One outstanding issue in brewing espresso drinks is the need for two different temperatures. Espresso tastes best when the coffee is extracted under pressure at temperatures well below boiling, typically in the 198-204F range. But what about frothing milk? That requires steam. Large volumes of steam for frothing are generated in a sealed boiler under pressure, producing temperatures that greatly exceed the boiling point of water.

There are several popular espresso machine designs, each with a fundamentally different method of handling the production of two temperature settings*. Single boiler, dual thermostat machines allow you to switch between brew temperature and steam temperature at the touch of a button... but can take several minutes for the boiler to stabilize at the new temperature. Although single boiler espresso machines may function very well for straight espresso shots (no milk), they are marginal for producing even one milk-based espresso beverage. Try to make cappuccinos for a crowd, and you're better off with another design.

Lever machines also have a single boiler, but typically only one thermostat setting. The boiler is kept at steam temperature, always ready for frothing milk. Brew water is magically cooled to the right temperature by contact with the cooler group head. The disadvantage? After a couple of shots, the group head heats up, and can no longer function appropriately as a heat sink.

Heat exchanger (HX) machines, like levers, have a single boiler that is kept at steam temperature. Fresh, cool water is drawn through a heat exchanger and flash-heated to brew temperature. The heat exchanger is a thin tube of metal that runs through the boiler. The large volume of hot boiler water, typically at least ten-fold greater than the volume of the heat exchanger, heats the brew water on the fly as the shot is pulled.

Double boiler (DB) machines take a brute force approach to temperature management. Need water at two different temperatures? Fine, let's just put two separate boilers in the machine. The brew boiler is kept at brew temperature, and the steam boiler at steam temperature. Primary disadvantages are additional cost and power requirements.

Most high-end prosumer and commercial machines use either the HX or DB design, because these allow you to crank out one milk-based espresso drink after another. So which is better? This question has become more relevant with the introduction of several new DB models during the last year. And so the debate rages on, occasionally reaching heights of absurdity. In a recent CG thread, one DB proponent flatly proclaimed:

ANY dual boiler machine beats ANY HX, hands down.

For the past few weeks, I have been in a position to compare the performance of a pair of prosumer espresso machines, one HX and one DB. These are the Quick Mill Vetrano and the La Spaziale Vivaldi S1. Pricing on these machines is similar (close to $2K, with the S1 typically costing a few hundred dollars more than the Vetrano). Both machines won CoffeeGeek Editor's Choice Awards in 2006. Both are fully plumbed rotary pump models.

But there the similarities end. The Vetrano uses the classic E61 grouphead: 58mm in diameter, heated by a thermosiphon, with lever-activated preinfusion. The S1 uses the La Spaziale grouphead: 53mm, heated by conduction from the boiler, and no preinfusion (preinfusion has been added in the S1V2).

Despite these fundamental differences, I thought it might be interesting to share some of my observations on these two machines. Take them for what they are worth: hearsay evidence, nothing scientific whatsoever. A valid comparison of HX and DB designs would require paired machines with the same grouphead, differing only in their use of a heat exchanger vs. dual boilers, and a double blind taste test to generate the data. Ian (cafeIKE) is already in a position to do this with his VBM HX and DB models. If someone wants to loan me, say, an Izzo Alex (HX) plus Izzo Duetto (DB), I'll be glad to report my findings.

Getting back to the Vetrano-S1 comparison:
First and foremost, both of these machines are capable of making truly excellent espresso. Forget the "hands down" hyperbole; that simply does not hold water. Earlier this week, a friend came over for a tasting session. After pulling several paired shots, he confirmed my impressions of the past few weeks. The pours are subtly different on the Vetrano and S1. The S1 shots taste slightly sweeter and flatter. The Vetrano shots are brighter, more complex, with a wider range of flavors. We both liked the S1 for straight shots, but preferred the Vetrano for Americanos and cappuccinos. Again, this is entirely consistent with my general impressions over the several weeks, using different blends, temperature settings, and brew ratios.

How much of this taste difference is attributable to the HX vs. DB design? Well, that's an interesting question. You could certainly argue that the Vetrano's preinfusion or larger diameter baskets are responsible for taste differences, but I believe it's more likely to be brew temperature differences. The S1 should be providing brew water at a fairly constant temperature. The Vetrano's grouphead thermometer (recommended) shows a typical "humped" HX brew temperature profile. The wider range of HX extraction temperatures may fully account for the wider range of flavors in the cup.

