Scace Thermofilter Temperature Device - Page 5

[barry]

In practice, the two methods produce different results. Temp variation within a shot is different, and also average brew temperature. Just as an example, in a few comparisons I did, thermofilter reported temp variation from 10th second on which was twice as large as one measured with coffee. Highest temp reached in the shot on dual boiler was also higher by ~ 1 f.

i'm still not clear about what you're trying to measure. if you know, and accept, that coffee grounds will attenuate the indicated brew temperature, then i don't understand why you have resistance to the data generated by the thermofilter. think of measuring in the coffee as watching a movie (or sunset, or whatever) with sunglasses on; think of measuring with the thermofilter as watching the same thing with sunglasses removed. or think of plato's allegory of the cave....

[Abe Carmeli]

It would be a welcome bonus if a coffee puck simulation accurately modelled a thermodynamic reality, but I would still taste the coffee before I'd believe it.

Yes Maan. I'm widja. Here's the thing: I think with experimentation, one can come up with a formula that will give us that value, without changing anything with the Thermofilter. It will take some time to do all those tests, but it is doable. Greg may be able to run some numbers and come up with an approximation of what the formula would look like. Say: If average thermofilter reading is 203, brew temp with coffee will be 202. If variation within shot is 2 f, with coffee it is 1f, etc.

[Abe Carmeli]

i'm still not clear about what you're trying to measure. if you know, and accept, that coffee grounds will attenuate the indicated brew temperature, then i don't understand why you have resistance to the data generated by the thermofilter.

What I want to know is by how much it will attenuate. I don't want to live in the Matrix. Let's get the simulation as close as possible to a real shot.

[barry]

What I want to know is by how much it will attenuate. I don't want to live in the Matrix. Let's get the simulation as close as possible to a real shot.

then you need to know the temperature of the coffee grounds for each shot and the temperature gradient of those grounds from top to bottom and from outside to inside. further, you need to know the amount of heat radiating from the showerscreen, and from the portafilter & basket. you need to know the space between the bottom of the showerscreen and the top of the grounds. you need to know the time between pf insertion and the start of brewing. you'll need to account for heat loss of the wave front as it travels through the puck, and you'll have to account for soluble and non-soluble solids transfer to the effluent.

if you want to know how much the coffee attenuates the brew water, then just measure the temp of the effluent at the underside of the brew basket. this is what some manufacturers do and then claim "stability". it's pretty meaningless.

[Abe Carmeli]

What I want to know is by how much it will attenuate. I don't want to live in the Matrix. Let's get the simulation as close as possible to a real shot.

then you need to know the temperature of the coffee grounds for each shot and the temperature gradient of those grounds from top to bottom and from outside to inside. further, you need to know the amount of heat radiating from the shower screen, and from the portafilter & basket. you need to know the space between the bottom of the shower screen and the top of the grounds. you need to know the time between pf insertion and the start of brewing. you'll need to account for heat loss of the wave front as it travels through the puck, and you'll have to account for soluble and non-soluble solids transfer to the effluent.

if you want to know how much the coffee attenuates the brew water, then just measure the temp of the effluent at the underside of the brew basket. this is what some manufacturers do and then claim "stability". it's pretty meaningless.

I don't think I need to know all that. There is no need to make it more complicated than it is. A simple test logging & comparing 50 shots of the Thermofilter against 50 of real coffee with the T/C in it on a PID'ed machine will give me a fair idea of the deviation between the two. All I want is a formula, I'm not sending it to space.

[barry]

A simple test logging & comparing 50 shots of the Thermofilter against 50 of real coffee with the T/C in it on a PID'ed machine will give me a fair idea of the deviation between the two. All I want is a formula, I'm not sending it to space.

and what does this tell you that a thermofilter measurement alone doesn't? are you just looking for an adjustment factor you can apply to a tc/coffee measurement so you don't have to use a thermofilter? "hey everybody, don't bother with the thermofilter, just take your plain ol' thermocouple reading and multiply by 5, divide by 9, and add 32, and you'll get the same results!"


--barry "if it was that easy, don't you think we would have done that?"

