I wrote a couple of months ago about pressure profiling. It remains a little divisive in the industry, with strongly polarised thoughts on its use and importance. Recently I’ve been thinking more and more about it, what I think is happening to the coffee, why lever machines produce the espresso they do and what the possibilities are for modern machines that can produce a range of pressures during the shot.
I warn you now that things might get a little nerdy. I will also do my best to distinguish between what is theory, what is opinion and what I think we can tentatively accept as fact. I’m also going to make some assertions about modern espresso brewing and the problems therein. I should also add that this is the longest post I’ve ever written, and covers a lot of ground…
The Modern Espresso Recipe
This may be stating the obvious but I think it is important to go over again before we get into pressure profiling. For a long time espresso machines were built to extract coffee using a fairly high water to coffee ratio (high by today’s standards). 7g of coffee per shot was used and the resulting brew weighed (and often still weighs) around two to three times that of the coffee. Because the coffee was rarely as freshly roasted or freshly ground as today that shot was around the 25-30ml mark when prepared into a shot glass.
Modern doses, for most quality focused shops, are much higher – usually around the 20g mark1. This likely started because people were grinding to order and dosing using the basket as a volumetric measurement. Most double baskets hold (without excess settling or compression) that much coffee when filled evenly and swept flat with a finger or dosing tool.
Along with the increased dose came an increasing preference amongst baristas for more concentrated/stronger shots. I know we don’t like the word strong but it is the right word. The ratio of dissolved coffee solids to water in the cup was much higher than ever before. Baristas were using freshly roasted coffee with a lot more CO2 present so their stronger shots had the appearance of greater volume. When freshly brewed, despite using significantly less water, they often looked around 25ml. These shots often weighed as much as the coffee used (instead of twice that) so a double espresso from 20g of coffee would often weigh no more than 20g.
So far, so obvious (I think) – but it is worth taking a step back and looking at modern espresso. We are using the same machines to brew 40-50% more ground coffee with 50-70% less water. As a result most brewed and served like this is going to be under extracted. It will be dense, thick, creamy espresso – but it could be so much better. Some argue that shots like these have an unpleasant acidity that we’ve gotten used to, the way that people who deal with very dark roasted coffee might become accustomed to high levels of bitterness.
In my opinion I think we’ve adjusted the way we roast for espresso to accomodate this. We roast a little darker, we roast a little longer. When we cup our espresso roasts we notice that we’ve introduced roast flavours we don’t want – but when we brew them as slightly underextracted espresso they fall into the background. Acidity is brought up in the shot, despite being muted by roast, and we get something pretty tasty. I’m not saying we’re all serving bad espresso – I am saying that there is a clear opportunity to serve better espresso.
This also explains the double hump phenomenon – roasting darker than we should, but underextracting slightly can produce a pretty good espresso with a relatively low extraction. As we creep up from 16% to 18% we start to pick up some of the unpleasant accidental byproducts of our roasts, until that too becomes balanced as the shots fall into 19% to 20% range. I can completely understand someone’s preference for relatively underextracted shots (as many have noted), but I think overall the shots can be improved with increased extraction. I am aware this theory probably won’t make me very popular. I’m not saying people are roasting incorrectly, and I believe we’ve gotten here based on taste. Maybe I’m completely wrong on this – but it will be interesting to explore some more.
Roasting darker also helps us extract the coffee more easily, as lighter roasts are usually very difficult to extract due to their density. We wrote off these light roasts as being bad for espresso because they lacked balance – much as we wrote off dense, high grown coffees for a long time. Too much acidity, not enough sweetness. However – this is the same description I’d give to any underextracted espresso…
The only way I’ve found to extract lighter roasts in a way that is balanced is to increase the amount of brew water used – essentially to pull a longer shot. By using more brew water I am able to drag enough of the tasty solubles into the cup – the downside to this is the loss of body, texture and intensity that sets espresso apart as a brew method.
I’ll come back to all of this once I’ve gone back to pressure profiling, starting with a few observations and some thoughts on lever espresso machines.
Why Lever Espresso Machines Work So Well
There is no denying that lever machines are capable of producing some delicious and distinctive espresso. The pressure profile of a lever has two distinguishing features:
- A period at the beginning of the extraction with low brew pressure (below 3 bars)
- A period of the extraction with a declining brew pressure as the spring expands up until the end of the brew
Interestingly a number of the people I spoke to who have pressure profilers have ended up in a similar place. While the exact nature of the profiles varied these two things tended to be in common. Perhaps these people were left as muddled by the infinite possibilities (like I was) and just reverted back to a “known” profile and interpreted it in their own way. Perhaps they were all just onto something. I want to look at each of these two points, and discuss them a little more.
Preinfusion is the term we tend to use for a period of extraction where water is soaking into dry coffee without really passing through under pressure. This is done in many machines using flow restrictors or with preinfusion chambers. I consider the following statement to be pretty accurate:
The flow of brew water through the cake will be faster if there has been preinfusion, compared to immediately brewing under high pressure.
If you have a flow restricted machine next to an otherwise identical unrestricted machine and you dial in for the flow restricted machine you’ll find that same dose and grind will choke the unrestricted machine. The prevailing theory (which I think is probably right) is to do with fines migration. Immediate high pressure pushes many of the finer particles to the bottom of the puck where they create a little sandbag of resistance. Unrestricted machines will often exhibit a slow down in flow rate towards the end of the shot as the fines continue to accumulate. The hypothesis with preinfusion is that the cake swells with water, trapping the fines in place before the pressure hits. This prevents the same level of accumulation.
