Scace Thermofilter Temperature Device - Page 6

[barry]

I also did that to calibrate the two to each other (i.e., set the one's offset so the delta is zero).

so you got the same relatively flat-lined curve with the tc as with the thermofilter?

--barry "go page 6!"

[BobY]

Part 4

I've all but finished testing the device and overall find the following:

1. The Device is amazingly consistent; it has the ability to repeat results to within less than 0.5 deg F when the exact same procedure is performed repeatedly. To test this, I used the "WBC Procedure for the Measurement of Brewing Water Temperature in Espresso Coffee Machines" authored by Gregory Scace, Barry Jarrett, Bill Crossland and John Sanders. I went through the test procedure four times with differences between the test results as noted above.

2. I measured the flow rate of dozens of shots and, after a thorough clean-out of the device, found the shot-to-shot differences to be nearly unmeasurable. I used 25-second shots and always got 70 ml of water plus/minus (too small to measure).

Note: On this subject just a word of caution. As reported earlier, during initial testing, I found that the flow rate would take sudden jumps and flushing the device's filter did not solve the problem. What did solve it, was a "poke" into the orifice with a 30 AWG wire. But that was just pushing the obstruction back into the space between the filter and the orifice and it soon appeared again.

Finally, taking Greg's advice, I removed the orifice end-cap for a clean out. Before cleaning it I took a good look inside the end cap with a high-power loupe and found tiny pieces of shredded Teflon sheared off from the tape that is used to seal the threads. Apparently they had caught on the micro-burrs on the brass threads and remained captured inside the orifice. They are not easy to remove but with the loupe and a sharp pick, I removed them and carefully re-sealed the assembly. After that, (and dozens of shots later) the problem was gone.

3. I did some comparison testing with a portafilter I assembled some months ago, with an embedded TC whose junction just reaches the very upper-most surface of the coffee puck. I adjusted the grind to produce shots of 70 ml in 25 seconds, ran some comparison plots and although the two brew profiles are not exact duplicates, they are very close - certainly close enough for my purposes and, I would assume, close enough for most professionals to do the analytic work for which this device is intended.

Below are comparisons of two representative plots.

4. I set 198 deg F as an arbitrary temperature target to see how long it would take me, given my personal routine for flushing, tamping, etc., to set up the machine for shot-after-shot consistency (temperatures of less than plus/minus one degree for 12 consecutive shots). It took about 1/2 hour to zero-in on the correct routine.

In short, this device, as claimed, is accurate, easy to use and provides fast results.

BobY

[HB]

so you got the same relatively flat-lined curve with the tc as with the thermofilter?

I adjusted the offset of the two inputs such that T1-T2=0. During the initial two-thirds of the shot, the delta display reading jumped erratically, I presume because the unattached TC was jiggling as the system pressurized. Once the temperature was at its peak, the T1-T2 value hovered around zero.

I did some comparison testing with a portafilter I assembled some months ago, with an embedded TC whose junction just reaches the very upper-most surface of the coffee puck.

Would you post a closeup picture of your TC portafilter without coffee? Judging from the two charts, it looks like yours tracks closer to the thermofilter's results than the over the lip approach.

PS: You had asked me to check the flow rate of the thermofilter. My results are the same as yours, i.e., consistently the same volume. I didn't have any of the clogging issues you reported.

[gscace]

Howdee:

Time to jump in on this business of which measurement is more accurate or whatever. Let's be clear on what we are measuring when we do the surface thermocouple on top of the cake (ST) or the thermofilter thing (TF).

In ST, you are measuring the temperature of water and coffee on top of the coffee cake. The cake starts cool, and warms up as water percolates through it. The percolation rate is approx 30ml / 25 seconds and the pressure is 9 bars.

With TF you are measuring the water temperature in a warmed, insulated environment as it leaves the dispersion screen. The cake facsimile is preheated. the flow rate in TF is identical to that of ST as is the pressure.