[Abe Carmeli]

are you just looking for an adjustment factor you can apply to a tc/coffee measurement so you don't have to use a thermofilter?

It's the other way around mister. I'm trying to get rid of the Pain in the Yass T/C in the coffee puck and use the Thermofilter instead.

<< Edited to be less obnoxious >>

[barry]

It's the other way around mister. I'm trying to get rid of the Pain in the Yass T/C in the coffee puck and use the Thermofilter instead. I'm wondering how many more times I'll need to repeat this.

it's hard to repeat something you haven't said.


if you want to use the thermofilter instead, then why do you need to know how the results from it translate to the other method? just use the thermofilter. easy.

i'm not trying to be difficult, i'm just trying to figure out what information you're really after and whether it is useful information and why (or why not).

[Abe Carmeli]

then you need to know the temperature of the coffee grounds for each shot and the temperature gradient of those grounds from top to bottom and from outside to inside. further, you need to know the amount of heat radiating from the shower screen, and from the portafilter & basket. you need to know the space between the bottom of the shower screen and the top of the grounds. you need to know the time between pf insertion and the start of brewing. you'll need to account for heat loss of the wave front as it travels through the puck, and you'll have to account for soluble and non-soluble solids transfer to the effluent.

I don't think I need to know all that. A simple test logging & comparing 50 shots of the Thermofilter against 50 of real coffee with the T/C in it on a PID'ed machine will give me a fair idea of the deviation between the two. All I want is a formula.

[barry]

We are experiencing a classic example of a failure to communicate here. I will take my leave now.

i just went back and reread the entire thread. the only point i can see which might have lead to the confusion was when you used the phrase "measuring a real shot". i took that as you wanted the thermofilter to mimic the thermal performance of pulling a shot through coffee. what i think you might have meant, and what i didn't get until now, was you want the results from the thermofilter readings to better duplicate the results you get from using a tc on top of the puck during a real shot.

if that is the case, then once again i have to ask, why? why do you want the thermofilter to mimic the results of a method which has known flaws? sure, the thermofilter isn't a perfect measurement tool, but it's better than the tc on top of the puck. so, why do you want muted data? sure, a .8F variation is better than a 1.6F variation, but only if it really happens and isn't just an artifact of the measurement method. if you get .8F with a cold portafilter and 1.6F with a hot one, then the .8F doesn't really happen, it's an artifact of flawed methodology. if you get .8F variance using a tc on top of the puck, and 1.6F with the thermofilter, then the tc results are due to flaws in the procedure. if it was the other way 'round, and you got 1.6F on the puck, and .8F with the thermofilter, then i would be worried about the thermofilter performance.

to give you another example, i used to use a 1/4" probe in my roaster as a bean temp probe. i logged lots of data and looked at lots of curves. one thing which always bugged me was the turning point on my roaster (the shift from decreasing bean temp to increasing bean temp) was well over 2 minutes, when most experts i'd talked with said the turning point should occur around 90 seconds. hmmm... i was convinced my roaster was radically underpowered. a while back, i put in a couple of more probes, and decided all the probes in the roaster should be of the same type/size, so i replaced the 1/4" probe with a 1/8" probe. guess what? my turning point shifted to just over 90 seconds. so, did my roaster suddenly gain power? no, of course not. the thermal mass of the probe decreased, so the system lag was reduced. if i stuck a bare wire bead probe in there, the turning point might change yet again. so, the problem wasn't with the roaster, but with the measurement system.

so, back to espresso. we know the coffee grounds suck heat from the brew water. we know that how much suckage occurs depends upon a number of factors i've previously listed. if we're trying to measure/track/log machine brew water thermal performance, then minimizing/reducing/eliminating that suckage will give more accurate results; it will better reflect what the coffee encounters initially, prior to the suckage. if your suckage rate is relatively constant, then knowing the initial conditions is what is needed. if you want to know how much suckage occurs, then put a probe on top of the puck and at the bottom of the puck and measure away! (but it still won't tell you how well your machine performs)

[Abe Carmeli]

i just went back and reread the entire thread. the only point i can see which might have lead to the confusion was when you used the phrase "measuring a real shot". i took that as you wanted the thermofilter to mimic the thermal performance of pulling a shot through coffee. what i think you might have meant, and what i didn't get until now, was you want the results from the thermofilter readings to better duplicate the results you get from using a tc on top of the puck during a real shot.

