Over a year ago, I developed a technique called staccato espresso where I used a sifter to separate the coffee grounds into fine, mid, and coarse. Then I made a layered (staccato) shot by first putting the fine layer, then the coarse, and then the mid. I found each layer provided certain flavors contributing to the overall flavor, and through this method, I got a better tasting espresso shot. Then I started thinking crazy.

一年多以前，我开发了一种称为staccato espresso的技术，在该技术中，我使用了筛分器将咖啡渣分为细，中和粗粒。 然后我先放上一层，然后再放上粗糙的，再放中间，进行分层(断音)拍摄。 我发现每一层都提供了一定的风味，有助于整体风味，通过这种方法，我得到了更好的品尝意式特浓咖啡。 然后我开始发疯。

I started thinking about coffee roasting. I didn’t like single origin coffees because I found them not to be balanced, and I had been roasting at home for three years. Typically, I would blend a bean from Africa and a bean from South America to produce a great blend. I blended before roasting because I didn’t want to invest the time into roasting two separate times and blending after the roast.

我开始考虑咖啡烘焙。 我不喜欢单一来源的咖啡，因为我发现它们之间不平衡，而且我已经在家烘焙了三年。 通常，我将非洲的一种豆子和南美的一种豆子混合在一起，制成很好的混合物。 我在烤之前要混合，因为我不想花时间在两次烤上然后在烤后混合。

However, a natural extension to separating particle sizes is to separate coffee blends to separate bean groups. I couldn’t resist making the process more complicated because at that point, I had the most complex espresso preparation routine. I present the history, the technique in both staccato espresso shots and staccato tamped shots, and show some data collected over the months to show how it produces a better shot in taste and EY.

但是，分离粒度的自然扩展是将咖啡混合物分离为不同的咖啡豆组。 我无法忍受使过程变得更复杂，因为那时候，我有最复杂的意式浓缩咖啡制备程序。 我介绍了断断续续意式浓咖啡拍摄和断断续续夯实拍摄的历史，技术，并展示了过去几个月收集的一些数据，以显示其如何在口感和EY方面产生更好的效果。

技术发展 (Technique Development)

To experiment, I roasted a typical blend of two bean types separately. I kept the blend separated, and I ground and sifted them separately. I started experimenting with a six layer staccato shot: 2 fine layers, 2 coarse layers, and 2 mid layers. I examined taste by alternately which layer came first, and which layers benefited the most from being split. I found only the fine layer needed to be separated, but which layer came first had an impact on how beneficial the technique was.

为了进行实验，我分别烘烤了两种豆类的典型混合物。 我将混合物分开，然后分别研磨和筛分。 我开始尝试六层断断续续拍摄：2个精细层，2个粗糙层和2个中间层。 我轮流检查口味，首先是哪一层，哪一层得益于拆分。 我发现只有细小层需要分离，但是最先出现的那一层会影响该技术的优势。

In a first experiment, I roasted Colombia Honey Process Buenos Aires Gesha and Rwanda Dry Process Rusizi Nyakarenzo separately. I then ground and sifted them separately, and I started testing. I immediately saw score differences. I didn’t know how significant they were, and at the time I started splitting all my shots, I would only change the orientation in the first two shots of the roast.

在第一个实验中，我分别烘烤了哥伦比亚蜂蜜Craft.io布宜诺斯艾利斯Gesha和卢旺达干Craft.ioRusizi Nyakarenzo。 然后我将它们分别磨碎并过筛，然后开始测试。 我立即看到分数差异。 我不知道它们有多重要，当我开始分割所有镜头时，我只会在烘烤的前两个镜头中改变方向。

I pulled the data I had across a few different roasts, and it is not enough data to draw any strong conclusions, but there is certainly variability across roasts. Here is a head to head:

我从几个不同的烘焙中提取了我的数据，虽然数据不足以得出任何有力的结论，但是各个烘焙之间肯定存在差异。 这里是一对一：

This comparison looks at No Split vs the two different orientations:

该比较着眼于“无分裂”与两个不同方向：

交叉应用到Staccato捣固 (Cross-Application to Staccato Tamping)

Staccato tamping was a derivative of the staccato shot. I was working with some darker roasted coffee from Hawaii, and I found putting all the grounds in the filter to be too messy. So I started putting only half, distributing, and tamping, followed by the other half. At the same time, I was developing a deeper understanding of staccato espresso, and I discovered the mid layer (top layer in staccato shot) was the key layer to determine how well the shot donuted or not based on its tamp.

