La Marzocco GS3 Prototype

The GS3 prototype, in the hands of a skilled barista and with great coffee, produces truly excellent espresso with shocking ease. It is a genuinely revolutionary espresso machine. The combination of the technology involved, the design and engineering, and the materials and components used create a tool that enables a skilled barista to do wonderful things. With this machine, the experience of exploring and understanding espresso becomes easier and more practical. For the average barista, the results will be less earthshaking, but still amazing. Average baristas should be able to produce espresso of higher quality than they are used to with dramatically greater ease and unparalleled consistency.

Over the course of this test, a large number of different baristas (both pro and amateur) pulled shots of more than a dozen different coffees from long list of roasters. Stumptown Hairbender was the most common espresso used due to my familiarity with it. In addition, more than a dozen other people assisted in the test by tasting shots, providing input and comments, and sharing their opinions on the machine and on the results. These people ran from average coffee enthusiasts to top baristas like Bronwen Serna, to expert coffee cuppers such asDuane Sorenson. The report that follows is heavily influenced by their opinions.

Espresso Performance

Drips of a naked pour Over the course of the evaluation of the GS3, it became a object of great desire and even lust for all who pulled shots on it. In every case, the discussion rapidly proceeded from "is it a good machine?" to "do I like it?" to "is it better than X?" (fill in "Linea" or "Synesso" for X) and eventually to "is it the best machine you've used?" and finally to "oh I want one!" This conversation was largely fueled by the results when pulling shots.

As with any espresso machine, to quote Dan Kehn, "The problem is on the handle side of the portafilter." This has never been more true than with the GS3. Finally, the espresso machine "gets out of the way" of the barista. If there are problems with your espresso, just accept that it's your own fault. Either you are using the wrong coffee, the wrong grinder, have maintained things poorly, or (most commonly) your barista skills need work.

So what is it about this machine that makes this possible?

The GS3 has a combination of attributes, engineering and technology that need to be clarified if you want to appreciate what a breakthrough this machine is.

First: It is a dual boiler machine. While it is entirely possible to get great espresso from an HX machine, it is far easier to do so consistently and repeatedly with a dual boiler machine. With a dual boiler machine you are more likely to see accurate and consistent and stable brew temperatures. By having one boiler for brew water and a second for steam, you avoid having to compromise on either.

Second: The boiler has a PID controller. Rather than using a mechanical thermometer and thermostat, the GS3 combines a sophisticated control system that includes a programmable digital controller, a solid state relay and a thermoprobe. This allows for two key features: (1) the barista is able to see on a digital display the current boiler temperature and program the target temperature; brew temperature is programmable on the prototype in 0.3F increments (though I gather than the production machine may well have a controller that can be programmed to 0.1F increments - which would be fantastic), and (2) because of the nature of this sort of control, fluctuation of the boiler temperature is dramatically reduced over mechanical solutions. A complete explanation is outside of the scope of this review, but suffice it to say that this sort of control is more "intelligent" than the old solution and thus minimizes the deadband, resulting in more stable brew temperatures.

Third: The La Marzocco saturated group results in more stable brew temperature. Because the LM group is integrated with the brew boiler (and thus "saturated" by the water within that boiler), there are fewer issues with the group cooling when idle, as one commonly sees from other groups.

Fourth: The GS3 has some tweaks to the way the group is designed and plumbed that result in even greater stability (shot-to-shot and in particular after idle periods). These seem to be at least in part borrowed from the "Hybrid" group design briefly prototyped by La Marzocco in 2004. By re-engineering the solenoid and three way as well as the way that volumetric dosing works, LMI has created a machine where grouphead temperature is not subject to the same degree of cooling (or heating) cycle seen on other machines.

Naked stream of espresso Fifth: The GS3 has a commercial Procon rotary pump. While many people have figured out ways to get great results from vibratory pumps, rotary pumps are easier to get great results from stock, are quieter, and are more robust. In the prototype GS3, this rotary pump is internal and coupled with an internal reservoir. This results in very stable brew pressure profiles. There is still the "vane flutter" profile you see from any rotary pump, but there is no additional oscillation or curve. As has been noted in other locations, I have theorized that the lack of stability in brew pressure is the primary culprit for the lack of "clarity" in many espresso machines. Suffice it to say that the GS3 can produce shots with incredible clarity.