We finally understand each other, and thanks for taking the pains to re-read that long thread. This is my take on it: The T/C over a coffee puck is flawed indeed, because of the flimsy placement in the puck, mostly. However, it does have at least one advantage over the simulated shot with the thermofilter: The coffee puck is cold. I suspect that the narrow range in variation in temperature within a shot is not due to faulty reading by the T/C, but mostly due to that difference. In that regard, it is reporting a more accurate behavior during a shot than the Thermofilter.

Now, how does this help me at all? In what way is the T/C reading better for my espresso? It all boils down to what I'm trying to measure. If I use a dual boiler machine for example & set temp tp 202f, and test it, the T/C method is likely to be more accurate in showing real brew temperarture. I will need to do It a few times and average the results because of the possibility of error in measurement, but it will be closer to real brew temperature.

The argument that I hear from you, and Dan, is that it does not really matter. What you are testing here is consistency. The scale can be adjusted to whatever you want. It is true, but only if everyone uses this device. For the greater community that still uses the T/C method, there is no standard to communicate with. Having a way to translate one to the other, keeps everyone happy. Hence my quest for the formula. When a roaster tell you that the optimal brew temperature for his/her beans is 203 degrees, we want all to be on the same page.

[barry]

a dual boiler machine for example & set temp tp 202f, and test it, the T/C method is likely to be more accurate in showing real brew temperarture.

no, it's not, for all the reasons previously mentioned. too many variables, the least of which is the tc placement.

It is true, but only if everyone uses this device. For the greater community that still uses the T/C method, there is no standard to communicate with. Having a way to translate one to the other, keeps everyone happy. Hence my quest for the formula. When a roaster tell you that the optimal brew temperature for his/her beans is 203 degrees, we want all to be on the same page.

unfortunately, without a standardized test procedure and equipment, the numbers don't necessarily translate. that was the major force behind the development of the thermofilter. numbers from my tests with my tcs are just the vaguest relative reference points for others. same with roaster temps, btw. just because i drop dsb at 438F doesn't mean that's the "correct" temperature for every other roaster. it might not even be an accurate temperature because of cumulative systemic inaccuracies. another issue is differences between machines, both machine types as well as unit to unit variations.

and, in the end, the "correct" temperature is that at which the coffee tastes the best, and it wouldn't surprise me if that varied from machine to machine (especially type to type, and model to model).

if i tell you that i like DSB at 195F, and you do your tc/puck measurement and come close, then decide that you like it better at 197F, then neither of us is wrong or inaccurate.

[HB]

This is a very good discussion and now I think we're understanding each other's viewpoint (perhaps not agreeing, but understanding). To give a concrete example of the effects under discussion, tonight I did a series of thermofilter versus over-the-lip comparisons:

Thermofilter setup: Warmed up in the group for over an hour. Pulled a series of shots with various intervals from twenty seconds to several minutes as part of my flushing fine tuning study. I noted the maximum temperature for each shot. When I was satisfied that my technique would give me a variance shot-to-shot of less than one degree, I recorded the next two pulls. They were essentially identical, the plot above is the first one of two recorded.

Over the lip (TC) setup: Getting reliable data proved tricky for a couple reasons. First, I needed to place the TC wire across the puck and lock in very quickly, since the A3 requires a "flush and go" approach for optimal temperature stability (i.e., the rebound time is around five seconds tops). Consequently the initial spike is likely higher than in normal usage. In addition, to speed up the placement without burning my fingers, I used cold baskets, which would further perturb the natural shape of the curve.

The latter setup problems do once again underscore one point we all agree on: It's a lot easier and more repeatable to use a fixed position TC. If I were planning a long series of measurements, I would instead route the thin TC wire under the gasket and wrap it around the dispersion screw. My apologies, I didn't think to do that before the machine was very hot.

