Monitoring Brew Temperature - E61 & Silvia

[erics]

Hi -

I have "designed" and Swagelok has manufactured (for me) a custom fitting to take the place of the M6 x 8 allen-head screw found on E61 groupheads and on Silvia (others also). I have termed this fitting the "Thermocouple Adaptor." The Thermocouple Adaptor is available in either brass or stainless steel and will accept ONLY a 1/16" thermocouple of your choice. It replaces the M6 allen-head screw by screwing into the grouphead and sealing with a copper washer exactly like the existing M6 screw in your grouphead. The 1/16" thermocouple is sealed to the adaptor by means of replaceable nylon ferrules. No drilling, tapping, soldering, brazing, cutting, teflon tape, or epoxy are required. Installation instructions for the adaptor can be found on my FTP "website" in the folder entitled "Thermocouple Adaptor": http://users.rcn.com/erics/. Click on Adaptor06.doc and save the file to your computer. Please read carefully as it is designed to answer most questions concerning installation and use. You might also want to read the Adaptor Tidbits file as it also contains useful information.

The thermocouple adaptor kit consists of the adaptor with copper washer, a spare copper washer, a spare nylon ferrule set, the original metal ferrule set, and a plug to seal the adaptor should you wish to suspend temperature measurements. After inserting your thermocouple and attaching it to your appropriate meter, you can easily observe and/or record brew and/or flush water temperatures approximately 1.25" upstream of the espresso puck/empty basket using your normal espresso brewing/flushing practice. EQUALLY IMPORTANT, you will be measuring a temperature that is very representative of grouphead temperature.



OK, I want one, what do I do now? - Read the installation instructions, then read the installation instructions and finally read the installation instructions again appropriate for the adaptor kit you desire. Complete ordering, inventory, and pricing information can be found on my FTP site in the document entitled "Pricing_Inventory_Ordering". You need to be somewhat mechanically inclined and understand the use of thermocouples and appropriate temperature meters OR have a good friend who is. I am requesting a $15 donation to CoffeeKids in addition to the cost of the adaptor kits. The adaptor kit prices, EXCLUSIVE of any donation to CoffeeKids but including shipping via USPS 1st Class are: Brass - $28.00 - $30.00, Stainless Steel - $53.00, Digital Thermometer - $66.00. Digital Thermometer with switchable degrees F/degrees C - $74.00 (limited availability). Shipping outside of the US (except for Canada) via USPS Priority Mail International is typically $11 additional. Two checks are required (if paying by check) -- one to me & one to CoffeeKids. Paypal is accepted for an additional ~3.6% (the PayPal fee).

COMMENTS/SUGGESTIONS on the instructions or questions of a general nature that could be included in the FAQ section of the install manuals would be GREATLY appreciated. Thus far, seventy-five [75] thermocouple adaptors and three-hundred & thirteen [313] digital thermometer adaptors have been shipped to users of varying ability in Australia (9), Austria (1), Canada (13), Denmark (5), England (5), Germany (4), Guam (1), Hong Kong (1), Japan (1), Netherlands (4), New Zealand (1), Norway (2), Poland (3), Puerto Rico (1), Saudi Arabia (1), Singapore (1), Sweden (3), and, of course, the US (332). Stock Status:

http://users.rcn.com/erics/

E-61 Stainless Steel - 1
E-61 Brass - 0
Silvia Brass - 1
Digital Thermometer - 34
Digital Thermometer in Celsius - 6

THE 100th ADAPTOR WAS SHIPPED TO SWEDEN - HOW APPROPRIATE !

Orders can easily be placed for additional adaptors but keep in mind that there is usually a 6 to 8 week lead-time from Swagelok and it is always best to check stock status BEFORE you place an order as this post is, maybe, edited on a monthly basis.

Edited 04/15/08 to reflect adaptors shipped, stock status, and/or improved drawings and new availability of digital thermometer reading in degrees Celsius.

Acknowledgements:

Thanks to Andy S. for suggesting I go with an E61 adaptor vice just one for Silvia after my original post on AC and to Sean L. for suggesting nylon ferrules and to Greg S., Sean S., and Dave S. for devising all sorts of innovative methods of measuring brew water temps. I also thank Lino V. for his unselfish OK to use several of his E61 group drawings contained in the thermocouple adaptor, digital thermometer, and adaptor tidbits documents.

The original posting on 03/06 was cross-posted in its entirety to AC and CG.

