Manual PID tuning oddities.

[Gameface]

*

 

[BitterSipper]

I'm going to ignore the last page or so of posts and just comment on the topic, because this is something I've struggled with too with a Mypin ta4.



By chance, do you know what the default settings for P & D are? I've autotuned mine several times and a quick google search didn't turn up much of anything.

Speaking of autotune, here is how mine operates after numerous autotunings (the results are often the same, maybe with a half a degree difference or so). Let's say I set up my HLT to heat tap/ground water to 165F. It will work perfectly fine, probably until about 158-160F. At that point it will really slow down and take quite a long time to get up to 165F - maybe 20 minutes. Once it reaches 165F, it will continue to switch on and off until the temp settles 2-2.5 degrees higher, at 167F-167.5F. This isn't so bad to micromanage, but it sure is a nuisance I don't think I should have to deal with. Note: if I want to set the temp such that the PID stops switching my element on completely, I'd have to set it to like 162 or 163F, which I don't understand, but that's just some anecdotal information.

I apologize that I can't add what my settings are right now; I'll add them later when I'm at home and can verify. I am hoping this post will lead someone to say "yep, that's a problem that setting I=1 should fix," heh.

Default settings are 3 for p and 240 for I, what you describe sounds fairly correct with p and I on. The point of PID is to get up to the temp quickly, without way overshooting, and then come down stabilized, if it just cut it off completely it thinks it would dip, so it's trying to softly bring it down. But it hasn't come up with values quick enough for you. If you want it to reach temp quicker you need "less" P (smaller number), but then you may overshoot more.

 

[TungstenBeer]

Default settings are 3 for p and 240 for I, what you describe sounds fairly correct with p and I on. The point of PID is to get up to the temp quickly, without way overshooting, and then come down stabilized, if it just cut it off completely it thinks it would dip, so it's trying to softly bring it down. But it hasn't come up with values quick enough for you. If you want it to reach temp quicker you need "less" P (smaller number), but then you may overshoot more.

How about D?

That makes sense. The thing is, it doesn't get up to temp that quickly (it gets up to temp minus 5 degrees or so quickly, but then it really slows), and it also never "softly brings it down," to use your words. At least not in the course of a 75 minute mash.

 

[BitterSipper]

How about D?

That makes sense. The thing is, it doesn't get up to temp that quickly (it gets up to temp minus 5 degrees or so quickly, but then it really slows), and it also never "softly brings it down," to use your words. At least not in the course of a 75 minute mash.

It's supposed to slow when it's that close to full temp. Try turning off all three P I and D and see what happens. That is straight on off, that will give you an idea of why it's supposed to slow as it gets closer. Just leave D off.

Mine never goes down either, even with only P on and no I, that has been my issue with this controller, it does not process like it should.

 

[TungstenBeer]

It's supposed to slow when it's that close to full temp. Try turning off all three P I and D and see what happens. That is straight on off, that will give you an idea of why it's supposed to slow as it gets closer.

I fully understand why it is supposed to slow down. I can't imagine it needs a full 10-20 minutes to make sure it doesn't overshoot though.

Just leave D off.

Roger.

Mine never goes down either, even with only P on and no I, that has been my issue with this controller, it does not process like it should.

Putting price completely aside, I guess this the overwhelming reason I want to move to Auber units. I've heard a ton of success stories using their units, and much more mixed reviews about anything else.

 

[augiedoggy]

Putting price completely aside, I guess this the overwhelming reason I want to move to Auber units. I've heard a ton of success stories using their units, and much more mixed reviews about anything else.

This is because of the amount of people using the aubers here vs other products, plus auber does have much better support to walk their customers who lets face it often know nothing about pids through every step to get them working correctly for this type of use. When someone pays good money they pay more attention to detail and are also more likely to follow through instead of just blaming it on cheap equipment and buying more expensive replacements when they get frustrated like often happens with cheaper stuff only to find it was something else they did or wired wrong... If I had a nickel for every time someone here replaced a mypin or even rex unit with an auber just to learn their temp probe or ssr was wired wrong..