断断续续捣固是断断续续射击的衍生形式。 我当时正在处理一些来自夏威夷的深色烘焙咖啡，但发现将所有碎屑放在过滤器中太乱了。 所以我开始只放一半，分配和夯实，然后再放另一半。 同时，我正在深入了解断层意式浓缩咖啡，我发现中间层( 断断续续射击中的顶层 )是确定击打效果的关键层。

I experimented with doing a harder tamp on the first half of a regular shot followed by a much lighter tamp for the top half, and the staccato tamping technique was born. I started tracking the bottom layer pressure, and I found somewhere between 300g and 500g of pressure you be optimal for the bottom layer. For the top layer, I use a leveler, and I track the depth of the leveler.

我尝试在常规射击的前半部分进行更硬的夯实，然后在上半部分进行更轻的夯实，断断续续夯实技术诞生了。 我开始跟踪底层压力，发现300g至500g之间的压力最适合底层。 对于顶层，我使用矫直机，并跟踪矫直机的深度。

After I felt more secure in the technique, I started splitting the roast for staccato tamped shots resulting in four layers. I would weigh and distribute the first two layers, tamp, weight and distribute the second two layers, and level.

在我对这项技术感到更加安全之后，我开始将烤肉分割成断断续续的捣实镜头，形成四层。 我将称重并分配前两层，夯实，重量并分配后两层和水平。

For Staccato Tamped preparation, I also found a benefit in splitting the beans. That preparation involves two layers, top and bottom, where the bottom layer is tamped harder than the first. Similarly to splitting a staccato espresso shot, the top layer doesn’t need to be split. I typically split it just for simplicity sake.

对于Staccato夯实的准备，我还发现了切碎豆类的好处。 该准备工作涉及顶层和底层两层，其中底层比第一层更坚固。 与拆分断层意式特浓咖啡类似，顶层无需拆分。 为了简化起见，我通常将其拆分。

数据与分析 (Data and Analysis)

I collected this data over the course of a year, but the difficulty was making sure I had controlled samples. A lot has changed in my technique in the past year, and I would take some samples with and without splitting on a regular basis. Recently, I curated the dataset, and I found I had a reasonable amount of samples to convince myself that the phenomena was not due to confirmation bias on my part, but it was observable.

我在一年的时间里收集了这些数据，但是困难在于确保我控制了样本。 在过去的一年中，我的技术发生了很多变化，我会定期取样和不取样进行取样。 最近，我整理了数据集，发现有相当数量的样本可以说服自己，这种现象不是由于我这一方面的确认偏差，而是可以观察到的。

绩效指标 (Metrics of Performance)

I used two metrics for evaluating the differences between shots: Final Score and Coffee Extraction.

我使用了两个指标来评估镜头之间的差异： 最终得分和咖啡提取 。

Final score is the average of a scorecard of 7 metrics (Sharp, Rich, Syrup, Sweet, Sour, Bitter, and Aftertaste). These scores were subjective, of course, but they were calibrated to my tastes and helped me improve my shots. There is some variation in the scores. My aim was to be consistent for each metric, but some times the granularity was difficult and affected the final score.

最终分数是7个指标(“锋利”，“丰富”，“糖浆”，“甜”，“酸”，“苦”和“回味”)的记分卡的平均值。 这些分数固然是主观的，但是根据我的口味进行了校准，并帮助我改善了投篮。 分数有所不同。 我的目标是使每个指标保持一致，但有时粒度很困难并且影响最终分数。

Total Dissolved Solids (TDS) is measured using a refractometer, and this number is used to determine the percentage of coffee extracted into the cup in combined with the output weight of the shot and the input weight of the coffee, called Extraction Yield (EY).