Eric S.
erics@erols.com

[Strugs]

I just wanted to weigh in on what Eric has done here. If you have an E61 machine, I have only one piece of advice for you....run out and get one of these bolt temp probes. While it is true that some espresso blends can be brewed within a range of ± 5 degrees, these blends are the exception, not the rule. In my opinion, the ONLY way to get the best out of your Spro is to ensure that your brew temps are (a) stable; and (b) consistent. As some of you know, the temperature monitoring mod on my machine is located further down the brew path than the one Eric has produced. However, I also made a similar version of Eric's probe a short while before doing my current mod, and there is not a whole lot of difference at the end of the day.

After using my probe for several months now, I would never dream of using an E61 without a temperature probe. Think of it as driving a car without a speedometer. Sure you can get a feel for how fast you are going, but try to explain that to a cop when he pulls you over.

So - major props to Eric and kudos for his insistence of the donation to CoffeeKids as part of the price. I spent well over $300 on the trial and error method of my current temp monitor rig. Eric has done all the hard work on this - his price is a steal for what is probably the most useful mod you can make on an E61 machine.

[Ozark_61]

I just posted a long reply at CG - but a very hearty second to what Strugs said. You'll not be able to live without this once you try it!
Cheers,
Geoff

[HB]

Greg milled his Astra to insert a temperature probe (discussion), Dave detailed a low-cost alternative in No More Guesswork! Naked Truth of E61 Temperature Revealed, Sean fabricated his own adapter as documented in Thermal Obsession...

... and now Eric has kindly done all the legwork to bring us a ready-made package:


E61 adapter with 1/16" sheathed thermocouple

Although I've been on respite from "gadgets" as part of my study of holistic barista techniques, the attraction of his offer was irresistable - I e-mailed Eric the same hour that I read his post. The adapter arrived last week and I installed it over the weekend:



Eric's instructions are thorough and packed with well-drawn diagrams. However, in my excitement to get it installed, I initially didn't insert the probe at the depth shown above, but rather near the bottom of the hole, directly into the valve chamber. Needless to say, the temperature measurements fluctuate dramatically in that location. Moving it closer to the location Eric suggests does produce more steady readings, but I'm intrigued by how accurately the sound of the "water dance" predicts the chamber temperature.

For HX owners who want to end their uncertainty about the proper cooling flush, this adapter / thermocouple combo looks very, very promising.

[RapidCoffee]

... and now Eric has kindly done all the legwork to bring us a ready-made package:

Although I've been on respite from "gadgets" as part of my study of holistic barista techniques, the attraction of his offer was irresistable...

I'm intrigued by how accurately the sound of the "water dance" predicts the chamber temperature.

For HX owners who want to end their uncertainty about the proper cooling flush, this adapter / thermocouple combo looks very, very promising.

Dan, is that a Vetrano install? Please continue to post your results. I've also been very interested in this device, just haven't had the time to follow up lately.

Thanks - John

[HB]

Yes, the adapter is installed on the Vetrano.

I plan to post temperature graphs reported by the thermofilter and adapter at different depths. For now, I'm evaluating it purely as an aid to managing the cooling flush. As my first test, I positioned the probe at the bottom in the valve chamber and followed my usual routine to see (a) what it read, and (b) if the end-of-dance technique was consistent shot-to-shot. The readings confirmed what we've said all along - nailing the temperature accurately based on sound is easy, especially if the shot interval is a few minutes minutes apart. At shorter, irregular intervals, it's difficult to judge with precision. That's where I am assuming the adapter's readings will prove most helpful.

[erics]

Hello Everybody -

I operate a PID'ed Silvia and observe "grouphead" temperatures via a Type T thermocouple and the thermocouple adaptor. I read the temps on an Omega HH506RA and am continually amazed at the grouphead's behavior from a temperature standpoint.

Silvia's performance is documented here: http://users.rcn.com/erics/ under temperature graphs and also in the FAQ section of the installation manual available in the folder entitled "thermocouple adaptor".

For now I make only Cappy's but they approach and, if I am patient, meet the quality of those served at Murky here in DC. For me, that's saying a lot because Murky is truly top quality.

Eric S.

[HB]

The wife was out for the evening, so I ran a series of thermofilter tests using the TC adapter to judge the correct flush. In other words, I want to answer the question: Assuming one forgets about observing the water dance or listening for HX sputtering, how consistent are the brew temperatures if one just flushes until the meter says temperature X followed by rebound Y?