Some other observations:
* Brew temperature management is easier on the DB. But it's not all that much easier, especially with a grouphead thermometer on the Vetrano.

Huh? What about the dreaded HX flush? Guess what, folks: you need to flush both machines for optimal results. Let me repeat that: you need to flush both machines for optimal results. Two 2-3 oz flushes are recommended when the S1 has been sitting idle, to bring the cooling grouphead back up to brew temperature. Only one flush is needed for short idle periods, and none if you are pulling one shot after another. The Vetrano requires a cooling flush after sitting idle, to eliminate overheated water that's been cooking in the heat exchanger. Flush volume? Typically 3-4 oz. In fact, I often found myself flushing more on the S1 than the Vetrano.

However, the DB flush volume is relatively unimportant, whereas it's critical to monitor the flush volume closely on an HX machine. This is not difficult, but it does require paying attention to either the water dance or the grouphead thermometer.

* Reproducibility is easier on the DB. The S1 (presumably) produces a flat brew temperature profile, pour after pour. Nailing the same humped profile on the Vetrano is a hit-or-miss proposition that depends on the idle period, flush volume, and delay prior to pulling the shot.

* Pulling shots for a crowd is easier on the S1. Just wait for the (small) S1 brew boiler to come back up to temperature (30-40 seconds after the previous pour) and you're ready to go. Recovery time on the Vetrano is longer. The (large) S1 steam boiler is capable of frothing large quantities of milk in short order, and milk frothing has no effect on espresso brewing. The smaller Vetrano boiler is great for frothing small quantities of milk, but is slower on large volumes... and heavy use of steam will require a recovery period before brewing espresso.

* The HX design lends itself very nicely to experimenting with different brew temperatures. Sure, you can program in a new brew temperature on the S1... and then wait for the machine to stabilize at the new temperature setting.** To change brew temperature on the Vetrano, all you need is a longer or shorter flush.

* Personally, I prefer the 58mm brew group. There have been claims that the narrower, taller Spaziale baskets reduce channeling. I see no evidence to support this claim. In my experience, taller baskets (like 58mm triples) are more susceptible to channeling, not less. There is also a much wider range of options in filter baskets in the ubiquitous 58mm diameter. Whenever I use the S1, I dearly miss my Synesso ridgeless doubles.

* The lack of insulation on the S1 steam boiler is, IMHO, inexcusable. The uninsulated brew boiler, only slightly less so. The saving grace (if you remember to do this): it's very easy to shut off the steam boiler when not in use. The boiler comes back up to pressure in just a few minutes.

These observations may not generalize to other HX and DB models. Some commercial HX machines with large boilers and heat exchangers appear to have excellent brew temperature stability. Other popular prosumer DB models feature the E61 grouphead (which may or may not be optimal for a DB). The large steam/small brew boiler configuration on the S1 makes sense to me, but perhaps a large PID'd brew boiler has even better thermal stability.***

So there you have it: some thoughts on HX and DB machines, and a first cut at comparing two popular HX and DB models.

Bottom line: I like both of these machines very much. Deciding which one to keep in my kitchen will be a difficult choice.



* Pardon the generalizations in these machine design descriptions. I'm well aware that there are several different variations on these themes.
** The S1 user interface for programming brew temperatures and water volumes was designed by a demented, evil, sadistic mind.
*** Coming from the lever/HX universe, I find ultra-precise brew temperature stability to be overrated.

John, whether fact or fiction, that was a darn nice read, and will (hopefully ) further confuse the noob attempting to decide between HX or DB!
But seriously, you answered questions, and maybe raised others.
I also wonder if the deeper basket would, due to the water having farther to penetrate the puck (that sounds a bit risque) would cause variances in taste...layering, if you will.
Th S1 is still my dream toy, though Mistress Cremina still has an ever willing slave.

Thanks Rob. I would certainly recommend the S1 to anyone with DB-itis. It's a nice-looking, well-built machine with one of the longest proven track records among home DB models (not sure which came first, the S1 or the Brewtus). One of these days I may spring for the preinfusion retrofit, just to see if it makes any difference in the pour. But it does just fine in stock configuration.

Basket shape is an open question, recently discussed on another thread. You've raised a good point, touched upon by Scott Rao in The Professional Barista's Handbook: a cylindrical basket shape is not ideal, because the top layers extract more than the bottom layers of the puck. A tall, thin basket will exacerbate this effect.

