Temperature Profiling on the Decent Espresso Machines

Beginner and pro baristas share tips and tricks.
User avatar
luca
Team HB

Postby luca » Jul 11, 2019, 5:24 am

I bought the crazy Decent Espresso machine a few months ago and have been having an experience with it that I'm sure many other users also face: being flustered at all of the options. For the moment, I have basically been alternating between a few very different temperature and pressure profiles that give me the flexibility to get shots that I like from different roast levels. I have been getting coffees that I enjoy, but haven't done much in the way of apples-and-apples comparisons changing only one variable at a time in order to isolate the impact of that variable and form a hypothesis that we can test for general application.

On this thread, Jim raised the issue of temperature profiling:

another_jim wrote:However, I'm very surprised that DE1 owners haven't experimented with severely falling temperature profiles, say down to 70C by the end of the shot. If there is gold in temperature profiling, that's the first place I'd look..


Of course, no one can say no to jim, so I said I'd do it and am now starting this as a separate thread so as not to pollute that thread. Jim elaborated on the other thread:

another_jim wrote:I'd be thrilled if you (and anyone else interested) could do comparisons of temperature profiles on the DE1.

There was a lot of discussion in the past, when the Scace first came out, about different kinds of temperature curve. E.g. Is a straight line 92.5 the same as starting at 95 and dropping to 90 by the end of the shot? This was before we were thinking about extraction, and before we learned about the sweet properties of lever machines and their pressure profiles. Based on all this new stuff, I was just thinking by analogy --- if dropping pressure is good; why not dropping temperature too?


I have set up an advanced flow profile to examine this. It is a modification of Rao's blooming espresso flow profile. Basically, it works like this: the first stage is 98C preinfusion at 4ml/s until the puck is saturated. The second stage has a flow rate of zero; it's the soak/"bloom" phase. Here the temp needs to be dropped to try to stop the group heater from bumping the temp up too much. The third stage is a ramp up to 1.7ml/s flow rate. The next couple of stages all maintain 1.7ml/s flow rate for chunks of a few seconds. These are multiple stages that are the same in terms of flow rate outcomes so that we can program successive temperature drops over these. The final thing has a flow rate of zero; its purpose is to tell the machine to idle with the group heater in the temperature from stage 1. Here's a photo:

Image


So the first thing that I wanted to see was how extreme a temperature drop I could get at ordinary flow rates. I'm currently doing a tremendous amount of work to re-learn how to use my Quest roaster, so fortunately I have no shortage of garbage coffee that I can throw at stupid things like this. The first order of business is to get an idea of how big a temp drop can we get across a shot with a sensible flow rate. I asked the machine to give me 98C at the preinfusion stage. It won't let you set less than 80C as a target temperature at the moment (maybe this is the sort of thing John and Ray could give us more range on via firmware update in future, were there demand for it), so I set 80C as the target temp for the remainder of the stages (except the final reset stage).

Here is a temperature curve:

Image


The solid red line shows the temperature profile and the vertical black lines show the end of each stage. So what we can see is that, even with the temp target set to the minimum in stages 2, 3 and 4, the measured temp hangs around near the 98C preinfusion temp for these first stages of the shot. I think that this is the reading at the temp sensor that sits immediately behind the showerscreen and is surrounded by the lower part of the brass dispersion block. As for roaster temp probes, some caution may be necessary in interpreting these results, since I don't know how they map on to what one would measure with a scace device. If someone wants to lend me one (I'm in Melbourne, Australia), I'm happy to look into that. It may be that the temperature that the puck experiences is lower; who knows if the thermal influence of the dispersion block and the cartridge heater drags the temperature up a bit locally at that probe.

With that absolute measured value caveat out of the way, at the end of stage 4, maybe 10mL had flowed out. This will vary depending on the coffee and the grind setting. With a finer grind, less seeps out during the bloom phase. With lighter roasts, more seeps out during the bloom phase. What we can see is that the rest of the shot experienced a temperature drop from about 98C to about 90C. So from phase 1 of this investigation, we can surmise that that is about the biggest repeatable temp drop I can get from the machine ... at that probe location ... using this profile.

The dotted red lines shot the temperature targets. The pale, solid red line in the background shows what I was using for my last shot.

So I'll leave that there as the background homework and try to fit in some actual comparisons this weekend. Of course, we need to remember that there are a lot of different variables, so there are many ways this could go from here.