I had already determined that the proper temperature at the well depth for my routine (flush... rebound 30 seconds... pull) was 206F. Keep in mind that for the purposes of this test, I don't worry about what the adapter temperature relative to the brew temperature might be, as long as it indicates the same readings every time. I postulated that since we're really tuning the flush, it's better to read the temperature BEFORE it's attenuated too much by the group, so I placed the probe at the bottom of the well. As I reported earlier, the reading is changing quickly at the bottom, but that's reality. I see a very distinct "knee" right around 205F. The temperature moves slowly prior to that and then plummets just past it. For this particular machine, my assumption is that it marks the end of the HX over-temp water exiting.

The series was proceeded by a "warm up" flush (flush... wait a couple minutes... begin series). Preamble out of the way, below is my Friday evening results:


Irregular pulls (3m, 1m30, 3m, 15m, in no particular order)

Some observations:

• Grouphead is definitely cooler and the HX hotter after the machine is idle for a long period (see line in red; rises more slowly at first, spikes higher)
• "Warm up" flush would spike well above the target brew range then settle down (this "high hump" profile was not recorded)
• Adapter feedback made it easy to stay within 1F, although occasional outliers still occur, especially if interval was short (see blue dashed line with 200.4F peak temperature)
• The adapter reading and thermofilter reading converge about midway into the shot; the thermofilter indicates a very slow decent towards the end and the adapter is reporting near flatline temperature from the HX (I did not show both readings on the chart above for simplicity).

The utility of Eric's adapter is plainly evident to me: By tuning the flush amount based on the adapter's TC reading, reproducible temperatures are easily obtained within ~1F. I wonder how well this holds up with the lowest-cost setup (approx. $20 digital thermometer + type K thermocouple) in place of my pricey rig (Fluke 54-II + special limits of error type T thermocouple).

[RapidCoffee]

Thanks for posting your observations! Very impressive brew temp stability and reproducibility, especially from an oft-maligned HX.

For comparison, how well did the water dance method fare? I.e., were you able to improve brew temp reproducibility with the E61 TC adaptor, or did it simply provide verification that you could perform a "good" flush by ear?

Gonna have get me one of these thangs. Any chance my TM-125 ("cheapest thermocouple and meter on the planet") is adequate? If not, any suggestions for a more sophisticated TC/meter?

- John

[erics]

Hi John -

These thangs require a thermocouple other than what (I believe) you have with the TM-125 and I suspect that the read rate of the TM-125 may be highly filtered and not readily adjustable. Does the TM have a standard thermocouple connection or a miniature one? Another good choice for a meter is one of the inexpensive Ebay 1/32 DIN PID's from Auber or ColdFusion, i.e. SYL1512 or TET-7100 - essentially identical.

Eric S.

[HB]

For comparison, how well did the water dance method fare? I.e., were you able to improve brew temp reproducibility with the E61 TC adaptor, or did it simply provide verification that you could perform a "good" flush by ear?

For the first week, I pretty much ignored the adapter readout while flushing. That's how I arrived at 206F as the "right" temperature to stop flushing. When I did peek at the numbers, I was usually within 0.5F if it was one of the later shots and the pace was not faster than one every three minutes (i.e., leisurely weekend mode). For the first shot, the adapter readout confirmed what I had long suspected - the temperature spikes higher than normal - and had already compensated by subtracting a few seconds from the rebound time.

So yes, I was able to improve on brew temperature reproducibility, especially for the situations where the water dance method is tricky (e.g., a shot pulled between 1m30-2m30 after the last). Bob Yellin's simple rule of "flush two ounces if it's less than three minutes" works reasonably well for that situation, but the adapter enabled me to improve on it.

In summary, those who worry about the complexity of HX temperature management have a solution that a half-blind one-armed barista suffering from a mild case of AADD could not screw up.

[HB]

The prior chart only showed thermofilter readings. Below is an example of the two location readings shown together. To demonstrate the attenuating effect of the group, the adapter reading was taken from the bottom of the well (nearer the HX exit), not turn "D" as Eric's instructions suggest.


Temperature within the valve chamber exit (blue) versus thermofilter (red)

[BobY]

This is really interesting stuff! I've received my adaptor but haven't had time to install it yet. I plan to run a series of data-logged simultaneous comparisons between Greg's and Eric's device on my Andreja Premium with many different "starting temperatures" (time between shots) etc., as soon as I get a little time.

Dan or Eric (or any other adaptor owners with experience): Can you comment on the position of the probe tip and its effect on the results, beyond what is stated above. IOW, before I go ahead and install the probe, given the experience to date, is the best place to position the tip still the one described in Eric's "04" instructions? Eric's instructions are very precise as to where the tip should be positioned.

I know you (Eric) don't own an HX machine.

BobY

[another_jim]

Thanks for the data, Dan.