It's not my intent to play "confuse the noob", as much fun as that sounds. I hoped to open a discussion on some of the topics in my post, and hear what others think. The HX flush, e.g., has never been a big deal for me. Flat vs. humped brew temperature profiles may be significant, and this deserves careful study at some point. Properly tuned HX models can probably approximate DB behavior, although the converse may not be true. Lots of interesting stuff to consider. Regardless, it's pretty clear to me that the "DB wins hands down" is an extremist view, and the reality is far more complex than that.

RapidCoffee wrote:One of these days I may spring for the preinfusion retrofit, just to see if it makes any difference in the pour.

John,
I've owned an S1 for just over 3 years. The pre-infusion mod is the best thing I've done for my shots in two years. The shots are MUCH more consistent in flavor and mouthfeel. I get a good bit less bitterness now. I wonder if it may be that the preinfusion chamber on my version I machine plus the initial hotter water out of the brewpath?
Anyhow, it is DEFINITELY a worthwhile upgrade.

BTW, my second biggest improvement came from the WDT. Thanks!

Jeff

Thanks John for this very helpful look at these two competing configurations. As someone planning an upgrade from Silvia, your analysis gives me a little more to contemplate.

Right now, the Quick Mill Alexia ranks high on my list, so I'm wondering about the following:

RapidCoffee wrote:Although single boiler espresso machines may function very well for straight espresso shots (no milk), they are marginal for producing even one milk-based espresso beverage.

Never having used an HX or DB machine, I haven't felt too inconvenienced when making one cappuccino with Silvia. Of course there's a waiting period while the machine comes up to steaming temp, but that gives me just enough time to get my pitcher set up. Yes, my crema is subsiding during this period, and the espresso is cooling off. But then my milk stays cold until just before the steam hits it. Does any of this affect the final beverage much? I'm not sure. Maybe I'm not according enough respect to milky drinks. But I'm curious, do those with HX and DB machines making just one drink at home feel the instant steam is significant? Or just a minor convenience?

Obviously, the single boiler design is highly compromised for making back to back milk-based drinks, or even following one with an espresso. No argument there.

Like others, I'm concerned about the temp surfing requirements of HX designs. I still anticipate a flush with Alexia or a DB machine, but it's helpful to know that you sometimes flush more with the S1. As someone who uses bottled water exclusively (due to marginal tap water), conservation is important to me. I would have thought HXs perform relatively poorly on this front, but that's obviously not true in all cases.

Yet I still find myself leaning away from HXs. One reason is the reproducibility issue you describe here:

RapidCoffee wrote:However, the DB flush volume is relatively unimportant, whereas it's critical to monitor the flush volume closely on an HX machine. This is not difficult, but it does require paying attention to either the water dance or the grouphead thermometer.

* Reproducibility is easier on the DB. The S1 (presumably) produces a flat brew temperature profile, pour after pour. Nailing the same humped profile on the Vetrano is a hit-or-miss proposition that depends on the idle period, flush volume, and delay prior to pulling the shot.

The other is brew temperature experimentation. I was surprised you found in favor of HX here.

RapidCoffee wrote:* The HX design lends itself very nicely to experimenting with different brew temperatures. Sure, you can program in a new brew temperature on the S1... and then wait for the machine to stabilize at the new temperature setting.** To change brew temperature on the Vetrano, all you need is a longer or shorter flush.

I would think it's somewhat laborious to develop different flush routines to nail various specific temperatures consistently. On most temperature controllers, changing the SV is a very simple matter (not so on the S1, apparently). Yes, there's a lag period while the machine stabilizes to a new temp. But that seems to me preferable to analyzing the effects of different flush routines. I don't think this is a huge deal, as with a grouphead thermometer you can get valuable feedback, and "a little hotter" or "a little cooler" should be easy enough. Still, I like the idea of punching in a different number while trying to optimize the brew for new beans.

I'll be curious to learn which machine you keep. Thanks for writing this up.

Jack

John

Here are some views from a recent convert to home espresso.

I wish I had the analytical approach then that I have now towards purchasing a machine. I had invested in a french press, vac pot, drip coffee maker and what I thought was a great burr grinder that made a discernible difference in my brewed coffee. When I decided to go to home espresso after partaking in great espresso at Blue Bottle Cafe in SF this past summer. I really had no idea of costs and would have laughed out loud if someone told me that I would be spending 2 grand for espresso.

