ... Now you have me concerned and I have had no problems (that I know of)

Wait, what problem are you talking about. The problem that created this thread was that my controller was choosing to heat past the SV even with I and D off, which is contrary to proper function.

I mean I can tell you setting I to 1 effects better control than the default 240, and in fact having I at 2 is even better (for me). But that's not the problem, the problem is this controller is making it impossible to go through the steps to manually tune it. I posted an example of that process in the first post (though I disagree a little with the suggestion on how to come to an initial proportional setting).

I can get the controller to hold around a degree by tuning both at once, but I know I can do better and improve the rise and settling time if it wasn't acting so retarded.

I don't think that by now anyone disagrees that under P only control, there should be no output past SV, yet I have two ta4s that both do it.

 

[muhteeus]

It is correct for error in general. I will give you that if you error is positive that you may want to do cooling, but now we're getting into a much more complex situation of double acting PID loops.

A heater can't actively cool. You need another device - e.g. a fan or coolant loop to do the cooling if you overshoot. That system has a different response than the heater does so you need another loop. Gets even more fun when you're tryign to cool something that is above the boiling point of your coolant. Been there done that.

I will disagree. We use reverse acting to control bubble diameter. This is an exhaust. So the delta output is inversely proportional to the delta error.

http://www.control.com/thread/1254819878 (maybe this is my rslogix5000 notation steering me wrong)

 

[TungstenBeer]

I cant believe we are on day two of debating and no one has yet bothered to try my suggestion to change the I to 1 and see how it performs... It seems its more fun to argue about the process than to attempt to correct the problem ... Now you have me concerned and I have had no problems (that I know of)

Oh, I'll be trying it! I just brew outdoors so the temps we've had make chilling very difficult through my frozen hoses, so it'll be at least a couple weeks until I can try it. And plus I'm a big wuss with the cold, haha.

 

[BitterSipper]

I will disagree.

You gone and done it now, he doesn't like it when you do that.

P.S. I figured it was something like this when you said integral is there to bring down the overshoot.

 

[muhteeus]

You gone and done it now, he doesn't like it when you do that.

P.S. I figured it was something like this when you said integral is there to prevent overshoot.

All about application. Most of the loops I have tuned were inverse .

 

[BitterSipper]

All about application. Most of the loops I have tuned were inverse .

I was really confused for a bit, I was like, this guy seems to know his stuff, wait... what did he say that's completely backw..... ohhhhh.

 

[augiedoggy]

Oh, I'll be trying it! I just brew outdoors so the temps we've had make chilling very difficult through my frozen hoses, so it'll be at least a couple weeks until I can try it. And plus I'm a big wuss with the cold, haha.

Which I why I built my full electric brewery and brew in a spare bedroom Although today we are back in the positive numbers yea!!

Sunday we broke a record set in 1889 with the -7 temps at this time of year. Sometimes I wish the weather was pid controlled...I could go for a little overshooting normal feb temps right now...

 

[muhteeus]

I was really confused for a bit, I was like, this guy seems to know his stuff, wait... what did he say that's completely backw..... ohhhhh.

Haha thank you. I have only been in the biz for 5 years, and sometimes I get pushback and think I am wrong. Glad to hear confirmation I can learn SOMETHING .

Did OP ever get anything solved?

 

[BitterSipper]

Haha thank you. I have only been in the biz for 5 years, and sometimes I get pushback and think I am wrong. Glad to hear confirmation I can learn SOMETHING .

Did OP ever get anything solved?

Op is me, tungsten has a piggyback problem and is trying to steal my thread

(my reply to augiedoggy before explains it)

Wait, what problem are you talking about. The problem that created this thread was that my controller was choosing to heat past the SV even with I and D off, which is contrary to proper function.

I mean I can tell you setting I to 1 effects better control than the default 240, and in fact having I at 2 is even better (for me). But that's not the problem, the problem is this controller is making it impossible to go through the steps to manually tune it. I posted an example of that process in the first post (though I disagree a little with the suggestion on how to come to an initial proportional setting).

I can get the controller to hold around a degree by tuning both at once, but I know I can do better and improve the rise and settling time if it wasn't acting so retarded.

I don't think that by now anyone disagrees that under P only control, there should be no output past SV, yet I have two ta4s that both do it.

 

[schematix]

You can do an inverse PID you just end up with negative gains. It's not wrong as long as you are consistent, but it's not common for your off-the-shelf controllers.

 

[TungstenBeer]

Which I why I built my full electric brewery and brew in a spare bedroom Although today we are back in the positive numbers yea!!

Sunday we broke a record set in 1889 with the -7 temps at this time of year. Sometimes I wish the weather was pid controlled...I could go for a little overshooting normal feb temps right now...

Great call. I'll most likely be building a new house next year and will be designing the brewery into the house layout, but for now... it sucks.

Op is me, tungsten has a piggyback problem and is trying to steal my thread

Haha sorry - it really is the same problem, you've just fine-tuned what your problem is a little further.

 

[BitterSipper]

Haha sorry - it really is the same problem, you've just fine-tuned what your problem is a little further.

Not really, yours is just a problem of low settling rate, which is solved by adjustments. You have P and I on and if the settings are not optimal, what you described will, appropriately, happen. It's likely your problem will be adequately solved by mucking with the integral.

I don't mind you hijacking my thread though, its the two haughty, holier than thou dweebs who don't provide anything constructive while flinging thinly veiled insults that I mind.

 

[augiedoggy]

Great call. I'll most likely be building a new house next year and will be designing the brewery into the house layout, but for now... it sucks.



Haha sorry - it really is the same problem, you've just fine-tuned what your problem is a little further.

last off topic comment (I promise) I'm actually looking to buy another house in the fall and am actually looking for something on the Niagara wine trail so I can grow some of my own hops and eventually (a couple years away) operate a nanobrewery from an outbuilding on my property. Either way I need dedicated brew space!