Coffee guy's oversimplified guide to setting your PID - Page 2

Beginner and pro baristas share tips and tricks for making espresso.
JimG
Posts: 659
Joined: 18 years ago

#11: Post by JimG »

cannonfodder wrote:IMHO, the problem with an E61 group and a PID is that as mentioned above, pulling shots in rapid succession will heat that big hunk of brass up. Even if your boiler temps are pancake flat, the group is still overheated and your shots start to cook.
The previous discussion about increasing grouphead temps was limited to Silvia's. That machine has an offset group and no thermosiphon.

The problem does not exist to nearly the same degree on high performance single boiler machines with E-61 groups. The testing I performed a few years back on the Alexia, and the testing I performed more recently on the Alex Duetto, shows that closely spaced shots do not create this overheating problem. Setting up a PID to recover quickly and maintain nearly flat line boiler temps is of value on this type of machine.
cannonfodder wrote:If you want to get real picky, the ambient temperature would need monitored as well.
I have found this to be helpful on my own Alexia. It is not very difficult to do. I have my PID programmed with a 23F offset, which I find is just right at ~75F. When the thermometer on the kitchen wall reads below 72F, I increase the offset by a degree, which bumps up both the boiler and the group temp. Likewise, when 78F or above, I decrease the offset. Really not very difficult or inconvenient.
cannonfodder wrote:I don't see any real benefit to putting a PID on a heat exchanger machine or a machine with an E61 group.
For the case of a high performance single boiler with an E-61 group, I completely disagree. Those machines (Isomac Zaffiro/Amica, Quickmill Alexia, VBM Domobar, etc.) are poster children for the benefits of PID control.

Jim

User avatar
cannonfodder (original poster)
Team HB
Posts: 10479
Joined: 19 years ago

#12: Post by cannonfodder (original poster) »

A PID for a single boiler machine under normal home conditions would make temperature management much better than something like an old school thermostat, I am not disputing that. My thoughts are specifically about a E61 grouped single boiler machine with an un-home like duty cycle. Any machine worth its muster should be able to pull a stable and repeatable temperature profile with shots spaced out every 3-5 minutes. But if you are pushing the limits and running a high volume back to back cycle that group is going to heat up. Eventually you are going to loose the heat sink effect you are getting from that big mass of brass and your temperatures are going to shoot up.

With a duty cycle of say one shot every 60-90 seconds with 8-10 shots you are going to start cooking coffee. How would you go about compensating for that rise beyond the obvious, don't run it that hard answer.
Dave Stephens

rama3i
Posts: 6
Joined: 15 years ago

#13: Post by rama3i »

cannonfodder wrote:A PID for a single boiler machine under normal home conditions would make temperature management much better than something like an old school thermostat, I am not disputing that. My thoughts are specifically about a E61 grouped single boiler machine with an un-home like duty cycle. Any machine worth its muster should be able to pull a stable and repeatable temperature profile with shots spaced out every 3-5 minutes. But if you are pushing the limits and running a high volume back to back cycle that group is going to heat up. Eventually you are going to loose the heat sink effect you are getting from that big mass of brass and your temperatures are going to shoot up.

With a duty cycle of say one shot every 60-90 seconds with 8-10 shots you are going to start cooking coffee. How would you go about compensating for that rise beyond the obvious, don't run it that hard answer.
IMHO, then its time to add another PID, or use PID with multiple input. AFAIK there are lots of Industrial grade PID that has more than one thermocouple input to determine the temperature. The other thermocouple is placed in the group head to force PID to hold the boiler to re heat and avoid overshoot in the group head. The result is more stable group head, and minimize the chance to get overshoot in the high volume environment.

User avatar
Stuggi
Posts: 440
Joined: 17 years ago

#14: Post by Stuggi »

How do you go about adding that input? Since the group head might be at such different temperatures during the span of a session, how do you keep the PID from overcompensating for a cold group? In my mind you would need to just set it as a maximum point, so that the PID only regulates with that input in mind if the group is too hot, not too cold. Still, since I haven't used a PID I might be overthinking this. :D

A couple months ago I had an idea for the ideal machine, a triple boiler one if you will. The idea would be to have a steam, a brew and a group head boiler. The group head would have a saturated loop from the group head boiler, which would be controlled by a PID to achieve optimum temperature, but the probe for it would be in the group head. The heating loop in the group head would have a small circulation pump to keep the heated water flowing, and the group and boiler would have sufficient mass to flatten out the temperature curve as well. Then you would PID the brew water boiler as well, but this water would just enter through a pipe inside the saturated group head and exit the head as fast as possible, much like LM has done it, except that this design could be much more flexible as the group head doesn't have to be physically attached to a boiler. The downside is that you now have two PIDs, three boilers and more plumbing than an average house to deal with, which would add to the cost of the machine making it quite unpractical to make, but it would indeed be a very controlled environment. Maybe one could modify a E61 to accept the forced loop instead of the thermosyphon, and then make a one off, since the only complicated part to make is the group head, all other parts could either be made from scratch or bought off the shelves.
Sebastian "Stuggi" Storholm
LMWDP #136

User avatar
CRCasey
Posts: 689
Joined: 15 years ago

#15: Post by CRCasey »

I guess I have a common question that is never answered, what happens with the controller and the boiler when you select to do an auto tune?

I am guessing that you get to a stabilizing point for the probe, followed by several cycles of ramp up, then a ramp down in temp.

Then you use those delta numbers to figure out a set of PID constants for that boiler/probe system. Is this anywhere near correct?

-Cecil
Black as the devil, hot as hell, pure as an angel, sweet as love-CMdT, LMWDP#244

User avatar
Stuggi
Posts: 440
Joined: 17 years ago

#16: Post by Stuggi »

Yeah, that's about how I've understood it. The PID measures the time it takes for the boiler to warm up, how it responds to heating and cooling and stuff. Basically the crucial measurements should in my mind be how quickly it heats up (heat absorption over time) and how quickly it cools down (heat dissipation over time). Then it calculates the numbers needed to work the P, I and D controls. Differences in water levels and suchlike should be compensated quickly enough by the different controllers that it shouldn't make a difference.
Sebastian "Stuggi" Storholm
LMWDP #136

User avatar
shadowfax
Posts: 3545
Joined: 19 years ago

#17: Post by shadowfax »

CRCasey wrote:I guess I have a common question that is never answered, what happens with the controller and the boiler when you select to do an auto tune?

I am guessing that you get to a stabilizing point for the probe, followed by several cycles of ramp up, then a ramp down in temp.

Then you use those delta numbers to figure out a set of PID constants for that boiler/probe system.
Auto tuning PID parameters is an application of an iterative algorithm that converges to optimum values of P/I/D. I presume it's similar to what you describe other than that I bet it guesses, applies, evaluates and refines each cycle until either it calculates no (or minimal) parameter change or it reaches a limit of iterations and determines that it's not influencing an environment that can converge (like, say, if you forgot to connect the element), in which case it gives up. That being as opposed to iterative observations followed by a single calculation of the parameters.

But I'm just guessing. I am sure you could do it either way. I know from the Fuji PXR-3 manual, though, that the tuning does have a limit of iterations before it bails out.

Of course, if you're auto-tuning the news, it goes like this: 1) click play. 2) laugh.
Nicholas Lundgaard