In the end I did what I thought was sufficient homework and dismissed Silvia and then Alexia as not being sufficient for my long range plans that included decafe Cappa's for the fiance/wife. I decided on Anita and I am quite happy though with some reservations now thanks to everyone here at HB

Learning to flush correctly and reproducibly to a desired temperature was random and difficult at first. With the help of Eric's probe it is a simple task to get fairly precise predictions of brew temps from a flush protocol. What I didn't anticipate so much and something not mentioned by John is just how much water I use to make a couple of shots in the morning. and each time I use the machine after a lag time between shots. This is the case almost all the time too. Now I currently have great water piped down from the reservoirs in upstate NY. Out of the tap I measure 38 ppm and I use a Brita filter (not sure its necessary) that takes it down to 32 ppm. I assume that my machine currently will not need regular descaling. This consideration is currently a big deal to me as we are going to build our new house in a beautiful plot but one with well water. The current water (with softener - don't know what kind but it uses salt) is 192 ppm and with a Brita filter it goes down to 102 ppm. We will need to dig a deeper more efficient well but I suspect that the water will be similar and we will need further treatment for espresso and probably general home use.

I was told when I bought my Anita that I would probably want and probably should buy a DB machine. I kind of laughed and could not conceive of spending more then I already was planning. However, they were probably right but I prefer a new house to a new espresso machine. The water requirements should be considered in my opinion for a newbie making a decision. My espresso with the HX is great so in the end my decision to go a step up from entry level was a wise one that I would recommend to those who can afford to make that commitment.

Very timely post, you examined half of my short list for upgrade. Fluctuations of the Euro/Dollar rate have something to say about it, but I digress.

I perused the HX temperature surfing posts, and while I'd have to examine the raw data to be sure, the graphs produced by erics seemed to indicate that a flush down on a Vetrano (long flush to ~183 degf, and recovery to ~197 for a brew temp approx 4 degf higher) seemed to produce a flatter temperature curve that sustained the curve shape for longer extractions. A shorter flush and go still produced a good profile but there appeared to be a couple more degrees (1-2) variation in temperature, and it appeared to tail off at the end of the pull. The group temp fluctuated much more for the flush and go method. Again, it is unknown if the scace temperature drop at the end was due to the shot being terminated, but the flush and recover grouphead and scace temperatures had a much smoother profile.

That said the results in the cup were a little surprising to me as I would have expected the S1 to produce a the brighter cup with the more stable temperature, but the slightly greater degree of fine tuning the Vetrano permits may have resulted in the development of the brighter notes.

Using the erics thermometer adapter and a temperature based flush & recover seems to produce a very repeatable temperature curve. The variable would be the point in the boiler heating cycle, but since you are monitoring the thermometer for group head temperature, that variable would seem to minimized.

I'd be curious how the S1 performs against DB machines with larger boilers, and insulated boilers, such as the Brewtus III and Vibiemme Domobar DB. I'm sure we're talking about very small differences in performance.

The threads addressing this convinced me to push the Vetrano to the top of my short list.

Thank you very much for testing effort. I'm counting the days until I can put either machine on my counter.

As someone new to home espresso I found a double boiler to be an excellent crutch for learning at least, and I can't say that I'm too displeased with it even after that.

Trying to figure out the correct dosing, tamp, etc is made harder when you have to worry about precisely how much water to flush through the group to get the correct temperature. You basically have to worry about all relevant variables at once. With a newer DB, you set the PID temperature and forget about it

With a double boiler you need to flush more to keep the grouphead up to temp than an HX, but you can continue flushing without detriment, your water stays the same temperature. Flush too much with an HX as I understand it and you're brewing cold.

EricL wrote:...the graphs produced by erics seemed to indicate that a flush down on a Vetrano (long flush to ~183 degf, and recovery to ~197 for a brew temp approx 4 degf higher) seemed to produce a flatter temperature curve that sustained the curve shape for longer extractions.
...
That said the results in the cup were a little surprising to me as I would have expected the S1 to produce a the brighter cup with the more stable temperature, but the slightly greater degree of fine tuning the Vetrano permits may have resulted in the development of the brighter notes.

The longer flush may indeed produce a flatter profile, but I'm too impatient for an extended flush regimen. Most of my Vetrano pulls are flush & go, or perhaps with a short wait (<20 seconds).

Describing taste accurately is still a challenge for me. My intent was to convey the greater range of flavors in the Vetrano pulls, including more sourness/acidity/brightness. This makes sense if there is a wider range of brew temperatures during the extraction. In concentrated form, some of those flavors become too intense, hence the (small) preference for S1 straight espresso shots. When diluted in an Americano or cappuccino, they added an interesting complexity to the brew.

There are many ways to manipulate taste, including blend/grind/dose as well as pressure and temperature. I imagine that for some beans and temperature settings, my preferences might change.