使用折光仪测量总溶解固体(TDS)，此数字用于确定提取到杯子中的咖啡的百分比以及小球的输出重量和咖啡的输入重量，称为提取率(EY) 。

分割分析 (Split Analysis)

Looking at some scatter plots of all the data, it is difficult to see a separation of regular and split shots. This is due to a data imbalance due to my data collection.

查看所有数据的一些散点图，很难看到常规镜头和分割镜头的分离。 这是由于我的数据收集导致数据不平衡。

So let’s reduce this data to just the best pairs of shots for each roast, regular and split.

因此，让我们将这些数据简化为每次烘烤，常规和分割的最佳拍摄对。

The pattern seems a little clear, but we have paired data, so let’s compare shot per shot. For taste (Final Score), split shots seem improve over regular shots, and for EY, they are often better.

模式似乎有些清晰，但是我们已经将数据配对，因此让我们比较每张照片。 对于品味(最终得分)，分割射击似乎比常规射击更好，而对安永而言，通常更好。

When viewed on a line chart instead of a scatter plot, we can sort the shots and see how they perform. There are some anomalies, but it isn’t random.

在折线图而不是散点图上查看时，我们可以对镜头进行排序并查看其效果。 有一些异常，但不是随机的。

We can look at a statistical test to understand if these differences in results are statistically significant. For taste, they are, but EY is still a bit too close in distribution. More data is needed, but at the same time, I’ve been working to increase EY, so my hope is that the taste is improving even if EY stays the same. I don’t have enough data to show a trend one way or another.

我们可以看一下统计测试来了解这些结果差异是否具有统计学显着性 。 就口味而言，它们是一致的，但安永在分销方面仍然过于紧密。 需要更多数据，但与此同时，我一直在努力提高EY，因此我希望即使EY保持不变，口味也会有所改善。 我没有足够的数据来以某种方式显示趋势。

We can further break this data down by roast and staccato espresso or staccato tamped. For taste, some of the roasts don’t get much of a benefit, but some do. For EY, one roast in particular is problematic (4/6/2020).

我们可以通过烤和断断续续的意式浓缩咖啡或捣固的断断续续进一步细分此数据。 就口味而言，有些烧烤并没有太大的好处，但有些可以。 对于安永来说，尤其是一次烤是有问题的(4/6/2020)。

For the 4/6/2020 roast, two of these shots split the roast based on the top or the bottom but not a split in both the top and bottom layers. So it was 9 grams of one roast, 9 grams of the other, instead of 4.5 bean 1, 4.5 bean 2 on the bottom and the same on the top. Removing those two samples, the EY now sees a statistically significant improvement (the threshold is a p-Value < 0.05).

对于4/6/2020烘烤，这些镜头中有两个是根据顶层或底层对烘烤进行分割的，而不是对顶层和底层进行分割的。 因此，一次烤9克，另一种烤9克，而不是底部的4.5豆1，顶部的4.5豆2。 除去这两个样本，EY现在在统计学上有显着改善(阈值为p值<0.05)。

We can expand the data to compare the same regular shots multiple times to different split shots because the number of split shots was much greater than regular shots. Not all split shots were great. I was merely comparing the best split shots to the regular shots. Some of these shots had lower Final Score and/or EY because they were involved in other experiments such as heat beans, cooling grinds, or modifying the distributor.

我们可以扩展数据以将相同的常规拍摄多次与不同的拆分拍摄进行比较，因为拆分拍摄的数量远大于常规拍摄的数量。 并非所有拆分镜头都很棒。 我只是在比较最佳分割镜头和常规镜头。 其中一些镜头的最终得分和/或EY较低，因为它们参与了其他实验，例如热豆，冷却磨或修改分配器。

完整性检查 (Sanity Check)

One of the variables that is not controlled, as well as I would have liked, is the output to input ratio. Ideally, this would be the same for every shot, but part of the artistry is ending the shot visually. Some people use a scale, but I’ve been using a measuring cup where I knew the volume range that would give the desired output level. At the start of this collection, the ratio was typically 0.9 to 1.1, and towards the end, the ratio has drifted to 1.2 to 1.4.