Which is more accurate? The answer is that they are both accurate and if you're using the same measuring equipment, let's say uncalibrated type T wire and the same readout device between the two, then they are equally as accurate, but the measurements are of slightly different things. The question is not one of accuracy, but a question of utility. Which of these measurements is more helpful to producing quality coffee. As measures of this we can look at ease of use, ease of reproducibility, uniformity of measurement methodology among different people making measurements, reproducibility of results if different people are making the measurements, elimination of outside events or phenomena that obfuscate the result, etc.

In my opinion, and of course I'm pretty biased here, development of a device that is more sensitive, less colored by outside influences (what happens when you decide to underdose your basket - something that a machine designer can't anticipate and compensate for in his design), extremely reproducible on its own and in the hands of completely separate groups, and is so easy to use completely raises the bar. This device, coupled with the WBC standard enables everyone to talk about the same thing wrt espresso machine temperature anywhere and anytime. A difficult measurement is now completely trivial. Now, machine designers can have better data, roasters can supply the temperature data that they think produces the best taste, espresso techs can set up new equipment more accurately, shops can institute real quality control programs, training new employees is simplified and both experienced and inexperienced baristas can quickly learn and improve machine flushing regimens and learn new machines more easily.

Dunno what more to say. If the water, whose temperature measured by the TF produces one profile, produces a similar but slightly different profile on top of the cake as measured by ST, then what is the problem? The measured boundary condition below the dispersion screen sets up everything else that's happening. And if that bothers you, why would you stop at measuring the temperature at the top of the cake. What about all of those other profiles that you haven't measured? Bury the thermocouple 5mm down inside the cake and you'll get a completely different profile. What is the correct profile there? Or 10mm off to one side and a little lower.

Get the measurement at a spot that makes good sense from an engineering standpoint. Do it in a way that is sensitive and easy to do. Make the procedure universal. Elevate the art of espresso.

-Greg

[HB]

What about all of those other profiles that you haven't measured? Bury the thermocouple 5mm down inside the cake and you'll get a completely different profile. What is the correct profile there?

To give readers a perspective on just how much placement makes a difference, see the chart below:


Cimbali Junior - puck top and bottom temperatures

That's one thing that really drives me batty about temperature discussions. Someone will say temperature X is THEE correct brew temperature for espresso Y. In reality the coffee is subjected to a wide spectrum of temperatures throughout the extraction.

[gscace]

Dan:

Thanks for the plot showing temperature profile distribution. This is exactly to the point.

We are trying to correlate what tastes best to some measureable parameter. For espresso, one obvious parameter is temperature. Since it varies all over the map with time and position, the logical thing to do is to pick a very easy point to measure and make the correlation with taste at that point.

-Greg

[another_jim]

Seems to me there's two real issues and one non-issue with the alternative ways to measure shot temperature:

The non-issue is which is the "right" temperature; barring a 3D temperature map of the entire puck for the duration of the shot, there isn't one.

The first issue is if the different methods are equivalent. Equivalence means that there's a function which will convert one graph to another with enough reliability to be worthwhile. If readings are equivalent, then it's simply a matter of standardizing on the simplest method. Greg's device looks like a front-runner in this regard.

The second issue is the relation between shot taste and precise temperature measurement of any kind.

Here I'm a sceptic. I fully believe in raising the temperature of sour and lowering the temperature of bitter shots, but this is not anything for which one needs accurate thermometry -- one doesn't need to know the absolute temperature, just have some serviceable way of lowering and raising it reliably.

For instance, I don't think taste to temperature relations are repeatable across different kinds of machines. And on the same machine, the hot:cold::bitter:sour relation seems to change for ristretto and lungo shots, although this could be because of the changed temperature curve during the longer ristretto shots.

Reading between the lines of the posts; I think the people using the device have zeroed in on the use that may show the most promise -- finding and fine tuning the flush/wait procedures that get reliable and repeatable temperatures, so one knows exactly how to go a "a little hotter" or "a little colder"

[BobY]

Would you post a closeup picture of your TC portafilter without coffee? Judging from the two charts, it looks like yours tracks closer to the thermofilter's results than the over the lip approach.