Mypins support is horrible. But the pids do work well. I just took brumateurs advice when I had the overshooting issue and set the i setting to 1 and it completely resolved my problems.. I have no temp spikes or dips more than a degree or two which last for a few seconds before the pid adjusts .. nothing that would have any actual effect on the brewing process.
There are plenty of auber pid issue threads like the one I linked earlier...

and you will see a lot of people warn others here not to buy the "cheap ebay pt 100 probes" and recommend auber instead even though they come from the same manufacturer and work exactly the same... same with these ssrs being recommended.
http://www.auberins.com/index.php?main_page=product_info&cPath=2_30&products_id=30

instead of these which are from the same manufacturer and come with a heat sink for almost half the price!( Very likely less than half with shipping )
http://www.ebay.com/itm/New-Solid-S...455?pt=LH_DefaultDomain_0&hash=item417e80a597

http://www.ebay.com/itm/New-Solid-S...064?pt=LH_DefaultDomain_0&hash=item35cbbd7228

Bottom line price changes perception... if someone gets it cheap they are more likely to not put as much care into wiring it up right in my opinion.. this can effect somethings reputation...

BTW I see your in buffalo... your more than welcome to check out my TD4 pid setup and see how it /they work if you like?

 

[TungstenBeer]

This is because of the amount of people using the aubers here vs other products, plus auber does have much better support to walk their customers who lets face it often know nothing about pids through every step to get them working correctly for this type of use. When someone pays good money they pay more attention to detail and are also more likely to follow through instead of just blaming it on cheap equipment and buying more expensive replacements when they get frustrated like often happens with cheaper stuff only to find it was something else they did or wired wrong... If I had a nickel for every time someone here replaced a mypin or even rex unit with an auber just to learn their temp probe or ssr was wired wrong..

Mypins support is horrible. But the pids do work well. I just took brumateurs advice when I had the overshooting issue and set the i setting to 1 and it completely resolved my problems.. I have no temp spikes or dips more than a degree or two which last for a few seconds before the pid adjusts .. nothing that would have any actual effect on the brewing process.
There are plenty of auber pid issue threads like the one I linked earlier...

Bottom line price changes perception...

I cut some stuff out that I didn't find relevant. SSRs, temp probes - I agree with you on those, they function the same no matter how much money you spend. It's like buying a $1000 alarm clock... just silly.

In my case, I've used a Mypin TA4 for over a year in a HERMS setup (boil with propane) and I've never been able to set it correctly. A very close friend of mine has had the exact same issue with the same oscillations. After all this time, I may have found a workaround that I have yet to try (setting I=1) but have really only seen posted by you on this forum. It isn't mentioned in the manual, and it's not a common enough product (compared to a TV or car or something) to where there is an absolute abundance of resources online. In my mind that makes for a really poor product; the unit doesn't work as it should, the instructions don't help, and the one or two things that might help are from users that happened to figure it out.

In the meantime I've been able to brew around my issue with relative ease, but it required quite a bit of monitoring and a handful of manual adjustments every mash just to get a stable temperature. THAT alone would be worth an extra $20 or $30 or more just to not have to deal with that headache.

The few brewers I've gotten to know that have gone electric based off Kal's design haven't used MyPins and haven't had these issues. Granted, the sample size of people I'm talking about is very small but it seems very validated by what I see on these forums and on others.

On a somewhat unrelated note - I wonder how much the "fuzzy logic" that some units use have to do with the perceived success rates. I've tried reading up on it; but since I feel like I have a rudimentary understanding ONLY of how a normal PID functions, I certainly can't decipher any of the limited resources I see on the web.

 

[BitterSipper]

Mypins support is horrible. But the pids do work well. I just took brumateurs advice when I had the overshooting issue and set the i setting to 1 and it completely resolved my problems.. I have no temp spikes or dips more than a degree or two which last for a few seconds before the pid adjusts .. nothing that would have any actual effect on the brewing process.
There are plenty of auber pid issue threads like the one I linked earlier...

Could you do an experiment for me. I assume you are heating mash water or heating a HLT? Starting from at least 20 degrees cooler than your set temp start your unit up. Press the "set" button so it displays output. What is it ~10 degrees from your set temp, 5 degrees, at, and above it? Repeat after turning I off (but write it down so you can change it back). I'm very curious if your controller is still outputting like mine, but your dissipation and volume is high enough that it doesn't factor.

 

[schematix]

Can these PIDs run in a thermostat mode? Wondering if anyone has had success holding +/-1F just using that...

I did want to make a few other comments on PIDs though based on some of the things i've read here that are incorrect:

1. P - the more aggressive the gain, the quicker you reach the set point, but more you will overshoot. The effort the controller applies is proportional to the difference between the set temp and the actual temp. For some controllers this is 1/Kp, and for some its Kp. Just know if increasing the value increases the aggressiveness, or lowers it. If you're overshooting, turn your I down low or off, then see how you're overshooting.

