The Impact of Frothing on HX Temperature Stability

[Ken Fox]

I'd always assumed that once you start frothing milk, you can kiss goodbye the temperature stability of your heat exchanger. This only makes sense, afterall, given all the heat that goes out through the steam wand. I'd never bothered to even look at shot temps while frothing with my Scace Device. Although I love my Fluke, having to type data into a spreadsheet, and then transfer it into a graphing program is a real PITA, and I wasn't willing to give Fluke another $150 for their rip off downloading software.

Omega has a cheaper (and deservedly so, I'm afraid to say) datalogger, the HH506RA, which comes with poor quality downloading software that *works*, and the whole package costs about the same as the Fluke software alone. I decided to buy one because I collect data from my roaster and my espresso machines regularly and typing the numbers in just sets off my tennis elbow (I don't play tennis, but I got it anyway).

I decided to do 5 separate "walk up shots" on my PID'd Cimbali Junior D1 rotary machine, 3 years old, modified with the aforementioned PID and a delay timer that allows 3+ bar preinfusion followed by a full pressure (9 bar) extraction. My machine can froth enough milk for a cappuccino in about 8 seconds but I decided to set up the experiment with 15 seconds of frothing as some may be slower and/or want to froth more milk.



This graph was obtained by flushing the group 50ml (my standard and programmed flush), then pulling a 30 second simulated shot with the Scace Device and Omega HH506RA datalogger. At the midpoint, 15 seconds into the shots, I opened up the steam valve all the way, and left it open for the remainder of the shot. Each shot was separated by more than 10 minutes, up to an hour and a half in one case.

What you see here is essentially no impact at all of frothing on shot temperature on any of the five datalogged shots. I don't believe that this finding comes from the presence of the PID in this machine, rather I think it is the inherent design of Cimbali heat exchanger machines. How transferable this information is to other heat exchanger designs made by other manufacturers, I do not know, and hope that others will test.

ken
p.s. this is cross-posted on alt.coffee; please don't respond in both threads

[Worldman]

Ken,

First: "nice machine!!!"

I know that many advocate not frothing while pulling a shot, but I have never found that it made any difference in the body, taste, crema, etc. of the whole process and usually just frothed away while pulling the shot(s). It is good to see that you have some evidence that it makes no difference (at least with your Cimbali). While I have no instrumentation on my machine, experience tells me that it matters not whether I wait to froth.

However, now that my nekkid' PF is new, I just keep watching the espresso as it falls from the bottom of the basket and frothing would just get in the way of that.


Len

[Ken Fox]

Ken,

First: "nice machine!!!"

I know that many advocate not frothing while pulling a shot, but I have never found that it made any difference in the body, taste, crema, etc. of the whole process and usually just frothed away while pulling the shot(s). It is good to see that you have some evidence that it makes no difference (at least with your Cimbali). While I have no instrumentation on my machine, experience tells me that it matters not whether I wait to froth.

However, now that my nekkid' PF is new, I just keep watching the espresso as it falls from the bottom of the basket and frothing would just get in the way of that.

Len

Jim Schulman sent me an email (which I need to respond to) that mentioned that he thought the major effect of frothing was to increase the time it takes in between shots, e.g. recovery time, but it does not effect the actual shot temperature control. I thought it would effect the actual shot temperature. I'm sure that the recovery time is effected but at the rate that most home baristas work, it is a non-issue.

Of course, if one is talking about a 110v machine, like the machines I have, it is silly to compare their recovery rate to what one would need in a busy cafe, being as no one in their right mind would use a 110v machine in a busy commercial setting. We home baristas may make some damn fine shots, but we don't work as fast as a good professional. And, why would we want to?

ken

[Matthew Brinski]

Ken,

I did a run on my machine (Vetrano) this afternoon. I followed my normal flush / idle routine to maintain a 199F brew temp with an intershot time of 3 minutes. I did a set of five simulated extractions with a Scace / Fluke setup and datalogged the results. I then repeated the set with full steam during the entire extraction. I don't have the results in graph form yet, but I can say that there was NO significant difference in brew temp variability. I find this extremely surprising given that I have discovered that if I deviate from a specific flush/idle routine in the least, the brew temp on my machine shows great variability.