Interpretation: With the caveat above in mind, note the red line shows the traditional declining curve one would expect to see for an HX machine. The thermofilter starts out already heated, hence its high initial temperature of 190.7F. The effect of the cold basket and cool coffee is pretty obvious since the starting temperature is 176.7F; generally the first measurement using hot baskets is closer to 185F mere seconds after the portafilter is locked in. Despite the less than ideal conditions of my test, Abe's point is plainly evident: The true shape of the temperature profile is quite different than that depicted by the thermofilter. Barry would counter that it's intentionally so since the device's purpose is to measure the boiler's performance, not the system dynamic that includes unpredictable elements like genuine coffee pucks.

Speaking of standards: Greg sent me a copy of the thermofilter instructions and the WBC protocol for its use. I converted it to post-able format and returned it to him for his verification. He should be posting it shortly so we'll all be able to work from the same instructions.

[Abe Carmeli]

Interpretation: With the caveat above in mind, note the red line shows the traditional declining curve one would expect to see for an HX machine. The thermofilter starts out already heated, hence its high initial temperature of 190.7F. The effect of the cold basket and cool coffee is pretty obvious since the starting temperature is 176.7F; generally the first measurement for hot baskets is closer to 185F mere seconds after the portafilter is locked in. Despite the less than ideal conditions of my test, Abe's point is plainly evident: The true shape of the temperature profile is quite different than that depicted by the thermofilter. Barry would counter that it's intentionally so since the device's purpose is to measure the boiler's performance, not the system dynamic that includes unpredictable elements like genuine coffee pucks.

This is an excellent illustration of the principles at play. Nice graph Dan.

[barry]

Abe's point is plainly evident: The true shape of the temperature profile is quite different than that depicted by the thermofilter. Barry would counter that it's intentionally so since the device's purpose is to measure the boiler's performance, not the system dynamic that includes unpredictable elements like genuine coffee pucks.

nope. read your comments about the curves and it might jump out at you.

"note the red line shows the traditional declining curve one would expect to see for an HX machine."

Q: why is this the "traditional" curve, and why would you expect it?
A: possibly because it is a curve generated with <however mildly> flawed methods, and you expect to see it because you've seen it so many times before.

my counter is that it is entirely possible that the actual brew water profile is better represented by the upper curve, and the "traditional" curve is a result of measurement artifacts.

in other words, perhaps we need to re-think our assumptions about machine performance.

--barry "stepping out of the cave is never easy"

[HB]

Barry, you may well be right. My assumption is that HX machines have an initial spike that I nickname the "HX hump". But that assumption relies on intuition and measurement techniques that could be fatally flawed. Nonetheless, the two curves above are reproducible at will, so whatever the cause may be, there is a difference afoot. The next question is then: Should I care?

By the way, if anyone is interested... I did sip a teenie bit of the last shot. It was good.

[barry]

Nonetheless, the two curves above are reproducible at will, so whatever the cause may be, there is a difference afoot. The next question is then: Should I care?

yes, if the flaw is repeated, then the curve will be reproducible. and yes, you should care, because understanding of how the machine really works depends upon it. just think, the hx guys might have been able to do schomeresque quality flat-line brew temps all along.

btw, here's a test for you: put the tc over the rim of the thermofilter!


--barry "naaahhhh... couldn't be!"

[Abe Carmeli]

btw, here's a test for you: put the tc over the rim of the thermofilter!

--barry "naaahhhh... couldn't be!"

I actually did that a week ago. The T/C over the lip reported the same temperature as the Thermofilter probe. The problem is not in the T/C reading but in the difference that the coffee puck makes, as I have already stated. Back to my original point.

This may raise a different question: should we measure brew temperature with or without coffee?

[HB]

I also did that to calibrate the two to each other (i.e., set the one's offset so the delta is zero). Nothing usual jumped out at me. OK, I'll guess that Barry's referring to the jumpiness of the TC's initial readings. Did I win?

[barry]

This may raise a different question: should we measure brew temperature with or without coffee?

i thought we answered that already.