PID Settings & Use

[russell]

I posted this on the Coffee Geek Machines Forum but got no response. Since Dan did a PID on his Amica (almost identical to the Eliane) I thought I'd try this forum.

I have now completed the installation of the PID controller on my Eliane and I have a number of questions. Maybe some technical types familiar with these devices could help with some of the questions about settins:


There are 20+ parameters that can be set. None of the discussions I have seen discuss more than a handful (target temp, high limit, low limit, fuzzy logic, and P, i, and d). The controller comes with a thick book that explains how to set the parameters but does not offer a layman's explantion orf what they do. I do not understand, for example, why fuzzy is preferable to Self control. The write-ups suggest that the auto-tune function (which resets P, i, and d) be performed once, but the book seems to indicate that auto-tune has to be rerun every time the target temperature is reset, and as you experiment with taste tests you are surely going to be changing the target temp. (Self--I think--does this automatically).


Is is not clear how relevant all of those parameters, even P, i, and d, are since what is being controlled is the temperature back in (or on) the boiler, whereas what matters is the temp down in the coffee in the basket which is a long way off and affected by the temnperature of the group head (especially an E61) and the portafilter.


Ultimately, considered as a technical problem, no solutionis really ideal because of this separation. We can control the temperature of the boiler water very accurate with the PID, but this is quite indirectly related to the temperature in the coffee in the portafilter. Or we can try measuring the temperature in the group head or even in the portafilter itself, but the only thing we have to work with as far as applying heat is a distant boiler.


In the end, perhaps the only point of the PID is to keep the temperature of the water in the boiler very steady thereby eliminating one of the major variables affecting coffee brewing temp. The principal other factors--group head and portafilter temperature--are controlled by flushing and by timing. Because of the staibility of the E61, this may be relatively stable.


Given all the imprecision resulting from the indirect link, does it really matter whether the thermocouple for a machine like the Eliane is immersed in the water in the boiler, put down the esiting temperature well, or simply screwed to the top (as with the Silvia)?


Two final questions: Can someone suggest a good target starting temp for the water emerging from the group head (so the PID target temp can be set to the value that produces this grouphead temp)?
Can someone technical suggest OK values for P, i, and d that can just be put in once and forgotten (given all the rest of the uncertainty)?

Russell

[HB]

I didn't see your post on CoffeeGeek, but did you read Cliffnotes for PIDing Isomac Amica / Zaffiro? It answers some of the questions you pose (FYI, I link the posts that I find most useful in the resources section of the HB homepage).

The write-ups suggest that the auto-tune function (which resets P, i, and d) be performed once, but the book seems to indicate that auto-tune has to be rerun every time the target temperature is reset...

Once you're auto-tuned, you're good to go. The book's advice was probably thinking of a larger change to the target temperature than a few degrees, which I suppose could affect the PID settings.

Or we can try measuring the temperature in the group head or even in the portafilter itself, but the only thing we have to work with as far as applying heat is a distant boiler.

Andy Schecter is the expert on the fine details; he hangs out on this board, CoffeeGeek, and alt.coffee (and apparently never sleeps). The short answer is the probe placement is some delta of the actual group temperature (e.g., for Silvia it's 229F > 201ish, Amica 219F > 201ish, etc.). Realistically you'll need some sort of in-basket measurement to confirm your setpoint.

For home shot-pulling speeds, the system has enough time to equalize such that the probe placement, while important, isn't hyper critical. The boiler well or side of yours should work well. IIRC, one Amica PIDer used the side and put some insulation overtop to steady the readings. I used the well of the Amica and it worked very nicely as long as you allowed about three minutes between shots.

[russell]

I guess I can follow up with a little exchange regarding Dan's comments. First, I did read your Cliffmotes and I did look at all the links in it. Given the disjoint between the heater action and what occurs at the grouphead, I would go beyond what you say and suggest that it may make absolutely no difference where the probe is placed among a. surface mounted on the boiler, b. in the well, or c. directly in the water. I may test this later by using different probes and comparing the results. I am using a digital thermometer in a plastic basket snug under the group head. One problem encountered is that, perhaps because of the preinfusion of the E61, the temperature starts out about 10 F lower then goes up to the maximum. I did not see this on the Gaggia machine nor on a Silvia.

Second, I doubt if there is a consistent delta in a stock machine because the E61 acts as a heat sink that evens out the temperature variation from the 10 degree F variation in the boiler water temp with the stock thermostat. If there were a constant delta, the water out the group head would also vary by 10 degrees and that is not the case. In fact, this must be one reason a stock Amica/Zaffilo/Eliane gives more consistent results than a stock Silvia or various HX machines (see Coffee Geek review of Amica). However, with a PID controlling boiler temp, maybe you can talk about a constant delta once the machine stabilizes after pulling a shot or steaming.

Third, if the P, i, and d values are more or less constant in this temperature range, why not just suggest a set and put them in directly. The main thing to avoid is overshoot and I think one of the parameters directly deals with this but I don{t understand it.

Finally, I think that a major difference between a stock Silvia and a stock Amica is the temp stability in the Amica resulting from the water circulating through the massive E61 group. Once we PID the Silvia, this advantage may be neutralized. If so, the question is how much the other advantage of the Amica, the preinfusion, affects shot quality. Is the quality improvement if any of a PIDed Amica over a PIDed Silvia striking is it a David Schumer sort of thing, too subtle for people like me_ I intend to test this out, but I would like to hear opinions.

Russell

P.S. Your BB code is driving me crazy''no question marks, etc.

[HB]

At least for the Amica, the boiler well saw considerable temperature swings as soon as the pump started. I think the designers intentionally positioned the well close to the water inlet so the stock thermostat, which has a larger deadband and much slower reaction time than a thermocouple, would kick in sooner. I worked around this post-PID by cutting off the output signal manually during and for a few moments post-shot to allow the system to restabilize. Locating the TC outside would likely introduce enough hysteresis to render this extra step unnecessary, but I didn't try it.

The delta I mentioned is the steady-state difference between wherever the sensor is located and the brewhead. Given an in-basket measurement and some patience, it's not hard to determine. However it's sort of a crapshoot as soon as the pump kicks in, but that's not important as long as you're not worried about rapid-fire espresso.

According to Bob Yellin, the "forgiveness factor" is non-negligible between a PID'd Silvia and an E61-style machine. If you're an espresso-only drinker, a PID'd Amica / Elaine is a good fit.

PS: Sorry about the quotes problem; I think it is corrected in the next version of phpBB. Thanks for your patience...

[russell]

I hope you'll look at my messages on this topic on the Coffee Geek bulletin board. There are just too many people with something to say using their board. I don't want to repeat my messages here, much as I appreciate your website.
Russell

[HB]

Re: PID Settings on CoffeeGeek.

Russell,

Judging from your last post, all seems well in the world again. I discourage cross-posting sites, whether they be HB, CoffeeGeek, alt.coffee, TMC, etc. I get an occasional "would you please look at this thread" request and I'm happy to do so -- way I see it, it's a compliment to be asked one's opinion.

The principle purpose of these forums is to provide a venue for the HB readership to provide feedback on articles, share site news, and try out new ideas (e.g., Marketplace, HB Blogs, Peer Review forum). As much as I love CoffeeGeek, I understand and respect that none of that would be appropriate on Mark's site.