This is good news. I was sceptical about how well a thermocouple sealed into a tube would work, since an experiment Andy and I did 2 years ago using a hollowed out 5mm screw to house the TC failed due to the lag between the leading and the actual temperature (that speed up of the flush temperature drop after it gets to around 96C or 205F is a killer for any slow temperature reading setup). Obviously, the 1/16th tube insulated by the nylon ferrule is a very different animal from a hollowed out brass screw losing heat to the rest of the grouphead!

... And this certainly beats getting a bare TC under the screen and gasket, since those get chewed up every few weeks.

[erics]

Hi BobY -

Certainly I believe installing the probe as specified in the installation manual is the "correct" position. I even went so far as to inquire about a CFD (computational fluid dynamics) program to optimize probe position but the fact that average fluid velocities are so low (0.953 fps) in this area rendered that idea moot.

The heat transfer antics in the water flow path within the grouphead of ANY machine are complex and the closer you can get to the actual point of delivery, the better off you are from an information standpoint. The best, to date, is certainly Greg's Thermofilter. Next would be his (and Sean Strugnell's) drilling/tapping the grouphead in the dispersion screw area.

Omega HTTC thermocouples are different from their TJC or TJ series in that the HTTC's are a hollow tube (low cost) while the TJ and TJC are filled tubes. Omega felt there was no difference in response and the benny is that the HTTC lets the wire do the bending. I have a 1.5" HTTC here (specifically for Silvia) that I hope to install today and run some simple tests as in the FAQ section of the manual.

Hi Jim - the nylon ferrules are just there for ease of thermocouple replacement and movement - not for insulation. The thermocouple junction itself is located within the tip of the probe. Yes, I too remember Andy's screw modifications.

Eric S.

[BobY]

Thanks Eric. I'll be very interested in your impressions of the difference(s) between the TJ and HTTC series after you do your testing because I have both options.

BobY

[another_jim]

Hi Jim - the nylon ferrules are just there for ease of thermocouple replacement and movement - not for insulation. The thermocouple junction itself is located within the tip of the probe. Yes, I too remember Andy's screw modifications.

Eric S.

Hi Eric,

I realize the ferrule is a seal; I was just speculating about its thermal properties. In Dan's graph, your sensor leads the Scace device's attenuated response by about 3 seconds, which seems about right (since the Scace device is affected by the baffling material in the basket). When I compared the hollow screw with a Schomer device (somewhat more attenuated than the Scace device), the graphs were reversed, with the screw slowing down the TC response to the point of uselessness.

The tube for the TC you're using may be more conductive than the brass screw, but I was speculating the nylon insulated it from the rest of the group, allowing the whole assembly to change temperature much more rapidly.

[erics]

Lots to reveal -

First, I had my old thermocouple (Omega TJC, Type T, 2.0 inches) positioned about 1/4" deeper than I should have. This is great, right? I write the instructions and can't even follow my own advice.

I replaced the TJC with the 1.5" HTTC, positioned correctly this time as per the instructions and I am a "happy camper". The HTTC, being a hollow tube, cools off quicker than the TJC, but responds beautifully to temperature changes. I had both the HTTC and the Thermofilter plugged into my meter but had the reading interval set to 10 seconds versus 1 second because of another project. It's a shame because that was a good shot, starting with the adaptor reading 190.2 and the thermofilter reading 183.7. The "my fault" 10 seconds readings were: Adaptor 190.2, 202.1, 204.6, 205.0, Thermofilter 183.7, 198.4, 199.4, 200.9.

I like the HTTC from a performance standpoint but the itty bitty wire would maybe bug me if it were exposed. I remain convinced that the proper probe location is as per the instructions but hey - that's what makes this a great forum.

The only way to get a SS braid is with the TJ series if you're big on looks.

The TJC offers wire size and Teflon covering identical to the Thermofilter.

The HTTC offers small wires but no bending necessary.

I think I'll order another TJC tomorrow and position it properly this time around. I am obviously interested to see how it does from a performance standpoint.

Maybe I post some graphs tomorrow.

Eric S.

[BobY]

FWIW I did some measuring today with just the two TC types to check out response times and hands down, the HTTC responds faster; in fact it's an almost instantaneous response. That's the one I'll be installing with 1 measurement per second set up on my Extech.

BobY

[erics]

At least you could have signed your response with "Bob (I'll give the other guy a chance)Y"

I agree with your comments on response, Omega seemed to think they were comparable. I'm also going to see if I can get the HTTC with fat wires.

Take this kindly as I know you will and as I believe you know - it takes a lot of playing around before you draw conclusions.

Eric S. (still set up to do measuring so I will suffer in the AM for the sake of science)