Matthew Brinski

[Ken Fox]

Ken,

I did a run on my machine (Vetrano) this afternoon. I followed my normal flush / idle routine to maintain a 199F brew temp with an intershot time of 3 minutes. I did a set of five simulated extractions with a Scace / Fluke setup and datalogged the results. I then repeated the set with full steam during the entire extraction. I don't have the results in graph form yet, but I can say that there was NO significant difference in brew temp variability. I find this extremely surprising given that I have discovered that if I deviate from a specific flush/idle routine in the least, the brew temp on my machine shows great variability.

Matthew Brinski

Hi Matt,

Thanks for checking and posting. I'm not familiar with your machine other than that I believe it is an e61; if not, please correct me. I think this is very good news. Has your machine been modified in any way, or is it stock? Perhaps I was naive in assuming that frothing would upset the applecart of temperature stability in a heat exchanger. I felt this way because it is part of the mantra in favor of double boiler machines, that you need a separate boiler for frothing because you can't get one to be right for both, which assumes in part that the act of frothing degrades shot temperature performance. Proponents of double boilers believe that since the froth comes out of a separate boiler, shot temperature is unaffected. An interesting side aspect of this question is how 110v operated machines of either design would function in the face of service demand approaching their design limit. The heat exchanger has but a single element and it is sized so that it doesn't draw more amperage than the circuit can put out. A double boiler presumably would rely on electronics to prevent the machine from drawing too much amperage and will have to preferentially shut off power to one or both elements in order to prevent the breaker from flipping. So, depending on the situation you could conceivably have better thermal performance from a 110v HX machine than from a 110v double boiler if both were stressed. An early double boiler, the Reneka Techno, would not allow frothing while pulling a shot, a somewhat inelegant approach but one they felt forced to implement.

With the PID in place I can get 3, possibly 4, temperature "bands," encompassing the desired brew temperature range of about 198 to 204F. These bands are not spot on, but rather provide a brew temp "approximately" what I want. It remains to be proven by blind testing how finely the human sense of taste can detect small differences in brew temperature; we can probably all taste 4 degree F differences, many can probably taste 3 degree differences, but how many can detect sub-2 degree F brew temp differences? I don't want to discount those supertasters in our midst, I just want to get a handle on how many of them are out there. Only with this sort of information will we know how much importance the average enthusiast should put on absolute temperature stability and reproducibility.

I believe I have reached the end of the road of what I can do with my present setups, until or unless the ensheathed TC probes are changed for longer ones that can be put in the immediate vicinity of the submerged bottom of the heat exchangers. There is a lag time period of uncertain duration that comes from sampling temperature in the steam column of a boiler when what is desired to be controlled is the water being expelled from the heat exchanger a number of centimeters away.

Still, it is nice to know that there are at least 2 designs of heat exchanger machines out there that can froth and still maintain the heat stability that they have. There are a lot of improvements that can be made on heat exchanger design, such as those reportedly present on the Aurelia, so there may be life left in the old design, yet.

best,

ken

[Matthew Brinski]

Ken,

Yes, the machine is an E61 thermosyphon type. Yes, it is stock ... well, I upgraded to a Sirai p-stat, but I personally don't consider that a mod. I considered installing a PID (it's still in the box along with an SSR) and even further considered installing a second boiler based loosely on the Brewtus configuration (I can hear the heckling right now), but abandoned the idea due to the probability of getting a GS3 - no, I'm not rich ... yes I'm insane. I want my current machine to remain stock for resale.