Here ya go...



BobY

[gscace]

The second issue is the relation between shot taste and precise temperature measurement of any kind.

Here I'm a sceptic. I fully believe in raising the temperature of sour and lowering the temperature of bitter shots, but this is not anything for which one needs accurate thermometry -- one doesn't need to know the absolute temperature, just have some serviceable way of lowering and raising it reliably.

For instance, I don't think taste to temperature relations are repeatable across different kinds of machines. And on the same machine, the hot:cold::bitter:sour relation seems to change for ristretto and lungo shots, although this could be because of the changed temperature curve during the longer ristretto shots.

But Jim, suppose you are trying to set up multiple machines to produce the same result with the same coffee, as in a coffee service supplying multiple shops. Or suppose your machine breaks and you need a replacement of the same type? Do you believe a fairly crappy pressure gauge whose history is unknown, or do you just get the temperature directly?

WRT different machines and different temps - I'm not so sure I buy into that. Most 58mm pf machines use pfs with very similar dimensions. They use similar, if not identical, pumps. As far as flowrate and pressure buildup goes I'd say that the flowrates are pretty much identical and the buildups may not be. But there are enough similarities that saying temperature is not transferrable is a pretty hard sell. My experience is that it is transferrable.

Now the business of shot volume is very interesting and opens up a good can of worms. As was shown here graphically, spatial temperature profile is gonna be volume flow rate dependent. So if you brew ristrettos you may need something different from lungos. But once again, in order to do it well, you'll be striving for consistency and once again quantifying boundary conditions makes it possible to transfer to other machines more easily.


-Greg

[another_jim]

But Jim, suppose you are trying to set up multiple machines to produce the same result with the same coffee, as in a coffee service supplying multiple shops. Or suppose your machine breaks and you need a replacement of the same type? Do you believe a fairly crappy pressure gauge whose history is unknown, or do you just get the temperature directly?

That's a very good point. For a shop using identical brand machines, I would agree that excellent setup instrumentation is the way to go if one is to have any prayer of getting a high quality product for all.

WRT different machines and different temps - I'm not so sure I buy into that. Most 58mm pf machines use pfs with very similar dimensions. They use similar, if not identical, pumps. As far as flowrate and pressure buildup goes I'd say that the flowrates are pretty much identical and the buildups may not be. But there are enough similarities that saying temperature is not transferrable is a pretty hard sell. My experience is that it is transferrable.

Now the business of shot volume is very interesting and opens up a good can of worms. As was shown here graphically, spatial temperature profile is gonna be volume flow rate dependent. So if you brew ristrettos you may need something different from lungos. But once again, in order to do it well, you'll be striving for consistency and once again quantifying boundary conditions makes it possible to transfer to other machines more easily.

Although Dan found no difference in his blind test; I keep being struck with how differently the same coffees come out when brewed on an E61 and the LM with brew pressures and temperatures set quite closely (using the conventional in-basket TCs). The Peppina and Silvia had other taste profiles again, although here the pressures were not be the same.

I'm beginning to suspect that the early phase of the shot, before there is any flow, may have a large influence on the taste. Hypothesizing way ahead of any data, just to get an investigation going, one would expect a machine that saturates the puck as evenly as possible to do better than one where some parts of the puck are wet for far longer than other sections.

Does anyone have a really crappy machine? One where you've set the temperature and pressure perfectly and still can't anything but swill? One that has a three way valve? It would be interesting to sacrifice some poor coffee, and stop shots 1 second, 2 seconds etc into the shot and examine the puck for wetting.

[barry]

WRT different machines and different temps - I'm not so sure I buy into that. Most 58mm pf machines use pfs with very similar dimensions.

baskets baskets baskets. the brew basket has to be the same, and it's often not.

