I recently acquired a Scace I thermofilter and would like to use it to determine the temperature offset between the brew boiler and grouphead of my LM GS/3.

GS/3 owners will be familiar with the brew temperature offset. The GS/3 measures temperature inside the brew boiler. Although the grouphead is directly connected to the brew boiler and is saturated with boiler water, it's slightly colder, presumably because it hangs out in space and isn't in full contact with the thermal mass of the rest of the boiler. So LM provides an "offset" figure which is programmed with the difference in temperature between the water in the boiler and the water in the grouphead. Typically, the grouphead is 4-7 degrees cooler. LM (or maybe Franke or Chris Coffee) programmed the offset on my machine to 4 degrees. I imagine they used a Scace thermofilter to do this. Unfortunately, the GS/3 doesn't automatically apply the offset to the boiler temperature setting, so the user has to do the math. For example, if the target brew temperature at the grouphead is 196.5 degrees, the brew boiler temperature on my GS/3 must be set to 200.5 degrees.

Below is a set of measurements taken with the Scace thermofilter using an Apollo DT302 digital temperature logger. The GS/3 target brew temperature was set to 196.5 (200.5 at the brew boiler), which is the recommended brew temperature for the Terroir Ademe Bedane SO I'm pulling this week, and the pull was for 25 seconds.



Here's the raw data, in one-second intervals.

177.9
178
178.2
180.5
190.7
194
196.2
197.1
197.6
198.3
198.5
198.6
198.9
199.1
199.1
199.3
199.3
199.4
199.4
199.4
199.5
199.5
199.4
199.5
199.4
199.5
199.4
199.5
198.6
197.4

I tried to start the logger at exactly the same time as the brew cycle, but it appears that I started it about 2.5 seconds early. You can see the temperature start to rise at about 2.5 seconds. If you count 25 seconds after that, the temperature starts to fall, which indicates the end of the shot. So, the actual shot data is the 25 values between the 4-second and 28-second marks.

The WBC protocol says that the first three seconds should be discarded, which puts the relevant data between the 7-second and 28-second marks. Those data points indicate an average temperature of 198.9 degrees, which is 2.4 degrees hotter than the target temperature I set.

It's not at all surprising that there's a difference between the grouphead temperature shown by the GS/3 at idle and the average temperature of the brew water: as hot water from the boiler replaces water that has flowed out of the grouphead, the grouphead temperature increases. But this means that you're always brewing at a higher temperature than the display indicates.

Now, the difference narrows quite a bit if we ignore the WBC protocol and just look at average temperature over the actual 25 seconds of the shot. In that case, it's 197.6, only 1.1 degrees higher than the target temperature. That's getting down to a fairly meaningless difference relative to the temperature stability at peak, but it's still higher than the target temperature displayed by the GS/3.

So the question is, how should I use the Scace data to adjust my GS/3 to duplicate temperatures recommended by roasters as closely as possible? Put another way, how should I use the Scace data to determine the correct offset between the brew boiler and grouphead temperatures? Should I use the average brew temperature computed using the data? If so, should I use the WBC protocol (reduce the offset by 2-3 degrees) or should I use all 25 data points (reduce the offset by 1 degree.) Does anyone know how GS/3 offsets are measured by LM or the distributor?

Of course, I have no idea what machine Terroir is using or how they determine and set the brew temperature. If they do use a Scace, presumably they're using average readings to set their machine temperature. But it's anybody's guess how they use that information to set their espresso machine.

I'm sure many responders will say this is a whole lot of worry about nothing. Just set the temperature for the best taste. Hey, back off! I'm not worried, OK? The temperature stability of the GS/3 is incredible and the difference I'm talking about, 1.1-2.4 degrees, is pretty minor. And I certainly do set the temperature according to taste (sometimes with very interesting results.) I'm simply asking the question out of intellectual curiosity (and, yeah, I'm OCD...)

I'm also asking the question because the possible discrepancy in how Scace data is used by roasters and customers makes the starting reference temperature something of a crap shoot if you don't know the relationship between your machine and that of the roaster. The Scace Thermofilter is supposed to provide a standard for comparing brew temperatures on different machines, but it seems to me that there's plenty of room for discrepancy in how the numbers are interpreted and used.

Peppersass wrote:So the question is, how should I use the Scace data to adjust my GS/3 to duplicate temperatures recommended by roasters as closely as possible? Put another way, how should I use the Scace data to determine the correct offset between the brew boiler and grouphead temperatures?

I don't know of a universally accepted standard, but I use the peak stable temperature, usually around the 15 to 20 second mark.