The Result? Combining all the above bits of techno-wizardry with the previously mentioned ease of use (stainless groups, automatic controls) and you get a machine that is as near to transparent to the barista as I've seen to-date.

On a quick side note, over time there has been a passionate debate about the merits of a flat brew temperature profile for espresso. Rather than relaunch this debate, I am merely going to point out that the GS3 has a very flat and stable brew temperature profile, commonly called intra-shot stability (see the cribsheet for details). More importantly, the GS3 also has great stability of brew temperature shot-to-shot (commonly called either reproducibility or inter-shot stability). The value of the latter type of stability does not seem to be a point of contention and, in fact, for many people in the commercial world it is a sort of 'Holy Grail' of espresso machines.

If you are an experienced barista, the learning curve for this machine is short. Other than learning to use the controls and learning to program the controller, little will be new to any barista. In fact, the hardest thing for most people to get used to is not having to do so much. It was hard for most everyone to stop their attempts to manage brew temperature. Repeat after me: Flushing isn't necessary.

One other key attribute of this machine is that it provides very flexible pre-infusion. This pre-infusion is not a "natural" pre-infusion like what you would see from a lever machine, nor is it a "progressive" pre-infusion as you would see from a true E61 group (or in its own way with a Mistral). The GS3 has what could be called "pump pulse" pre-infusion. When the extraction begins there is an initial period where the pump is activated, followed by a pause, after which the pump restarts to begin the full extraction. It is important to note:

  • During that initial "pulse" activation, the pump is at full pressure. There is no progressive ramp-up.
  • Both the initial pulse and the subsequent "off" period are programmable in duration (and per button on the control pad).
  • Pre-infusion is only available when the machine is used in automatic mode.

There is no pre-infusion available if you use the Continuous Brew button to mimic a semi-automatic machine. Initially I had mixed feelings about the theory behind this sort of pre-infusion since I'd assumed a more gentle form of pre-infusion would produce better results. The first shots I pulled from the GS3 made me suspect that my intuition was correct. Comparing shots pulled using the auto mode with pre-infusion to shots using the machine in semi-auto mode left me feeling like, at the very best, there was little difference in quality. After some more experimentation, however, I discovered that re-programming the pre-infusion for a slightly greater initial pulse and a far longer "off" period resulted in noticeably superior shots (to my palate).

In terms of the quality of the espresso, the GS3 produces shots that have a lot of similarity with shots from any of the top temperature-stabilized dual boiler commercial machines (by which I mean a PID'd Mistral, Linea, or Synesso). There is the same dense, creamy and hyper-saturated mouthfeel. In fact, if there is any difference, it is that the mouthfeel of shots from the GS3 is even denser and heavier. In addition, there is a sense of "concentration" at the low end of the flavor profile. The emerging theory is that the combination of the GS3's active pre-infusion and its low water debit (which, in itself could be called another form of pre-infusion) deliver this result. In addition to the similarity of this (desirable) mouthfeel, there is also the aforementioned cup clarity.

In the past I have posed various theories to explain this clarity issue. But, given that the sense of clarity is so elusive and isn't consistently available even from machines like the Mistral or a Synesso, I have started to see it as a sort of a mythic attribute. Over time I concluded that the most logical theory is that this clarity results from a combination of intra-shot brew temperature stability and intra-shot pressure stability (assuming, of course, solid barista skills and great coffee).

Measuring espresso extraction With the GS3, shots exhibited this clarity with incredible consistency. From the very beginning, I was able to get that "separation" of flavors that I desire. Imagine tasting all the unique and individual varietal characteristics of the coffee—each clearly defined and differentiated. Interestingly, I found that this was true at different brew pressures and shot after shot. I had not found either to be true in the past on other machines; in some cases the clarity was brew pressure dependent (as with the Bricoletta), and in all other cases, it was hit and miss on a shot-to-shot basis (as you'll see with a temperature-stabilized machine in a commercial setting). I don't have a great theory for the consistency across brew pressures, but I think that the incredible repeatability may well be from the lack of any brass in the system.