[MattSimpson]

Well I am very excited to start using this device - I have been waiting for it since I saw Barry showing it off in Seattle! I have placed my order and am in suspense waiting for brown to deliver. - Thanks Greg!

I've been intrigued with the over the lip measurements compared to straight water measurements using a bead junction placed in the water flow.

I once thought I'd reached temp nirvana with my pidded silvia when over lip measurements revealed a flat temp profile for 95 C for a 30 second pull.

as the experiments progressed though I've realized that means a much higher water temp at the beginning of the pull than would be recommended!

I am very eager to start seeing far more accurately the actual temp profiles of silvia, my 3 group linea and Yes Yes Yes, finally, the NS Oscars I use in the coffee bar at church!

In fact, I expect my initial tests will be with the Oscars and that is where I will have the most to gain as those low volume Hxs are likely where flushing routines are the most critical to consistent and proper temps!

I'll be plugging this device into my pico-08 usb TC device - that is a cool unit - Thanks Barry for the tip there... I love using it with my roaster and have written an excel vb program to capture data in real time so that I can record gas and air settings and follow a roast profile curve - all in real time - some day I'll do a write up of that to share...

Matt Simpson

[MattSimpson]

Oh - and any progress on a full or abbreviated version of the WBC standards to post?

[HB]

Hey Matt, I look forward to your feedback on the thermofilter once it arrives. It is definitely a big step forward towards standardization and I'm especially intrigued by the training / tuning aspect.

BTW, Greg posted the instructions and procedures yesterday.

[barry]

have written an excel vb program to capture data in real time so that I can record gas and air settings and follow a roast profile curve - all in real time

oh that's spiffy!

[HB]

Although Dan found no difference in his blind test; I keep being struck with how differently the same coffees come out when brewed on an E61 and the LM with brew pressures and temperatures set quite closely (using the conventional in-basket TCs).

Jim, what blind test are you referring to? In the Cimbali Junior versus La Marzocco Linea group taste comparison, the goal wasn't to discern a difference, but gauge whether one sample was equivalent, better, or superior to another.

Interesting side note: While not mentioned in the report, Peter Guiliano recused himself from testing because of his familiarity with the unique characteristics of Junior's espresso. He developed Counter Culture's Aficionado blend on it and was certain the appearance of the shots alone would be a "tell" for him.

[another_jim]

Jim, what blind test are you referring to? In the Cimbali Junior versus La Marzocco Linea group taste comparison, the goal wasn't to discern a difference, but gauge whether one sample was equivalent, better, or superior to another.

That's the one. I figured if the shots were systematically different, people would have had preferences and you would have mentioned it.

[HB]

Thanks for the clarification, Jim.

That group test was a surprise for me. I really expected the "pros" to pick the shots from their own equipment straight down the line, but they split their votes. One thing a little different about this test than say an SCAA barista competition is that tasters were specifically asked not to split hairs. That may have skewed the results towards more "equivalent" evaluations than if they were required to pick one over another, no matter how small the distinction. The purpose of constructing the evaluation in that fashion was to avoid the case where somebody's buying decision hinged on one-tenth of a point, which in reality means as close to "no difference" as any (non-super) taster can say.

[malachi]

The machine is only a tool and is rarely the weak link.

[HB]

For the benefit of those who've never seen the thermofilter in action, I've created a video as part of my study of "extreme precision flushing" for the Elektra A3 writeup. Intelligentsia sent me some Kid O's Organic Espresso and I've been playing with the temperature ranges to see how it reacts. For the video, the target temperature is 201.5F and the flush timing is for when the machine has recently pulled a shot (link).

I plan to include a table similar to Abe's "lazy man's flush chart" that new owners can use to jumpstart their learning curve. The thermofilter doesn't eliminate the need to learn and pay attention, but rather confirms your intuition. With the added feedback, it enabled me to quickly and confidently advance to a whole new level of HX temperature accuracy. Really, really cool.