My motivation for going to a PID'd dual boiler is in regards to obtaining a specific brew temp on demand and staying there rather than for steam capacity/performance. The steam / brew temp results were surprising to me, and it really speaks to the engineered balance of the heat element, boiler size, and hx circuit, but I have a challenging time maintaining a repeatable brew temp with the same flush machine from one day to the next. I try not to rant too much about it in the forums because I am continually learning about this machine, but I've discovered how ambient temperatures and small variances in flush/idle methods can significantly alter the brew temp. I can add further observations and go into details, but I don't want to go too far off of your OP.

In short, the results I obtained with today's tests are both surprising and promising, but I have other reservations about home/small HX machines that make me question their benefits and the extent of their usability ... at least the machines that are similar in construction to mine.

Matthew Brinski

[Ken Fox]

Ken,

Yes, the machine is an E61 thermosyphon type. Yes, it is stock ... well, I upgraded to a Sirai p-stat, but I personally don't consider that a mod. I considered installing a PID (it's still in the box along with an SSR) and even further considered installing a second boiler based loosely on the Brewtus configuration (I can hear the heckling right now), but abandoned the idea due to the probability of getting a GS3 - no, I'm not rich ... yes I'm insane. I want my current machine to remain stock for resale.

My motivation for going to a PID'd dual boiler is in regards to obtaining a specific brew temp on demand and staying there rather than for steam capacity/performance. The steam / brew temp results were surprising to me, and it really speaks to the engineered balance of the heat element, boiler size, and hx circuit, but I have a challenging time maintaining a repeatable brew temp with the same flush machine from one day to the next. I try not to rant too much about it in the forums because I am continually learning about this machine, but I've discovered how ambient temperatures and small variances in flush/idle methods can significantly alter the brew temp. I can add further observations and go into details, but I don't want to go too far off of your OP.

In short, the results I obtained with today's tests are both surprising and promising, but I have other reservations about home/small HX machines that make me question their benefits and the extent of their usability ... at least the machines that are similar in construction to mine.

Matthew Brinski

Matt,

Given the variation in shot temps you've observed; how much of a fluctuation in temperatures do you think it takes before you can taste the difference in brew temp?

ken

[barry]

make ten drinks in a row, steaming with each drink, and see if your results hold.

[Ken Fox]

make ten drinks in a row, steaming with each drink, and see if your results hold.

These were random walk up shots, which in my experience are more variable than shot series. The issue with the shot series would be the intershot interval needed to get a semi-stable system. But no one in their right mind would use a 110v machine in this way. In a business setting busy enough to do 10 milk drinks in succession, 110v machines are not generally used. In a situation where you had to make 10 milk drinks, in a home on rare occasions for example, I would froth enough for 2 or 3 drinks each time I frothed, so there would be periods of frothing and then 2 or 3 shots made without frothing.

Earlier today I played around with one mixed frothing no frothing scenario, where arbitrarily I made a simulated straight shot, then frothed while making shot #2, another straight shot, then frothed while making shot #4, then two straight shots, #s 5 and 6. I'd actually meant to froth on shot 6 but forgot to do it , and I'm too lazy to repeat the series. This is at an inter shot interval of 2 minutes, e.g. pull a 30 second shot then wait 90 seconds and start the next:

There are definite limits to what you can accomplish (temp-stabilitywise) on a heat exchanger, and I think I'm bumping up against them. There might be some improvement with a longer TC probe, one that could be bent to be in the water column, just up against the bottom of the heat exchanger where the water is being taken out. I don't know how much improvement one could get from this, and it would a be a bit of a PITA to implement as the HX would need to be removed to see that your were positioning the probe correctly. Perhaps I could thread a TC down the water intake siphon from which the water exits the HX, but aside from the fact that I can't figure out how to get the wire out there without making a leak through the seals, I'm concerned that the TC could become malpositioned and somehow fail to register boiler temperature changes and this could cause the boiler to overheat, resulting in blowing the pop valve, at an unpredictable time.

ken

[Matthew Brinski]

Matt,

Given the variation in shot temps you've observed; how much of a fluctuation in temperatures do you think it takes before you can taste the difference in brew temp?

