Digital E61 thermometer adapter - Why does group temperature nosedive after flush?

Beginner and pro baristas share tips and tricks for making espresso.

#1: Post by quar »

Hi all,

Just installed a digital thermometer adapter and am very impressed with the unit. However, I'm seeing some (to me) strange behavior. Let's say that I want to flush to 197 and let it rebound. If I stop flushing immediately at 197, my temp continues to decline to the lower 190's and takes at least three minutes to rebound to 197. I've been playing with it so much, I suppose that my problem could be that I'd overflushed and hadn't let the GH fully recover (30 minute wait?), but I'm somewhat worried that my thermosyphon is clogged somewhere. Always in fear of scale due to our hard water, but I've been keeping that in check with bottled water and descaling.

Is that normal?

I ran through the test protocol that Eric S. recommends in his .PDF after installation, and my machine recovered from 185 to 198 in a little under three minutes. Somewhat slower than his, but in the same ballpark. This was after fully coming to temp and idling at 212.5. P-stat was sat to ~1.25 BAR but this MA-TER stat I'm using has a .25 BAR deadband, which is a little more than I'd like. PF is removed from group.

Quick update: I'm letting her heat back up now and I've adjusted the stat to 1.3 BAR. GH seems to be coming up to temp a little faster on the initial warmup.


User avatar
Supporter ★

#2: Post by erics »

Mike -

The observed grouphead temperature of 212.5 at a boiler pressure of ~1.25 bar is certainly reasonable given the tolerances of the thermometer and the boiler pressure gage.

Flushing to 197 F looks like this when you graph it:

I stopped graphing after about 4 minutes but obviously the temperature would eventually return to where I started. Normally my Anita runs at 212 and 1.20 bar after real long time idle but I had just pulled a shot about 15 minutes before so Anita had only recovered to around 210 when I saw your post.

More information on the digital thermometer is on my FTP site ( but the reason the temperature goes down after your flush is that heat is being drawn out of the group via a reverse thermosyohon action, i.e. the group is hotter than the hx.

Eric S.
E-mail: erics at rcn dot com

Urnex: 100% dedicated focus on coffee and tea cleaning
Sponsored by Urnex
quar (original poster)

#3: Post by quar (original poster) »

I've been reading through the thread you pointed me toward earlier and the docs on your FTP site. What I was seeing this afternoon, was when getting ready to pull a shot, my GH temp would be around 206F (10 to 15 min after a previous shot). I'd get the large spike upwards as one would expect when the hot HX water enters the group and then a fairly smooth decrease in temp till stopping at 197ish. Once stopping the flush, I might have a couple of seconds rebound followed by a steep decline down to the lower 190s. I was then taking about three minutes to come back up to 197-198. I'm concerned that this rebound may be way too slow. At the same time, I was playing around with it quite a bit and may have inadvertently overflushed.

What I'm really looking for is a way to use the thermo adapter to replace my current technique of flushing 2oz past the end of the water dance and then waiting 40 seconds or so, much like Dan's technique in the HX Love article. I tried the flush and wait technique this morning with my machine at rest for over 30 minutes and it worked, but the two minute+ pause is just too long. I guess I need to figure out if my machine is taking an abnormally long time to recover though. When using flush and wait for a subsequent shot, do you follow the same routine? (i.e. 185 deg flush recovering to 198 in ~2 minutes? What kind of wait between shots?


User avatar
Supporter ★

#4: Post by erics »

Actually, lately I have been flushing to around 183 because I bumped the pstat up to 1.20 for one reason or another. My particular methodology for preparing the espresso is such that the machine and I are usually ready at the same time +/- 5 seconds.

My rountine is usually 3 to 4 shots with 20-30 minutes between shots in the AM and every shot follows the same routine - flush & wait. The wait time is devoted to preparing the puck, rinsing the steaming pitcher, cleaning the steam wand, etc., etc.

One of the uses of the digital thermometer would be for you to "map out" YOUR current successful routine and then adjust (or not) YOUR "flush to" temperature or YOUR "start shot" temperature to taste.