Coffees were reproduced beautifully - with no muddying of high notes, no blurring of low tones and clearly detailed aromatics. This has some interesting implications. I found some espresso blends that tasted noticeably worse with the GS3 than with other machines. In all cases, these were espressos where there was a bean in the blend that was either defective, damaged, mis-roasted or inferior in some way. With many other machines the flavor of this individual bean would be "blended" into the overall monolithic flavor of the espresso and its flaws would be muted as a result. With the GS3, I was able to clearly identify all the flavors of the various beans, and in some cases this is not a good thing. Do I consider this a problem with the GS3? Not at all! I consider it a great strength. But I also consider it a wake-up call to the roasters (both commercial and home) of the world. Currently, with many machines it is entirely possible to "hide" one bogus bean within the blend. Those days may well be coming to a close. In the past, you could use a little gauze to create an attractive soft-focus portrait. The GS3, however, shows real life—warts and zits and all.

As one might expect, the results in the cup were consistently good across coffees, but were most noticeable with the "intolerant" coffees. Coffees like the Intelligentsia Black Cat, the Doma Vito's Blend or the Artigiano Espresso tasted fantastic, but felt like less of an "accomplishment" than some of the other coffees. With these tolerant coffees, the shots tasted very much like what I would expect from a very good espresso bar. Where this machine really shines is with the finicky or difficult espressos. With a coffee like the Vivace Dolce, the trifecta of stability, reproducibility, and very discrete adjustment capability allowed one to fine-tune the shots to an unprecedented degree. Bronwen and I were both able to pull shots of the Dolce that were as good if not better than any shots we'd been served at Vivace itself. These shots were incredible thick and dense and sweet while somehow still being balanced and complete. Simply wonderful!

This is no slight to the GS3. It's just that pulling really good shots of a tolerant and forgiving blend with ease seems like less of something to be proud of than pulling really good shots of the Ecco Caffe Espresso Reserve (for example). On the other hand, it is fair to note that even with the tolerant blends, the GS3 has great value. The Vito's Blend, for example, tasted like a different coffee with this machine. Was it easier to get great shots of it than with, for example, a dialed in Bricoletta? Not really. But the flavors are different. With the GS3, I became aware of some wonderful tropical fruit and floral aromatics. One of the experimental espressos from Ecco Caffe didn't merely show intense chocolate in the body—it exploded into a layered range of chocolates (sequencing from initial notes of bittersweet Belgian chocolate through sweet milk chocolate to a final powdery Dutch processed cocoa finish).

Each coffee truly reveals itself when prepared with the GS3. This is largely due to the ability the machine gives you to control your environment. What is really amazing about the GS3 is the opportunity it gives a barista to really explore the results of small and controlled changes to the brew parameters. By providing such a high degree of stability and control, and by freeing the barista from having the manage the machine itself, the GS3 allows you to start to really pay attention to the results in the cup from small and inter-related changes to everything from brew temperature to dose to basket... on a coffee by coffee basis. This has the potential to enable new understanding of espresso at both a personal level and an institutionalized one.

On a side-note - the performance of single origin espressos on the GS3 was eye-opening. With all other home machines I have tried to-date, it seems like they perform best when using either blends or single origin coffees, but not both. With the GS3 this is not the case. The same ability to reproduce flavors and produce bell-like clarity in the cup results in gorgeous single origin espresso. And the dense mouthfeel and low-end concentration is particularly valuable with many single origin espressos, which often lack that deep bottom end. On the other hand, reproduction of "off" flavors in blends noted above applies to single orgin espresso too. Drinking shots of inferior single origin beans prepared using the GS3 can be a horrifying reminder of Dr Illy's comparison of bad espresso to "30 minutes at the dentist."

As with all my other reviews, I tested espresso by pulling numerous shots in different preps, temperatures, baskets and extraction times. With the GS3, however, this process finally became more structured and organized. With each coffee I would initially dial in the brew temperature (easy with this machine) down to 0.3F (the current minimum). I would then dial in dose and then extraction. Finding volunteers was easy, so I used other tasters as much as possible. This is mostly due to my desire to get as wide a sampling as possible given the subjective nature of taste and flavor. But it was also due to my fear of palate fatigue.

I tested the machine with a large range of coffees - from blends to single origins, from light roast to dark roast, from a huge range or origins and roasters.