ken

Ken,

I honestly don't know the answer to that in terms of temperature per say. What I have realized is that if I'm not "dead on time" with my flush/idle routine, I can taste differences in the cup, mostly with SO type coffees. A favorite coffee of mine is Terroir Northern and it's a love/hate deal because I get some amazing shots from it, but if I vary as little as 15 - 20 seconds on the intershot time, everything goes to hell. I also use the Terroir to make machiattos which just come through incredible at times, but even with a short milk drink such as that, my flush / idle routines have to be near perfect or I lose the up front almond/ floral? type taste. Call me lazy, but it just gets to be tedious at times to maintain "on the mark" flushes and extractions. Having said that, now knowing what I do about simultaneous steaming/brewing, maybe the PID will come out of the box ... I don't know.

Matthew Brinski

[Matthew Brinski]

... also, I repeated the simulated extractions today with near identical results. I'm still amazed.

Matthew Brinski

Perhaps I could thread a TC down the water intake siphon from which the water exits the HX, but aside from the fact that I can't figure out how to get the wire out there without making a leak through the seals, I'm concerned that the TC could become malpositioned and somehow fail to register boiler temperature changes and this could cause the boiler to overheat, resulting in blowing the pop valve, at an unpredictable time.

ken

Sounds like putting in a pacemaker ... :)

mb

[Ken Fox]

... also, I repeated the simulated extractions today with near identical results. I'm still amazed.

Matthew Brinski



Sounds like putting in a pacemaker ...

mb

Well, here's another series I just did, this time with my 10 year old vibe pourover machine. This time I really mixed it up, to try to simulate a situation where you might have a bunch of people over and were making some straight shots and some milk drinks; since the frothing lasted 15 seconds, it would be enough for 2 drinks (on my machine, which froths like hell), so even though there were only 3 frothings, it would have covered 6 drinks if need be.

I had intervals of 5 minutes, after which there was a 50 ml flush, or 1.5 minutes (shotmaking pace of 1 drink per 2 minutes), mixed in, with 3 frothings of 15 seconds each. A total of 9 simulated shots were pulled over about half an hour. I began writing down the order of the shots and which were frothed and which weren't, but I lost track midway and decided to just omit that information as it doesn't really matter. The shot temp curves are all similar, with the familiar heat exchanger hump at the start. The part that might be attributable to the PID is the shot temperature stabilization for the last half of each shot.

Here's the results I got:

The qualification here is that I can't tell how much of these results are attributable to the basic design of the machine, and how much to the PID. The temperature variation appears to be confined within about 2 degrees F.

What this says to me is that in a normal, low volume home usage setting, at the speed that a good home barista normally works (slower than a pro in a busy cafe), that at least some home 110v heat exchanger machines can maintain a respectable shot to shot temperature variation in spite of how much frothing is being done.

Froth away!

Best,

ken

[barry]

In a business setting busy enough to do 10 milk drinks in succession, 110v machines are not generally used.

yes and no. there are plenty of busy settings where a 110v machine is required, and bumping up against the capacity of a machine is not limited to 110v machines. running a 220v machine flat out making capps/lattes can significantly deplete the steam supply, possibly sufficient to change the brew temperature.

[Ken Fox]

yes and no. there are plenty of busy settings where a 110v machine is required, and bumping up against the capacity of a machine is not limited to 110v machines. running a 220v machine flat out making capps/lattes can significantly deplete the steam supply, possibly sufficient to change the brew temperature.

Of course we are largely talking apples and oranges. You are speaking of a situation dealing with "the public" who thinks that espresso is a bitter black beverage that needs to be disguised in 20 oz. of milk so it will be drinkable. I won't serve a drink like that in my house no matter what; the biggest I'll go is maybe 6 oz. of milk with a double shot and I try to talk the person out of it. So if you are making more reasonably sized drinks where the milk isn't the overwhelming ingredient as one finds in many N. American cafes, then you would probably be a lot less likely to overwhelm a 220v. or even a 110v. heating element since the problem is going to be volume dependent.