2. I - This is the portion of the controller that "learns" the system over time based on the history of the error. A more aggressive I-gain leads to more overshoot (it has to accumulate negative error to counteract the positive error when it was below the set point), but better long term steady state tracking. It sums the error (set - actual) over time, and weights it with the I-gain. Again, depending upon the form of the PID the controller uses, increasing the number can either increase of decrease the aggressiveness.

3. D - This portion of the controller attempts to slow the rate of change. Just turn it off. Sometimes its helpful, but usually its not. PI controllers are the way to go for 99% of applications.

4. Temperature systems are inherently slow, which leads to problems for pure PIDs. The purpose of a controller is control an output to match an input. However, the nature of the input has a huge impact on the controller and the output. In general, a tight controller is needed to hold a system at a set value (this is called disturbance rejection). However, a controller that is changing inputs (called set point tracking; ramping from 145F to 158F to 170F) needs a "soft" controller to prevent overshoot. This is made even worse by the inherent slowness. There are many solutions to this problem. The better solutions use dynamic gains and consider the systems rate of change. Some simple solutions like Auber's "fuzzy logic", uses a different control technique when its close to its set point to prevent the dreaded overshoot associated with set point ramping and the integrator winding up.

This thread is clearly user error. This stuff isn't simple, but its not rocket science either.

 

[augiedoggy]

Keep in mind I use one for control of my rims tube as well so if it overshot that would be a big deal....
I wish I could post videos to show how its working (which is why I offered Tungsten the opportunity to check mine out in person if he wanted since I'm 10 minutes from buffalo...)

in any case I still recommend someone change the i setting to 1 and try it...

 

[schematix]

My Auber controller didn't like my RIMS loop until i put I=1000.... The auto-tune thought it should be about I=50, and it oscillated constantly. Moderate P control, soft I control and no D control should tune all of these loops to a good level of satisfaction. Unfortunately i can't tell you what constitutes moderate, tight or soft without a load of details about your system.

 

[PlexVector]

I found the link below helpful. Although it includes equations the descriptions for each of the PID parameters are decent.

http://www.thorlabs.com/tutorials.cfm?tabID=33653

 

[TungstenBeer]

Keep in mind I use one for control of my rims tube as well so if it overshot that would be a big deal....
I wish I could post videos to show how its working (which is why I offered Tungsten the opportunity to check mine out in person if he wanted since I'm 10 minutes from buffalo...)

in any case I still recommend someone change the i setting to 1 and try it...

I will try that - and I just may take you up on that (when it warms up of course)! First I am going to try I=1 though, because I won't be able to upgrade until late summer 2016, and I certainly want to brew by then :cross:

 

[schematix]

The wikipedia article is a really good resource. Even looks like they might have been copying each other since their diagrams and tables are practically identical.

http://en.wikipedia.org/wiki/PID_controller

 

[schematix]

FYI Turning OFF the integrator is a BAD idea except for debugging. No integrator means you need error before you'll have control output. With the slow system response your performance will suck.

 

[augiedoggy]

FYI Turning OFF the integrator is a BAD idea except for debugging. No integrator means you need error before you'll have control output. With the slow system response your performance will suck.

I set it to one which slows the response down and avoids overshooting. 0 would be off... I suppose I could experiment but the setting I have seems to work very well over the factory settings which did overshoot.

 

[PlexVector]

The wikipedia article is a really good resource. Even looks like they might have been copying each other since their diagrams and tables are practically identical.

http://en.wikipedia.org/wiki/PID_controller

You're right. I like the wikipedia version better. especially the graphs.

I'm running a mypin (4 or 7, but its too cold to go to the garage to check) in a DIY RIMS at 120V. I ran the AT with water before brew day and during the actual mash it tracked quite well. Since I'm running 120v I use propane to get the HLT to temp and the RIMS to keep it at temp during mashing. Part of the system is also the flow rate. I estimated what that would be during AT with water. Also during AT with water the water volume was based on the total mass I expected to have in the MT on brew day. So, not only do you have P,I,&D, but also flow rate and mass to consider for the PID to work properly for a specific set of control constants. I hope I said all that right.

In other words: It depends. ipe:

 

[BitterSipper]

Can these PIDs run in a thermostat mode? Wondering if anyone has had success holding +/-1F just using that...

The mypins do by shutting all three off.

This thread is clearly user error. This stuff isn't simple, but its not rocket science either.

Yea, your attitude is deplorable at best. You cannot be the least bit confident in that and it's just more petty crap.

You're claiming "incorrect information" but you just restated several things I already stated, but in more ambiguous ways. You shouldn't be using "aggressive" to describe P, most people would think you mean "more" when in fact you are trying to describe "less". NO P = off which is the most aggressive you can be. That's why "more" and "less" are better descriptors, and in this case happen to coincide with "higher" and "lower" for the values of most controllers as well.