What is interesting is that the longer your preinfusion (to a point) – the faster the resulting flow rate. To give you an idea of preinfusion times on lever machines I’ll borrow Gwilym’s technique of an 8 second preinfusion. The rule of thumb is that you want to completely soak the cake before the brew. This is going to be a function of your dose (the depth of the puck) and the preinfusion pressure. 20g at 2 bars takes around 8 seconds.
This means that you can grind your coffee finer, for the same dose, when using preinfusion. This, I think, is excellent news if you are brewing a modern recipe where you need to extract a lot of coffee without using too much water.
The Ramp Down
The reduction in pressure in the middle to end of the espresso is an interesting one that is still causing me some problems. I really ought to use diagrams to explain this better. The following diagram is not accurate – merely a mixture of an approximation of what is happening, mixed with a bit of a guess.
The diagram shows extraction on the vertical. I’ve highlighted the range between 18% and 21% where I think good extraction lies. The horizontal axis is brew time. The blue line represents a good 9 bars extraction (though not accurately – I know Vince Fedele has done accurate work on this!). Obviously extraction is much faster at the start of the shot (we know that from the colour of the first part of the espresso alone). As we extract the resulting extract is getting weaker (growing paler) and our rate of extraction decreases. In most good espressos the window of tasty is between 25 and 30 seconds – correlating to the range of tasty extraction on the diagram.
With decreased pressure comes a decreased flow rate. My first thought on why this might be good is illustrated on the diagram. As the pressure decreases the rate of liquid hitting the cup slows, so therefore the rate of extraction would decrease. Perhaps this would extend the window of tastiness wider? Perhaps this would make it easier to serve better shots?
This may be the case but something else popped into my head. The liquid flowing through the coffee puck would be moving slower and therefore be carrying more solvents with it. This would mean that if we look at mass in the cup, and not time, then we’d see a different diagram. Again extraction % is on the vertical, but on the horizontal there is the mass in grams of a single espresso. Also again – this is not accurate, merely a visual aid!
This diagram is a little confusing, because it is weird to think about espresso brewing and not be thinking about time. Imagine you’re watching the scale you put under your spouts (because doing that – evenly occasionally - is a good thing.) With a ramp down the extraction is increasing at a faster rate per gram of espresso, compared to a straight 9 bars shot.
For an accurate chart showing TDS, Extraction and espresso mass I am grateful to Vince Fedele. This chart represents what should be a pretty tasty espresso – brewed under standard pressure conditions.
So far the above idea is based on my own empirical observations, and my increasing need to better understand the effect of flow rate through coffee.
Back to the Lever Machine
So – does the theory match my own observations of shots from lever machines? Yes. I noticed, though didn’t give much thought to it at the time, that the shots that Gwilym would pull me on his lever were surprisingly strong and balanced. (Charting above 13% strength and being tasty is surprisingly rare!)
The theory would be that lever profiles produce better extraction when using a large amount of coffee and a small amount of brew water – exactly what most quality focused shops around the world are trying to do!
So should we all be using lever machines? Should all pressure profiling machines be dialled into to exactly replicate that curve? I don’t think so – unless you want a good solution immediately with a minimum of work. The lever profile is something of an accident resulting from the initial design of the machines. You have to have a period of preinfusion because you need to fill the chamber above the coffee with brew water. The expansion of spring probably wasn’t considered to be interesting or important early on. It was just a way of exerting more pressure than steam, line pressure or air had been able to achieve in previous machine designs. Put simply – a spring wasn’t built into the design because a decline in pressure was considered important. It was more an economic way to achieve very high pressures in a simple, mechanical way. I think there is a great need to better understand the soak/preinfusion phase as well as the ramp down phase too.
It would have probably seemed quite defeatist if I’d opened this post with the idea that the main benefit of pressure profiling would be that we could pull dense, strong but balanced shots of espresso. It seems so anti climactic, especially when pressure profiling offers such possibilities to explore. When you can do anything the idea quickly emerges that hidden somewhere out there – amongst infinite squiggly curves and shapes of profiles – there is something totally new, something that will blow our minds. I’m not so sure, and I’m actually pretty excited about the increased extraction idea for another reason:
Introducing denser, difficult to extract, coffees to this kind of profile means that we might find it easier to achieve an excellent extraction of light roasted coffees. Extracting them more efficiently we may well get great balance, and all the strength and mouthfeel of great espresso – with the benefit of an even cleaner finish. We might be able to roast all our coffees for espresso a touch lighter, keep a little more complexity in the coffee, bring out even more sweetness from our shots. That is pretty exciting to me.
Where do we go from here?
I remember writing in that last post on profiling that it would be good to get a little cohesive, collaborative work on this front. I’m going to propose a couple of experiments that I think would be immensely useful because we’d start to understand how we are influencing our espresso brewing. I certainly think some good results can be obtained from focusing on those two elements of the brew. However, I’d rather open the floor up to some feedback. I think I’ve written enough for now!
- I’ll be honest and admit my cynicism about dose. Most people who say they use 18g are using 20g and most who say 20g are using 22g+ [↩]