In evaluating the GS3, various sorts of sessions were involved:

  • A "typical" home morning session of a double espresso and two double cappuccinos
  • Sustained "working test session" periods where four to five different coffees would be cycled through, with brew temperatures changed often on a per shot basis with total sessions often exceeding two hours,
  • "Catering" style sessions where drinks were made for up to 18 guests.

In all cases, the GS3 handled the work without a single hiccup. Unlike what you see with almost any other machine, it is hard to differentiate between the results from the various sessions. With the big commercial machines, the quality gets better with heavier usage patterns. With most home machines, quality begins to degrade with heavy sustained use. But the GS3 seems to have no weakness when it comes to this type of session.

During my evaluation of the GS3, I had an interesting opportunity. For the long Thanksgiving weekend, I rented a house on the Oregon Coast with a bunch of friends. Included in this crew were Bronwen Serna (Hines and Vivace Barista and 2004 USBC Champ), Valerie Hoecke (beginning barista, coffee fiend and Stumptown regular), Amanda Bryan (noobie coffee fiend) and James Cooper (coffee fiend and Vivace regular). I decided to take the GS3 with us. This turned out to be an amazing and informative experience. In many ways, this shows off one of the greatest things about the GS3 - its compact nature. It was no problem to throw the machine in the back of the Subaru and take it to the coast (though carrying it down two flights of stairs made me wonder if the boilers were, perhaps, made of lead rather than stainless steel). Finding room for it on the counter in the kitchen of the rental house was no trouble. We didn't have to worry about plumbing or special wiring. And for a long weekend, in the middle or nowhere on the Oregon Coast, we had world-class espresso whenever we wanted it. Now that is cool!

The scores below compare the GS3 to the machines I know the best—a stock La Marzocco Linea three group semi-auto, a Schomer modified (temperature stabilized) La Marzocco Line two group semi-auto, and a three group Mistral (see comparative equipment for more details). These scores are based purely on the opinions of the tasters involved (myself included) and not on any suspected or assumed experience of a mythic "average user." I'll score each machine on a 0-10 scale using the following attributes:

  • Ease of use—how much skill the barista requires to get consistently good shots
  • Aromatics—rate the presentation of aromas
  • Mouthfeel—rate the feeling on the palate
  • Reproduction—how faithful to the beans
  • Clarity—how well defined are the flavors
  • Flavor—rate the overall taste and aroma.
  GS3 Stock LM Linea Modified Linea Mistral

Ease of Use



































I can hear the screams already. Some may accuse me of going overboard or hyping the GS3. Honestly, these are the results as I see them. When it comes to the quality of espresso being produced, the GS3 is superior to any of the commercial machines I have extensive experience with including the temperature-stabilized Mistral. The perfect score on Ease of Use is what put it over the top, but even so it's a huge accomplishment.

In my humble opinion, when it comes to producing high-quality espresso the GS3 is the current state-of-the-art.

Steaming Performance

Steam joystick closeup When I first saw the GS3, one of the things I was most excited about was the cool-touch steam wands (I still have scars from steaming on the Mistral), and one of the things I was most doubtful about was the "joystick" steam controllers. Needless to say, I was both excited and apprehensive about evaluating the steaming performance of the GS3.

Admittedly, I'm not a big "milk guy." I drink the periodic short milk drink - but I'm mostly all about the coffee. I get a kick out of seeing really good latte art, but I'd rather have a great tasting drink than a great looking one.

At first it seemed like my apprehension was well-founded. The stock (and only available) steam tip is a variant of the LM acorn tip that I don't like. In addition, the wand rotates side to side, but doesn't articulate in any other direction. It traps a lot of condensation within the wand. And finally, as feared, I was struggling with the joystick controller.

I thought a swap to a different tip would be a fix - but it turns out that the stock tip is the only one that will (safely) work with the cool touch wand. Other tips are likely to result in water being "sucked" up through the interior of the wand and potentially damaging the machine. In talking with Bill, I found out that they have already addressed this issue and that there is a fix that allows you to use any of the La Marzocco tips with the GS3. Whew!