For what it's worth, I'm currently drinking an americano made with a filter roast kenyan coffee extracted on the above profile. It's delicious; sweet, no grassyness, lots of tomato type aroma/flavour (others say that this is indicative of an underdeveloped roast; in the absence of grassiness, bitterness, astringency, etc, I don't really mind it). I haven't got much experience with this particular coffee, so I have no idea if this result is because the temp profile is good, bad or otherwise.

EDIT: Image showed wrong pressure profile, so I replaced it.
LMWDP #034 | 2011: Q Grader Exam, Brewer's Cup #3, Australian Cup Tasting #1
★ Helpful

cunim

Postby cunim » Jul 11, 2019, 8:26 am

Luca, I think most DE1 owners start out playing with all sorts of pressure/flow profiles, find something they like, and settle on that while they manipulate other things. For example, I use just a few lever profiles (specific for coffee weight and degree of roast). These have given me the best shots I have ever made, so why deviate? Basically, my DE1 is a lever machine that shows me what is going on and that lets me play with prefinfusion times, fill rates, temperatures, etc.

I have not gotten into temperature profiling just because I have not seen the need. However, I have modified temperature settings (as opposed to creating profiles) in ways that I couldn't with my E61. I now tend to use much lower temperatures, for example, and I can easily find out what that is for each coffee. A given lever profile is not optimal with all coffees, or even with one particular coffee as it ages and temperature becomes a way to address that. Recently, for example, my 3-week old decaf started to have a bit of astringency (other flavors remained good) and I found that increasing the temperature from 87' to 89' helped a lot. I have no idea if that is a general rule (increase temp to lower astringency), or if there is just some specific component in that coffee that wants a higher temp than it did originally. However, the DE1 made it easy to make the adjustment

I will be interested to see how this topic develops.

crunchybean

Postby crunchybean » Jul 11, 2019, 9:25 am

Good write up, but you would have an easier time with more precision by controlling temp through pressure rather than temp settings. But personally I do not believe that temp matters when it comes to extraction, temp matters when it comes to flavor. As for the tomato flavor everything is a combo move so I think you can get around it, but I am not sure. First I think it depends on what side of the roast the tomato is coming from? Did the roaster just eeak into the realm of tomato or didn't get enough propulsion to get out of it. Depending on that the barista can overwhelm the extraction or pull back in order to pull out the other flavors adjacent to mask or accentuate the other flavors developed through the roast. Though after saying all of this, I can be wrong.

User avatar
another_jim
Team HB

Postby another_jim » Jul 11, 2019, 1:01 pm

Hi Luca,

I guess thermal storage at the group overrides the ability to get any temperature you like from the manifolds. It's a bit of a relief to find the DE1 is a physical object, not some visitor from the sphere of Platonic espresso machines.

If what I've seen of the Rancilio machines that temperature profile; you might want to set the group temperature at the final temperature, and the water flow at the initial desired temperature. This assumes the water gets to the puck fast, roughly at the initial temperature, then gets influenced more and more by the group temperature as the shot progresses.
Jim Schulman

ira
Supporter ♡

Postby ira » Jul 11, 2019, 2:30 pm

Since the group to the best of my knowledge does not have active cooling and the volume of cold water flowing in is limited to the volume of espresso being made, there is likely a limit to how fast the temperature can be made to drop. I would suggest you try one with one 3 stages, with the last stage being set to the final flow and 80 degrees. I would guess you'll see a similar temperature profile to what you have with the multiple descending steps. If not you'll probably get a much better idea of what can be successfully requested.

Ira

User avatar
luca
Team HB

Postby luca » Jul 11, 2019, 2:41 pm

crunchybean wrote:... But personally I do not believe that temp matters when it comes to extraction, temp matters when it comes to flavor.


another_jim wrote:...If what I've seen of the Rancilio machines that temperature profile...


You both jog my memory that, of course, there has already been some work done on the subject here by Dr Monika Fekete and Craig Simon using the Rancilio machine. The article doesn't report how many shots were pulled on each condition, but at least we get a few box and whisker plots. At a glance, looks like there's a lot of overlap.

Regarding extraction, unfortunately I lent my refractometer to a friend; I should have it back next weekend, so extraction measurement will need to wait. And I should say I'm also not planning on pulling a lot of shots to test statistical significance, at least until we find something worth exploring further.