Obviously, if you want to stay in business then you give the customer what they want, but if you come over to my house and demand a 20oz latte I'd just as soon you never came back so refusing to supply it makes it easier on the two of us

If you were making more reasonably sized drinks, say 4 oz. of milk in a milk drink, maximum (which I'm sure is more than you find in a cappa in Italy), do you think it would be possible to overwhelm a 220v. heating element?

ken

[miKe mcKoffee]

Ken,

I honestly don't know the answer to that in terms of temperature per say. What I have realized is that if I'm not "dead on time" with my flush/idle routine, I can taste differences in the cup, mostly with SO type coffees. A favorite coffee of mine is Terroir Northern and it's a love/hate deal because I get some amazing shots from it, but if I vary as little as 15 - 20 seconds on the intershot time, everything goes to hell.

I find the flush & go technique Chris talks about in his Bric' review much more accurate for nailing shot temps time after time +-<1f of target, ~198f to 206f on demand on my Bric' running 1.0bar boiler. This method of tuning the count down from flash pace to get 1f drops per count takes a bit of practice, then becomes pretty second nature. My count is a bit faster than 1sec per FWIW. Thanks to having use of a Thermofilter when I got my Bric' dialing in shot temps was a breeze. (a bit time consuming breeze, but a breeze) After initial HX cooling flush, doesn't matter if it takes 35sec or a couple minute to build the shot (or next shot in a series), the flush & go flash will just be a bit longer or shorter before beginning the count down.

I've also found starting steaming right after starting the shot pull not a problem with Bric'. However, this is using a slower GP2 tip, which suites my weak capabilities! (About 35sec for 3oz for cap. Yeah I know that's slow, nothing like commercial ~10sec speeds) If just doing back to back shots the recovery time is ~minimum 35sec after post shot group flush and PF wiggle in a long series to maintain shot temps. (tested out to 30 shot series multiple times). When also steaming recovery time is longer but nothing that holds back my meager skills! (I don't recall the exact test time results off the top of my head, seems like it was 50sec or something like that) I'd suspect the Vetrano recovery slightly slower because of it's lower powered heater. When I had the Bric' up around 1.3bar steaming was stronger, but the flush & go flash also much longer so I tuned for a compromise favoring straight shots.

[HB]

I find the flush & go technique Chris talks about in his Bric' review much more accurate for nailing shot temps time after time +-<1f of target, ~198f to 206f on demand on my Bric' running 1.0bar boiler. (emphasis added)

Not to put too fine a point on it, but what you say is true for that particular HX machine, not all. I made the mistake when first evaluating the Cimbali Junior of applying the flush-n-go technique I preferred for La Valentina. The results were disasterously bad. I contacted Ken and he was kind enough to set me straight (Jim Schulman also provided a number of Junior-specific insights). Some semi-commercial machines perform better with the "rebound technique" documented in HX Love (e.g., Andreja Premium), others the flush-n-go technique (e.g., Elektra A3). And others are have unique expectations (e.g., Cimbali Junior).

[miKe mcKoffee]

Not to put too fine a point on it, but what you say is true for that particular HX machine, not all. I made the mistake when first evaluating the Cimbali Junior of applying the flush-n-go technique I preferred for La Valentina. The results were disasterously bad. I contacted Ken and he was kind enough to set me straight (Jim Schulman also provided a number of Junior-specific insights). Some semi-commercial machines perform better with the "rebound technique" documented in HX Love (e.g., Andreja Premium), others the flush-n-go technique (e.g., Elektra A3). And others are have unique expectations (e.g., Cimbali Junior).

I'd have expected a flush & go count speed to vary machine to machine, but are you saying it just doesn't work accurately at all with some like AJ? Interesting. Since I haven't had the opportunity to play with zillions of machines I guess I got lucky when I read about the technique and bought the same machine!