ChiaroScuro wrote:i did some temperature profiling of my new GS3 over the last few days, using a scace2 with an extech 421509 (calibrated in ice and boiling water).
first i did 10 trials following the WBC protocol, with varying times between trials. the results looked great: all temps fell within a window of +/- 0.55F, with a standard deviation of 0.34F. based on this, i calculated that, for my particular GS3, the "coffee boiler temperature" setting should be (desired + 4.1F).
next, i decided to try a longer pre-shot flush of 1 oz (about 5 sec). this raised the average temperature by 0.2F, but the stability over 8 trials was exactly the same as before, so i concluded the longer flush was no better than the 2 sec flush specified by the WBC protocol.
next, i thought i'd try some trials with no flushes at all. all of a sudden, i completely lost temp stability! in 4 out of 8 trials, the measured temp was 2 to 3 degrees under the set point.
for my next session, i went back to 2-second flushes, and i had excellent temperature stability for 8 trials.
since that session, i've done 3 separate sessions of several trials each, some with and some without flushes, and the measured temps are now ALWAYS 1 to 3 degrees colder than expected (never hotter).
during the course of the above work, besides recording the scace temps, i also watched the temps being reported through the GS3's display, dictated them into a voice recorder, and later transcribed them. for each trial, i calculated the average temp, from time 10 to time 25, as reported by the scace and as reported through the GS3's display. i calculated that my GS3's "coffee T offset" should be -5.4F (when using 2-second pre-shot flushes).
the temperature profile of a single shot reported through the GS3's display has a very different shape from the profile measured by the scace. interestingly, i noticed that, once i had my "coffee T offset" calibrated, i could just read the temp on the GS3's front panel any time between 10 to 15 sec into the shot, and that number would be very close to the overall shot temp measured through the scace (average from time 10 to time 25), almost always within +/- 0.3F!
now here's where it all gets very interesting. i've noticed that, as my scace temps vary wildly between 1 and 3 degrees too cold, the temps reported thru the GS3's display have stayed rock-solid, and if you believe them, the machine is working perfectly. also, the handful of actual shots of espresso i've made during all of these experiments have been wonderful. and i'm drinking the notoriously finicky vivace dolce. i assume i'd know by the taste if it were running 3 degrees too cold (but this is just my first pound of dolce).
so i'm strongly tempted to think there's something wrong with my scace measurements, not my GS3. i even called up greg scace, to get help disassembling and cleaning out the scace (cleaning it didn't fix the problem). he also suggested i try bleeding the group head (i didn't find any air) and re-checking the extech/scace in boiling distilled water, correcting for elevation and current atmospheric pressure (i did this just after the trouble started and again just now, and the calibration has remained perfect, within 0.1F).
at this point, i'm pretty perplexed. i'd really appreciate any ideas anyone has. i'm pretty convinced that my GS3 is working properly and is now calibrated as well as it can be, but i'd feel a lot better if i were seeing consistent numbers through my extech/scace. i'd also like to do some more experiments, like seeing if the machine remains stable without any flushing at all.
That does seem pretty perplexing. Let me address the difference between your boiler measurements and the Scace first. The measurement displayed by the GS3 is the water temperature adjacent to the probe inside the boiler. The Scace is measuring the temperature of water exiting the group. These are two very different places and it is entirely reasonable to have different answers and different profiles. The temperature probe in the boiler is close to the heating element and has a very short thermal path to the element, so heat is very efficiently transferred to the water by the probe. On the other hand, the group is hanging out in space quite a ways from the element. It is heated by water within the group neck. Hot water from the boiler travels up toward the group along the top of the neck and warms the group by giving up heat. This makes the water denser and it flows back to the boiler along the bottom of the neck where it is reheated. The flow is sort of like a thermosyphon, except that there are no tubes used. This sort of free convection loop is possible because the group neck is large in diameter compared to the length of the neck. For what it's worth, the smallest ratio of diameter to length that supports free convection is usually around 1:40, and the group neck on a Marzocco must be around 1:4 ish. So there's not much resistance to formation of the convection cycle. The heat from the free convection loop then gets transferred to the dispersion screen and other associated bits by heat conduction through stainless steel. Stainless steel is a crappy conductor of heat - on the order of most ceramic coffee cups. That means that the dispersion screen and other parts on the coffee side of the group must absorb heat when brewing commences. The heat lost to these parts from the brewing water is what gives the Marzocco its typical brewing temperature profile, which is to rapidly increase in temperature over the first three seconds, then increase more slowly, levelling off for the most part by the end of the extraction.
The profile you observe is related to the machine, and not to the Scace. I've done my own tests to confirm it, as have other people in the industry, including folks like Cimbali / Faema , who are now using Scace devices instead of modifying groups in order to get temperature data.
I'm glad that you successfully disassembled and reassembled your filter / flowmeter. I didn't think you would find the cause of your problem there. Sorry you still have a discrepancy here.
I think you should now try 4 things.
1) If your Extech has batteries inside, then you should change them out for new ones.
2) I'm not sure if your logger is connected directly to computer? If it is, I have observed grounding issues in the past with Picotech TC-08 dataloggers that use serial port connection rather than USB. In order to fix this I attached a wire to a screw on the case of the espresso machine to the screw attaching the serial port cable to the computer. I'm not sure that this is applicable for Extech, but it's worth checking if appropriate. Prolly worth checking even if the logger is USB connected as well.
3) You should do several checks of the thermometer using your boiling water scheme. Be careful not to immerse the bottom of the probe (the spring end) in the liquid. See if you get variable answers.
4) Put your Scace in the group and leave it in there for a day or so. It might be possible that liquid water entered the probe from the outside, near the spring. I'm not sure if it can happen because I've never observed it with one of these probes, but if electrically conductive water (most all water is unless it's distilled or RO water) got inside the probe it could form a thermocouple junction that was not at the end of the probe. That would give you the wrong answer. I've never seen it with a Scace, but I've seen it happen to supposedly water proof probes that were completely immersed, including leads. By leaving your Scace in the group you will help drive out any water that accidentally entered the back of the probe.
Lemme know what you come up with. If you can't sort it out, pm me and I'll give you an address to ship it to. I'll sort it out and send it back.
-Greg