Anyway, in the article, the doc does say that brew temperature affects flow rate. But the graph that she uses to support this statement is a graph of shot weights, not times ... and the intro to the article says that they used the double shot volumetric to ensure equal amounts of water were added. Am I missing something? I worked with Craig for a while and he's a wonderful guy and the three of us are in Melbourne, so it's easy enough for me to get in touch with them, but I thought I should post here first before asking anything before I go asking stupid questions and totally making a fool of myself. I probably should let them know I've got the crazy machine in case they want to test something or can point out something based on their tasting that we should look into.

crunchybean wrote:Good write up, but you would have an easier time with more precision by controlling temp through pressure rather than temp settings.


(I'm assuming that the second "temp" in this is meant to be "flow".) I'll admit that I created the profile without much thought, and this does raise the issue of dependent variables to which I adverted above. I had a few practical reasons for preferring a flow profile: (1) I've been using it a bit lately, so one is just familiarity; in particular, I like to dial in shots based on the pressure peak height, (2) limiting the flow rate means that I'm guaranteed to get a certain amount of volume and time in each segment, whereas doing it by pressure and time means that the amount of water in each segment (and therefore at each temperature setting) will vary when switching coffees and grind settings and (3) the characteristic of the grinder and the burr set that I've got seems to be that the increase in flow rate (for a pressure profile) or the decrease in pressure (for a flow profile) is higher across a shot than for other grinders that I've used on the DE1P, which exacerbates the time variation I referred to in (2). Now of course pressure and flow are dependent on each other, so I can't affect one without the other, but, to spell out the converse of 2, on a practical level if I pressure profile, then, theoretically at least, I can get different volumes at different temp stages. However, all of this is probably moot on a practical level, since the graphs are showing that the laws of physics still apply and we can't expect to use the profiles to draw temp curves like an etch-a-sketch, so probably where we'll get to is that, like the rancilio machine, we set start temps and end temps and see what we get. But since it's not much work to just pull some shots and see what the graphs look like, I'd like to actually do it. (On this point, though, it's nice to see both the target and the actuals displayed on the DE and it would be interesting to know the critical point regarding the absolute temp values the rancilio machine bands about: yes, but measured at what point? And, further, if it is measured far from the puck and with an offset factored in, does that offset hold true for what the puck actually sees?)

Anyway, one thing we may need to consider is the impact of temperature on flow rate, which may well lead to people making the comment "that's all great, Luca, but you should've adjusted the grind setting to compensate when comparing X and Y". Which sort of is a question that brings us back to what is the purpose of the exercise anyway? For the moment, I'm just poking around to try to get a feel for what might be relevant and what might merit further investigation.

Lots of digressions from me in that bit, as usual, but coming back to the quote, totally happy to give a pressure profile a go. Also totally happy to remove the "bloom" phase. We're going to have to approach this like eating an elephant: one bite at a time. I think the next logical bite is probably looking at what we can do in changing the temp the other way. It looks like we've established that there is a limit to how much temp the system can lose across the shot (at least with the current firmware). What about how much temp the system can gain? So the next step is for me to set stage 1 to 80 C (the minimum) and the remaining relevant stages to 98C (I think that you can actually choose 99C if you want) and see what we get!

another_jim wrote:I guess thermal storage at the group overrides the ability to get any temperature you like from the manifolds. It's a bit of a relief to find the DE1 is a physical object, not some visitor from the sphere of Platonic espresso machines.


Yeah, there's also presumably some amount of water volume held in the tubes from the manifold to the group that also confounds things. I think John mentioned to me in passing that they had selected tubes with a small internal diameter to reduce this effect and give them more control.

another_jim wrote:you might want to set the group temperature at the final temperature, and the water flow at the initial desired temperature. This assumes the water gets to the puck fast, roughly at the initial temperature, then gets influenced more and more by the group temperature as the shot progresses.


Alas, the tablet software - at least as far as I can tell - only allows you to set a target temp for a stage. That's it. The machine controls the rest. I can control the cartridge heater temp independently at the bloom/soak phase only because the flow rate is zero (interesting separate question here; at flow =0, what are the water heating elements doing?). They can control the group head heater separately, so this is another if-there's-demand-they-could-push-a-firmware-update thing and I know John has actually done it for a few things he wanted to check out at MICE earlier this year.