English language is hard, I know.

 

[augiedoggy]

English language is hard, I know.

It hard enough that the guys at Myin couldnt master it instead our directions sheet comes in engrish... :cross: likely the cause of this thread...Sorry I saw this and thats the first "on topic " thing that popped into my head..

 

[PlexVector]

instead our directions sheet comes in engrish...

Poor engrish at that...

 

[BitterSipper]

It hard enough that the guys at Myin couldnt master it instead our directions sheet comes in engrish... :cross: likely the cause of this thread...Sorry I saw this and thats the first "on topic " thing that popped into my head..

I threw those out. You see my post about looking at your output? Curious since you said you had little overshoot and good control if you have any output while above set point with your normal settings, and with I off.

 

[augiedoggy]

I threw those out. You see my post about looking at your output? Curious since you said you had little overshoot and good control if you have any output while above set point with your normal settings, and with I off.

I will play with it the next time I run pbw through the system since I usually heat the water prior to doing this. And I will report back.

 

[BitterSipper]

I will play with it the next time I run pbw through the system since I usually heat the water prior to doing this. And I will report back.

Thanks, I have a feeling you do but the heat capacity of your system vs element wattage and transfer rate absorb it.

 

[augiedoggy]

Thanks, I have a feeling you do but the heat capacity of your system vs element wattage and transfer rate absorb it.

well I know the output fires a few times after the desired temp is met but at that point the temps are still fluctuating a bit and all in all I would say after about 30 seconds of it firing on and off while the temp is reaches it stops completely... and the output temp never goes more than a degree or two max over the setpoint for any period longer than a second or two... basically after about a minute its holding at the proper temp.
same with my rims... it will climb maybe a degree and flicker on and off 2 degrees higher than setpoint and within 30 seconds or so its settles to the correct temp.
My understanding it the pid operates close to thermostat mode yet still with minimal pid logic buffering things a bit this way... I will admit since its worked fine I havent researched this aspect of my build much.

 

[schematix]

You're claiming "incorrect information" but you just restated several things I already stated, but in more ambiguous ways. You shouldn't be using "aggressive" to describe P, most people would think you mean "more" when in fact you are trying to describe "less". NO P = off which is the most aggressive you can be. That's why "more" and "less" are better descriptors, and in this case happen to coincide with "higher" and "lower" for the values of most controllers as well.

This thread has gotten very contentious...

I have tuned HUNDREDS of PID loops in my career, many of them temperature loops. I know what I am talking about, and your response indicates you don't understand the nuance. In fact your terms are only valid for PID controllers expressed in a particular form, which although common, is not universal.

 

[BitterSipper]

This thread has gotten very contentious...

I have tuned HUNDREDS of PID loops in my career, many of them temperature loops. I know what I am talking about, and your response indicates you don't understand the nuance.

Really, so if I said "aggressive resistance" You would think less resistance? Of course you wouldn't. Proportional gain is a reduction in output, if you call for it to be used aggressively you're calling for more reduction, not less, that is the locution of the term.

If you want to use the word aggressive you need to attach it to temperature delta, not the control.

The only one making this thread contentious is you with conceited comments and claims of "certain user error" without providing anything constructive.

Also, I find it amusing that you are upset at being contradicted when you did the same thing to someone else earlier in the thread when they were using the less ambiguous terms.

In fact your terms are only valid for PID controllers expressed in a particular form, which although common, is not universal.

Again "more" and "less" are not the same thing as "higher" and "lower" English is hard, I know.

 

[schematix]

Proportional gain is a reduction in output

You are spreading false information and I am trying to correct it. This statement is completely dependent upon the controller you have and the form of the PID controller that is used internally.

 

[BitterSipper]

You are spreading false information and I am trying to correct it. This statement is completely dependent upon the controller you have and the form of the PID controller that is used internally.

You still think I'm talking about values, I am not. I am sorry you are missing this. What proportional control does, does not change, though the values required for a certain outcome might.

 

[schematix]

You still think I'm talking about values, I am not. I am sorry you are missing this. What proportional control does, does not change, though the values required for a certain outcome might.

I am not sure how you are misinterpreting my previous comments but you clearly are.

Did you get your controller working yet?

 

[BitterSipper]

I am not sure how you are misinterpreting my previous comments but you clearly are.

I said "proportional gain is a reduction in output" You said "This statement is completely dependent upon the controller"

There is no "misinterpretation" there.

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.