正如我所希望的那样，不受控制的变量之一是输出与输入之比。 理想情况下，每张照片都应该相同，但是部分艺术性是在视觉上结束镜头。 有些人使用秤，但我一直在使用量杯，因为我知道可以提供所需输出水平的体积范围。 在开始收集时，该比例通常为0.9到1.1，到最后，该比例已漂移到1.2到1.4。

Let’s assume there is a linear relationship between EY and ratio, we can do a best-fit line as seen below. There is a variety of wiggle, but we could still use this to adjust all the EY’s to what the EY would be at the 1:1 ratio. There are more shots used for this linear relationship than the paired shots because I included as many shots as I could from those roasts even though they didn’t have a pairing.

假设EY与比率之间存在线性关系，我们可以如下所示做一条最佳拟合线。 有多种摆动，但我们仍然可以使用它来将所有EY调整为1：1的EY值。 用于这种线性关系的镜头比配对的镜头要多，因为即使没有配对，我也会从这些烤肉中添加尽可能多的镜头。

Then we could take a look at the pairs with adjusted EY’s and test for statistical significance. Again, we run into a similar issue with two shots from the 4/6/2020 roast due to them not being in the same manner of staccato tamped shots. After removing those, the difference between split and regular shots is statistically significant.

然后，我们可以查看调整后的EY对，并检验其统计显着性。 同样，我们遇到了一个类似的问题，即从2020年4月6日的烘烤中拍摄了两张照片，这是因为它们的断音方式不同。 删除这些镜头后，分割镜头和常规镜头之间的差异具有统计意义。

This results should not be too surprising because the paired shots were pulled in the same time frame, so the range of ratio was usually similar.

这个结果应该不会太令人惊讶，因为配对的镜头是在同一时间范围内拉出的，因此比率范围通常相似。

I presented a work here built on the staccato concept to improve espresso even further by splitting a roast by the bean. The result has been an improvement on taste and extraction. This split roast staccato technique has had a lasting impact on my espresso routine, and these experiments have furthered the notion that the espresso process is still mysterious with room to grow.

我在这里介绍了一种基于断断续续概念的工作，可通过将烤豆切成薄片来进一步改善意式浓咖啡。 结果改善了口味和提取。 这种拆分烤断断续续技术对我的意式浓缩咖啡常规产生了持久的影响，这些实验进一步证明了意式浓缩咖啡的过程仍然具有增长空间的神秘性。

If you like, follow me on Twitter and YouTube where I post videos of espresso shots on different machines and espresso related stuff. You can also find me on LinkedIn.

如果您愿意，请在Twitter和YouTube上关注我，在这里我会在不同的机器上发布浓咖啡拍摄的视频以及与浓咖啡相关的内容。 您也可以在LinkedIn上找到我。

我的进一步阅读： (Further readings of mine:)

Pre-infusion for Espresso: Visual Cues for Better Espresso

特浓咖啡的预输注：视觉提示可提供更好的特浓咖啡

The Shape of Coffee

咖啡的形状

To Stir or To Swirl: Better Espresso Experience

搅拌或旋流：更好的意式浓缩咖啡体验

Spicy Espresso: Grind Hot, Tamp Cold for Better Coffee

香浓意式特浓咖啡：磨碎热，捣碎冷可制得更好的咖啡

Staccato Espresso: Leveling Up Espresso

Staccato意式浓缩咖啡：升级意式浓缩咖啡

Improving Espresso with Paper Filters

使用滤纸器改善意式浓缩咖啡

Coffee Solubility in Espresso: An Initial Study

浓咖啡中的咖啡溶解度：初步研究

Staccato Tamping: Improving Espresso without a Sifter

Staccato捣固：无需筛分器即可提高浓咖啡

Espresso Simulation: First Steps in Computer Models

浓咖啡模拟：计算机模型的第一步

Pressure Pulsing for Better Espresso

压力脉冲使咖啡更浓

Coffee Data Sheet

咖啡数据表

Artisan coffee is overprice

工匠咖啡价格过高

The Tale of the Stolen Espresso Machine

被盗咖啡机的故事

Espresso filter analysis

浓缩咖啡过滤器分析

Portable Espresso: A Guide

便携式意式浓缩咖啡：指南

Kruve Sifter: An Analysis

Kruve Sifter：分析