HB wrote:I don't know of a universally accepted standard, but I use the peak stable temperature, usually around the 15 to 20 second mark.

Well, that was my first instinct, even though it implied an offset of only one degree. But after reviewing the WBC protocol, I decided the espresso world must use the average temperature. Perhaps not. But if peak temperature is commonly used, then the typical offset range reported by GS/3 users seems rather high. Either that or the grouphead on my GS/3 has unusually low heat loss (doubtful.)

Have you calibrated the Scace and logger together?

Had a chat with Bill Crossland today in Anaheim at the SCAA Expo. I haven't been pre-flushing my GS/3 at all, and I asked him what he thought of that. He said I should flush just for a moment, not for brewhead temperature, but to heat the screen, which is relatively isolated from the saturation area. He said just long enough to turn the paddle on and off would do it.

shadowfax wrote:Have you calibrated the Scace and logger together?

Yes. First thing I did. Found the boiling point for water at my altitude using a website recommended by Eric Svendson (210.8 degrees at 680 feet.) Then inverted the Scace over rapidly boiling distilled water so only steam touched it. Left it that way for about 15 minutes and calibrated the logger for 210.8 degrees. The adjustment factor was -0.6 degree.

Marshall wrote:Had a chat with Bill Crossland today in Anaheim at the SCAA Expo. I haven't been pre-flushing my GS/3 at all, and I asked him what he thought of that. He said I should flush just for a moment, not for brewhead temperature, but to heat the screen, which is relatively isolated from the saturation area. He said just long enough to turn the paddle on and off would do it.

Pretty much what I do with my 3 second flush -- takes the chill off the exposed parts so to speak. Also, as Nicholas advised me, it's a good thing to do from a sanitary point of view.

When I did the measurements posted above, I also checked the temperature profile after a much longer flush of about 2 oz of water. I picked 2 oz semi-randomly: someone posted on Gs3cafe that he programmed one of his volumetric buttons to do a 2 oz flush before pulling a shot. Here's the curve after running 2 oz of water and then pulling a 25 second shot:



Here's the data:

197
196.8
196.9
197.5
198.2
198.6
198.8
199.1
199.2
199.4
199.5
199.4
199.6
199.6
199.6
199.7
199.6
199.6
199.4
199.6
199.3
199.4
199.3
199.4
199.5
199.5
199.4
199.3
199.2
198.7
198.3
198.2
198

As you can see, the effect was to shorten the ramp up time. This pushed the average temperature for the full 25 seconds up to 199.26, or 199.41 if the first three seconds are omitted. Basically, the curve gets flattened such that the average and peak temperature are about the same.

Bear in mind that the 2 oz flush wasn't chosen scientifically. I suspect a shorter flush can be used to produce a similar temperature profile. I'll have to experiment with that.

In any case, the longer flush eliminates most of the temperature differential during the first part of the shot. So if one is interested in keeping the brew temperature as constant as possible during the shot, as opposed to relying on the average brew temperature, then it would seem a longer flush is the way to go.

That said, it's not clear to me whether constant or average temperature should be the target from a taste perspective. Guess I'll have to experiment with that. Of course, the two different approaches require different adjustments to the GS/3 temperature offset.

another_jim wrote:Not sure about the espresso world. I think the WBC norm is to drop the first five seconds, and then average.

I got the 3-second figure from WBC Procedure for the Measurement of Brewing Water Temperature in Espresso Coffee Machines posted here on HB by Scace, Crossland and Sanders on September 1, 2005:

"In the case of manual data collection, the average brew temperature shall be the temperature observed most often during a specific simulated brew cycle, ignoring temperature observations during the first three seconds of the cycle. Ignoring results during the first three seconds negates the effect of thermometer lag on the result."

Is there a newer version of the document?

There was quite a bit of discussion in the thread that followed about how Schomer throws out the first 8-10 seconds, and that this is valid because the machine doesn't get up to peak temperature until then. Greg's response was that some people delete the data because they don't want to see it. This really gets to the heart of my question about which temperature to target: peak or average.

My gut feeling is that average makes sense. Seems to me the effect of water temperature on extraction must be taken into account during the entire time the water is in contact with the coffee, not just the time that the water is at maximum temperature. I would think that different solubles are extracted at different temperatures, so the resulting cup is a complex composite of the effects of different water temperatures at different saturation levels during the course of the shot.

But as the second set of data I posted shows, the GS/3's water temperature variation during the shot can be minimized by doing a long enough flush. This will bring peak temperature water in contact with the coffee for a longer period of time and there will be less variation. It remains to be seen, however, if this is a better approach from a taste perspective than a short flush or no flush.