In terms of how the system behaves, I think before there is liquid in there, there's a bit of temperature loss from the heat capacity of ... stuff ... everything; coffee, metal; all of it. (I assume) you can see the influence of that on the probe at the very beginning of the graphs where I request 98C water and the reading starts at 98C and dips to 96C before flattening out at 98C for the remainder of preinfusion. So I think the group heater definitely needs to be on and at some temp hotter than the end temp around the beginning of the shot, at least if we want preinfusion temps to be fairly high.

Interesting question: We see from the temp graph that some volume of water comes out pretty close to the 98C target from stage 1, even though I ordered 80C water from the remainder of the stages ("waiter, this isn't the brew temp I ordered"), then we get the steep decline to 90C. I wonder if the hotter than requested water is the water sitting in the tube between the mixing manifold and the group? I mean, after it has been mixed, it's not like this water can be cooled any more than it cools when sitting in the tubes and apparently the tubes are pretty good insulators (presumably they have to be for the machine to work).

Further speculation: The brass dispersion block, and presumably the cartridge heater above it, have a lot of brew temperature water in and around them (eg. the 98C water I ordered at preinfusion stage), plus whatever air insulation the group itself provides over it. Presumably the fact that the dispersion block and heater are filled with and surrounded by this stuff limits how much temperature they can actually lose.

... anyway, one thing's for sure: there will be plenty more to talk about ...
LMWDP #034 | 2011: Q Grader Exam, Brewer's Cup #3, Australian Cup Tasting #1

User avatar
luca
Team HB

Postby luca » Jul 11, 2019, 2:45 pm

ira wrote:Since the group to the best of my knowledge does not have active cooling and the volume of cold water flowing in is limited to the volume of espresso being made, there is likely a limit to how fast the temperature can be made to drop. I would suggest you try one with one 3 stages, with the last stage being set to the final flow and 80 degrees. I would guess you'll see a similar temperature profile to what you have with the multiple descending steps. If not you'll probably get a much better idea of what can be successfully requested.

Ira


Hi Ira,

I was typing the above reply to the others as you posted this; good to see we had some similar thoughts. Whilst I programmed a lot of stages for flexibility, I basically set preinfusion to 98C and the remainder of the stages all to the minimum of 80C, so, functionally, from a temp point of view, it was basically a two-step profile. So the solid red line in the temp profile above is pretty close to the max temp drop that we can get, at least for the 98C starting temp, and for a 20g dose at a normal flow rate. I'll double check the pressures came out OK, too.

Cheers,
Luca
LMWDP #034 | 2011: Q Grader Exam, Brewer's Cup #3, Australian Cup Tasting #1

crunchybean

Postby crunchybean » Jul 11, 2019, 5:04 pm

@Luca "Am I missing something?", they were probably measuring time to weight amount, aka flow, maybe not taking into consideration what exactly they were extracting more of and how that impacts flavor. And so they affirmed that temp will effect dissolution or dissolvement rates. Which I would agree that increasing temp would, but now we live in a world of flavors and volatile compounds where the emphasis of extraction may necessitate a more exactness than just a "yield of x" and must answer: how, what and why.

But I feel I am missing something:
Where is the flow being measured at? Is it at the top of the puck or the exiting of the portafilter? When I said "temp" the second time, I meant it because pressure will inevitably change the contact of the puck and thus the theorical actual heat flux between grounds and water. Flow will affect the erosion y/n?

Have you established a control?

On a side note I feel an empathy for this experiment, since I am currently testing the effect of my fan speed. Like you are with temp/flow/pressure. In a way it is a difficult situation. Because for me, fan speed is an accompaniment to the overall profile and an alteration in my heat setting would inevitably need an alteration to my fan. What I have seen previously in my roasting that the fan can accentuate a profile but it is not the main driving force and thus needs an established profile to test against. Though very complicated. Since different beans will require different heat profiles and thus a different fan.

User avatar
luca
Team HB

Postby luca » Jul 11, 2019, 6:22 pm

crunchybean wrote:@Luca "Am I missing something?", they were probably measuring time to weight amount, aka flow, maybe not taking into consideration what exactly they were extracting more of and how that impacts flavor. And so they affirmed that temp will effect dissolution or dissolvement rates. Which I would agree that increasing temp would, but now we live in a world of flavors and volatile compounds where the emphasis of extraction may necessitate a more exactness than just a "yield of x" and must answer: how, what and why.