Eric S.
E-mail: erics at rcn dot com

User avatar

#5: Post by HB »

quar wrote:What I'm really looking for is a way to use the thermo adapter to replace my current technique of flushing 2oz past the end of the water dance and then waiting 40 seconds or so, much like Dan's technique in the HX Love article.
My approach using the thermocouple / thermometer adapter is different than Eric's. His instructions say to install the probe at a depth equal to the turn towards the exit:


This gives a good approximation of the actual brew temperature and the reading is not volatile. However, I prefer the probe deeper because the reading changes more quickly, allowing me to flush more accurately. This approach works very well with the type T thermocouple / Fluke 54-II combination because they react rapidly, though it's not bad for the thermometer adapter despite its slower response time. More recently I've been using the thermometer adapter exclusively and Jon / RegulatorJohnson's "blind flush" technique, excerpted below from Need hints on using E61 thermocouple adapter:
HB wrote: I tried this technique on the Vibiemme Domobar Super and it works very well! The "blind flush" drops the temperature more slowly than an unimpeded flush, which plays to the strengths of a E61 thermometer adapter versus the thermocouple adapter. Less water in the driptray, less refilling and emptying.

As the unofficial keeper of heat exchanger nomenclature, the techniques to-date are:
  • Flush and rebound - from the original HX Love article, a flush followed by a 30-45 second wait while the heat exchanger recovers.
  • Flush and go - for fast recovery HX machines, this is a flush followed by the extraction within 10 seconds.
  • Flush and wait - Eric's flush to well below the target temperature, then waiting for the thermosyphon to re-establish itself until the thermometer adapter reading matches the desired brew temperature.
  • Blind flush - Jon's "flush until you're close, and then flush with blind basket" approach above; requires Eric's thermometer/thermocouple adapter.
I believe the particulars of Jon's approach will vary from machine-to-machine because of the tuning of the return versus direct flow from the thermosyphon during an extraction. That is, the water reaching the brewhead isn't only from one leg of the circuit:


Although this diagram depicts a one-way street, there's no shutoff solenoid blocking the return (blue) leg during an extraction, so the water reaching the group will be a mixture of the water that's passed by way of the heat exchanger (red) and some that's pass through the return (blue) leg. If I understand correctly what's happening, Jon's "blind flush" is forcing the water to rapidly circulate, thereby equalizing the system's thermal balance without expending water. Neat idea, Jon!
Jon's approach makes it much easier to nail the flush precisely because the temperature drops in "slow motion" compared to an unimpeded flush. One caveat though: It works exceptionally well on E61 HX machines like the Vibiemme Domobar Super that have over-pressure valves venting from the cold side of the HX, I assume because the turbulence drives the thermosyphon and equalizes the group temperature. The Quickmills vent the OPV from the hot side of the HX; I don't have one to test, but I assume their temperatures will drop more rapidly as a consequence.

Perhaps Quickmill owners with the thermometer adapter would benefit from a hybrid approach? That is, flush until the flash boiling stops, drop in a blind basket, continue with a "blind flush" to the target temperature. Remove the blind basket, drop in the prepared basket, wait maybe 10 seconds for good luck, pull. Anyone willing to try it?
Dan Kehn

quar (original poster)

#6: Post by quar (original poster) »

Something along those lines is what I was thinking of. I'll give it a shot in the morning. I was wondering if the blind flush would have the same issues that the flush-n-go has with Quickmill machines.

I'm just very concerned that something is causing my GH to recover more slowly than it should. If I flush my Anita down to 185 as in Eric's previous graphs, I don't recover nearly as quickly as his machine does. Takes me more like four minutes. However, I ran the same test that Eric posted in this thread, flushing to 197 and it was close to the same. I'd assume the recovery from 197 is quicker, because the GH isn't cooled as much as it is when flushing to 185. Not sure what to think here.

That being said, it seems to take a lot of water to flush down that low. Wonder if for some reason, that much flushing is just way too much for my machine and is cooling the head too much. I do have one of the very first Anitas. Wonder if there's been a change in the design since the first batch.


Supporter ♡

#7: Post by SLC »

All of this is quite intriguing to me.

I visit Jon time to time in order to annoy him and learn from him., and he is a really nice guy. I observed his method and I like it and it hits the temp really well. The only concern he has and I have is when using it on his or my Vetrano the more frequent than usual blind flushing takes the oils out of the lever. The lever then squeaks too much which most likely is not good over time.

What works well for me is a double flush, wait and pull.