As time went by and I became accustomed to the machine, I grew to like the joystick. After the first couple of weeks, I was entirely comfortable with it; after six weeks of testing, I find that I actually prefer it to the Linea style knob controllers. Its immediate response is very valuable—not only can you go to full pressure very quickly, you can go to zero pressure just as fast. And that allows for far finer control of milk texture. In addition, the multi-directional nature of the joystick allows this to work well for nearly all users.

The GS3 on full power has impressive steaming performance. While it is not on par with a commercial 220V Linea, it's damn close. In fact, I am afraid that most home users are going to have a hard time getting used to working with this sort of power. A six ounce pitcher of milk takes about 3 seconds to stretch and another 3 seconds to steam. You have to work quickly and very, very smoothly if you want good texture. On the other hand, once you know how to use the power there is no going back. The ability to stretch without over-heating the milk allows for the creation of incredibly sweet textured milk.

Even after getting used to the joystick, however, I was having real problems getting the milk I wanted. In tracking the experiences other testers were having with this machine (in particular Mark Prince), it was clear that I was hardly alone. Nobody seemed to be having any luck steaming milk with this machine. So I put my failures down to poor design and prepared to write a little report on the weaknesses of this machine. And then three things happened.

Rosetta First, I heard from Bill Crossland that they were re-engineering the wand (see above). While this didn't address my needs with this prototype, it at least set my mind at ease when it comes to the production machine. Then I found out that the fix was done and I could start using any tip (once I had the requisite parts). But most importantly, Bronwen (re)taught me to steam milk over the Thanksgiving weekend on the Coast. To net it out— I was lazy and I was overthinking it —all at once. With a simple change in technique and a more important change in mindset, great milk was not only possible but easy.

With the GS3 it is absolutely vital that you move the pitcher as little as possible. You need to have "still hands" while steaming. In addition, you need to steam by using the tip in the center of the pitcher rather than angling the wand along the side of the pitcher. If you relax and keep your hands nearly motionless while focusing on correct and centered positioning of the tip, the milk quality will be nearly of the quality that you can get from a Linea.

The stock acorn tip (currently the only tip option) is, as noted earlier, was nobody's favorite. The dispersion pattern is wide and the holes are really large. This makes it difficult if not impossible to get a good rolling action in the milk in small to normal pitchers. I would much rather see the now standard bullet tip. And, of course, this is now possible.

Compared to most home machines, the milk quality is dramatically more sweet. I've been able to get milk of equally good texture from many home machines, but I've never been able to get milk of this sweetness with the high quality texture as well. With most home machines I seem to have to choose between getting ideal flavor or getting ideal texture but cannot get both.

The GS3 can easily steam milk down to three ounces in a twelve ounce pitcher and up to ten ounce in a sixteen ounce pitcher (and perhaps beyond that but I had no real interest in steaming that much milk). It's fast and it's reliable.

By the way, Greg Scace has been working with Bill Crossland on a modification to allow the machine to provide even more steam power. If this makes it into the production version, you can assume that the GS3 will truly be a milk texturing monster.

A couple of things to keep in mind. First, while the wand is "cool touch" that doesn't mean it stays cool under all circumstances. If you steam large quantities of milk back-to-back, the wand can eventually get quite warm. In addition (as one poor soul learned), just because the wand stays cool under most circumstances doesn't mean the tip does so as well! Second, I was never really able to resolve was the condensation in the wand. From talking with the folks at La Marzocco, I gather that this is one of the reasons behind the re-engineering of the wand and will be fixed in the production version. Finally, it is incredibly important to keep the wand clean. Because of the construction of the cool touch wand, it is possible for milk to collect and condense at points within the wand. In addition, the acorn tip has enough "texture" to it to create numerous points where milk can collect and dry on. Given all of this, it is very advisable that you clean the wand with a commercial dairy cleanser like Rinza on at least a daily basis.

As with any machine, the GS3 takes some getting used to. Even with the improved design of the steam wand and the availability of additional tips, this will remain true. You're going to have to assume that it will take you a couple of weeks to master steaming on this machine. At first, it is likely that you'll be frustrated as a result. Personally, I think this is simply due to the fact that making espresso with this machine has such an incredibly truncated learning curve. By comparison, the GS3 is a beast to learn to steam with. But once you get the hang of it, the GS3 is close to the current standard bearer - the LM Linea - when it comes to quality of milk.

Next Page...