That's sort of not what I was asking. I was saying that Dr. Feteke seemed to use a graph of weight only to support an argument about flow rate. That doesn't make sense to me as a matter of logic. The weight measured in that graph seemed to be the total shot weight, so I don't think that you need to consider individual constituents at that stage. It seems to me that they must also have measured shot times, but they didn't put the shot times in the article.

crunchybean wrote:But I feel I am missing something:
Where is the flow being measured at? Is it at the top of the puck or the exiting of the portafilter?


In Dr Feteke's article, or in the DE? More to the point, does it matter? Is the compressibility of the water and/or the system (eg. teflon tubes) such that the point of measurement of water is likely to introduce measurement error to a meaningful amount? I get that if you measure temperature in, say, a boiler, that's a very different matter from measuring it at the group, but is that also true of pressure?

crunchybean wrote:When I said "temp" the second time, I meant it because pressure will inevitably change the contact of the puck and thus the theorical actual heat flux between grounds and water.


I really don't understand this. It seems like what you wrote then is:

crunchybean wrote:Good write up, but you would have an easier time with more precision by controlling temp through pressure rather than temp settings.


You need to tell the machine (1) what temperature you want and (2) one of either (a) what flow rate you want and/or (b) what brew pressure you want. It's not like, say, we are using a pressurestat to control the temp in a closed boiler and then releasing that boiler pressure to brew with. Are you suggesting a thing I can do with my machine? If so, how do I do it?

crunchybean wrote:Have you established a control?


A control for what? The temp measurements? I wrote that I don't have a scace device and am happy to use one if someone can get one to me.

The actual stuff that I'm going to test? I haven't tested anything yet! I'm just trying to get a feel for what temps ranges I can get out of the machine. When I get around to actually testing some stuff, yes, I'll have some controls. I'll test a flat temp vs some temp profiles on the same pressure/flow profile. But let's be clear on something: at the moment, I'm just poking around trying to get some broad ideas. I'm trying to follow up on Jim's comment that the low hanging fruit here is to test the super extreme temp profiles and see what they are like. I'm not going to get all of the answers, and, like I said, I'm not going to have beautiful, robust data sets with lovely statistical analysis. I'm just going to pull a few shots and report back, then we can take it from there.

Regarding the roaster front - funny you should say that; I'm also doing the same on my roaster at the moment! Yes, it certainly requires a bit of time and being methodical. I've done about 30 roasts and have just started to zero in on some combinations that seem pretty good. I was getting some really worrying variation on my BT final readings, but seem to have brought that somewhat into control with some increased airflow.

I guess I want to make one closing remark: I'm trying to write my posts cautiously, so that I don't over-reach with generalisations that I can't substantiate. It would be easy for readers to think that this level of caution means that I'm not confident in what I'm doing. I've been making espresso for probably 20+ years, I majored in chemistry, I passed the Q, I'm a national cup tasting champion and I have a podium finish at the world brewers' cup. In spite of this, I still don't want to speculate about particular chemical and physical causes and effects and I am extremely sceptical of most of the technobabble laden explanations, regardless of how confidently stated, that I read all over the internets. You can come up with some plausible explanation for just about anything; it's not worth anything unless it's backed up by actual observations, since there could equally be countervailing things that run against it. We drink coffee, not theories, and that's what I'm going to do. I'm the FBI on this one: observe and report!
LMWDP #034 | 2011: Q Grader Exam, Brewer's Cup #3, Australian Cup Tasting #1

pizzaman383
Supporter ♡

Postby pizzaman383 » Jul 11, 2019, 9:14 pm

The physics of temperature change come into play enough that they should be well understood as profiles are built. The rate of change of the temperature at the puck will be most strongly impacted by the flow rate of water to the puck. If you want fast change you will need either fast flow, a big temperature delta, or a combination of both factors. In a group design like the Decent which has low mass you should see more rapid change than other, heavier groups but there will be a time lag from when the mixed temperature change happens to when the new water temperature hits the puck. Lots of short profile phases will likely give the best control but the behavior will need to be worked out by experimentation.

It is easier to add heat to the group more quickly than it is to let it cool down so it might be worth trying rising temperature profiles.
Curtis
LMWDP #551