If the machine has been idle for sometime I flush until the water dance stops (about 4 oz. .9-1.1pstat setting). I then prepare everything but the grind, tamp and pull. After about 1 minute post first flush, the temp is at 202ish. I then pull until the water dance stops again (about 2oz). It then takes about 16 seconds to grind and tamp. The temp is at 199 where I want it. I pull and then at about 15 seconds into the shot the temp is at about 199 and then stabilizes there until I stop the shot.

That is what works for me, feedback/comments appreciated.

Baratza: skilled in the art of grinding
Sponsored by Baratza
User avatar

#8: Post by RegulatorJohnson »

HB wrote:Perhaps Quickmill owners with the thermometer adapter would benefit from a hybrid approach? That is, flush until the flash boiling stops, drop in a blind basket, continue with a "blind flush" to the target temperature. Remove the blind basket, drop in the prepared basket, wait maybe 10 seconds for good luck, pull. Anyone willing to try it?

hey dan that is what i have fallen into with the vetrano. :D

serious that works very well for me regardless of idle time. i think it is a more gentle way to flush.

if you try it let us know how it works out.

2012 BGA SW region rep. Roaster@cognoscenti LA

User avatar

#9: Post by RegulatorJohnson »

SLC wrote:I visit Jon time to time in order to annoy him and learn from him., and he is a really nice guy.
it may be true that i am a nice guy.. but it is completely false that it is annoying to have you over.

2012 BGA SW region rep. Roaster@cognoscenti LA

User avatar
Supporter ★

#10: Post by erics »

Jon was one of the three original field testers of the thermocouple adaptor back in March of 2006 so, for sure, he and I have had and continue to have many pleasant email exchanges.

That said, I don't particularly agree with his methodology because it is very dependent on the type of machine you're flushing and the way that machine is set up. There is a world of difference between doing a blind flush on a Vetrano as compared to an Anita or Andreja because as Dan pointed out, the OPV on Anita/Andreja is at the "exit" of the hx and thus you are sending the "blind flow" (unheated water) through the hx and back to the tank. On a "blind flush" with Vetrano, you are simply recirculating water within the rotary pump and on other E-61 machines, the blind flush flow is simply heading back to the tank because the OPV is located on the entrance side of the hx (Valentina, Giotto, Vibiemme, etc., etc.). Note that Jon described and discussed his method in response to a post by Elan S. who also has a Vetrano. The additional complexity of plumbed-in machines is that the incoming water temp is ~70-75 F whereas it is ~95-100 F with Anita and Andreja.

Jon's method for his Vetrano is repeated here for convenience:


1. Remove PF & do a short initial flush "until the bubbling and/or hissing stops"

2. With a rubber disk blank in your normal PF basket, do a backflush until such time that the pump reaches full pressure (pump sound will change). Continue this backflush for a time period of ~3 seconds after the pump reaches full pressure. Return brew lever to the off position. Note grouphead temperature during this backflush.

3. Pause for ~10 seconds with the brew lever in the off position and the PF w/ blank disk in place.

4. Repeat step (2) above until the grouphead temperature is equal to your desired shot temperature. Stop the pump when this temperature is reached.

5. Remove PF, wipe basket dry, and "build shot" in basket as per your normal procedure.

6. The process of grinding, dosing, distributing/leveling & tamping is defined as "building the shot" in the basket. A total of 25-45 seconds is allotted to this process.

7. Pull shot at any time within these 25-45 seconds.

This is my interpretation of his procedure and certainly I will send him an email about this thread. Needless to say, there are numerous ways to operate an espresso machine and what counts is that method which is easiest and most repeatible for the given scenerio.


The "proper" probe protrusion (no offense to Dan of course) for the digital thermometer on all Quickmill machines is 3/8" as detailed in the install manual. This protrusion is set by me on all thermometer adaptors that are shipped to Chris' Coffee. This dimension should always be checked prior to installation because shipping can be a little upsetting.


Data I have shows a 60 second differential in recovery time when the flush to temp is changed by 4 degrees. A gradual lengthening of the recovery time would prompt me to at least do a partial inspection for scale buildup; likewise a substantial reduction in recovery time after a descaling operation would be cause for a minor celebration.

Eric S.
E-mail: erics at rcn dot com