Dialing in a new espresso machine, a step by step guide

[cannonfodder]

Help, I just got a new espresso machine and my shots are horrible!

If you frequent the forums, you have no doubt seen this reoccurring topic. Dialing in a new machine for an inexperienced Barista can be a bit daunting, especially if you have a new grinder to accompany that shiny new espresso machine.

Many novice users start with too high an expectation and the incorrect assumption that if I purchase a quality espresso machine and grinder I can automatically make good espresso. Just because you own an espresso machine does not mean you can make espresso.

It takes time and a lot of practice to develop the basic Barista skills of grind, dose and tamp. In practice it sounds simple, grind your beans, dose them into the portafilter, mash it flat and make your cup. Unfortunately, this is deceivingly simplistic. It is more akin to an algebraic formula, each variable being dependent on the others. By changing one variable, you change the dynamics of the entire equation.

Too often a new Barista goes about the process in an over exuberant fashion changing the grind, dose, tamp, brew pressure, boiler pressure, coffee blend etc... all at the same time. The key to make one change at a time, observe the results and make another small change based on your observation.

I am the new owner of an Elektra A3 espresso machine. So I decided to document my dialing in of the machine and grind in hopes of helping others with the same process. My kit did not include a new grinder; I am still using my La Cimbali Jr. grinder so I already know how my grinder will respond to a change.


I loaded up the grinder with some two week+ old coffee; this is well beyond its prime but should still have enough serviceable life in it to get started. I set my grinder on 4.5 which I knew was too coarse. These are the frightening results.

Shot one, the grinder set at 4.5, 50 pound tamp, and 20 gram dose in the stock double basket. As you can see, 20 grams was too much, the machine was choked and the extraction barely started. The pucks surface was grinding against the shower screen due to the over dosed basket.


Shot two. I kept the grind and tamp the same and reduced the dose to 18 grams. My scale was jumping between 18 and 19 grams so this was a heavy 18 gram dose. I do have to say that having read Dan's review on the A3, I already knew it did not respond well to overdosing.

Let me also add that this dose did not make heavy contact with the shower screen but was very, very close. To check that, dose and tamp your portafilter then lock it into the group. Then remove the portafilter. If you can see even the lightest impression of a screw head or other parts of the shower screen, you are still overdosed. This dose was just touching on the outer edges of the puck.

You may want to stand back from your computer for this one. You don't want to get sprayed in the eye.


Boy, that one was nasty, but we now know what happens when we change the dose from 20, to 18+ grams. If you look closely at the photo below, you can see several pinholes in the puck, those are the sources of the channeling.


Shot three. I kept the grind and tamp the same as shots one and two but further reduced the dose to 17 grams. The headspace was just about right on this one. The massive channeling was reduced although the shot was way too fast and I still had donut extractions. The perimeter of the basket starts first but we have a slow spot in the center of the basket.


After these three shots we have learned that the A3 does not like to be overdosed. With the given grind and tamp, the 17 gram dose provided adequate headspace and the best extraction of the three. The next step is to fine up the grind one notch in an attempt to slow the extraction. I start with a dose of 18 grams while keeping the distribution and tamp the same. Why go back to 18 grams? As I stated above, all of these variables are dependent on each other. Since we changed the grind, the dose may have been effected as well. So I am bracketing the extraction again to see what effect it has.

The grinder is now set at 4.25, the dose is 18 grams and the distribution and tamp remain the same. The extraction was better than the original 18 gram dose. The shot starts more even but still favors the front of the basket. It develops nicely but half way through the shot we see signs of channeling, then suddenly the flow runs fast and blond.


For the fifth shot I retain the 4.25 grind setting and down dose to 17 grams, once again keeping the distribution and tamp the same. This time the shot starts more abrupt and flows faster. We still have some channeling, the 18 gram dose worked best with the finer grind.


So after 5 shots what have we learned. The A3 does not like to be over dosed. The 17 and 18 gram dose worked best depending on the grind that was used. We also learned that two week old beans are not going to produce an acceptable shot. So for those of you that spend $2000 for you kit and decide to use cheap grocery store beans to dial in your machine, you are wasting your time. If two week old beans are too stale to produce a good shot, what do you think two month old beans will do. Please, do yourself a favor and invest in two pounds of good beans to learn with. We also discovered that my tamp favors the front of the basket. That has nothing to do with the machine or grinder, but a flaw in the Barista's (my) technique.

More to come.

[another_jim]

Nice intro to working a new machine.

BTW, you might consider sometime trying 14 grams and a grind roughly 15 degrees of a turn on the grinder collar finer; that works better for the Semi, and the group bell is the same on both machines. The donut extraction was a headache on the semi doubles too; and it gets worse the deeper the puck. If you like 18-19 gram shots, consider a triple basket; I have a friend in Evanston who uses that dose/basket on this group.

[cannonfodder]

Thanks for the heads up on the dose. The donut of death extractions are usually a symptom of too much coffee with too coarse a grind and too hard a tamp. I would not have guessed a 14 gram dose though. That is really cutting it down but the Elektra has a relatively thick dispersion block which reduces headspace.

Under normal conditions, I would have jumped to a finer grind and lower dose but that would defeat the purpose of the thread. Hopefully, this will help provide systematic flow process to assist the first time home user.

I had planned on adding a frothing guide but steaming on the A3 is so easy there is not much to show. Just stick the wand in the milk and flip the lever. No surfing, no standing on one foot while angling the pitcher 15 degrees while chanting 'be one with the milk'. Just stick the wand in the milk and go. Wonderfully simple and the best wet microfoam I have ever made. I really like this machine.

I also must point out that using grams for the dose can be deceiving. Some blends are denser than other so the same weight dose may have different volumes in the basket. The trick is volume not weight. This is where I like ridged baskets. That basket ridge gives me a consistent point of measurement. The tamper piston is also a good indicator. Make note of how deep in the basket your tamper sits. That will let know if you up or down dosed and help you gain dosing consistency.

[another_jim]

I also must point out that using grams for the dose can be deceiving. Some blends are denser than other so the same weight dose may have different volumes in the basket. The trick is volume not weight. This is where I like ridged baskets. That basket ridge gives me a consistent point of measurement. The tamper piston is also a good indicator. Make note of how deep in the basket your tamper sits. That will let know if you up or down dosed and help you gain dosing consistency.

Hmm, I wonder. Cellulose is cellulose. If the coffee is denser the cause is that it has less vacuoles (air pockets) in its cell structure. Less airspace means a thin layer may resist the flow just as much as a thicker layer of a less dense bean. So it could be dosing by weight provides a more consistent taste and grind than dosing by volume. I'm not sure about any of this, but it's something that needs testing (unless I missed reading about it)

[chelya]

Great videos and descriptions. It is very educational. I am looking forward to part 2.

[edwa]

Great thread! After you finish part 2 my suggestion would be to post this in the How Tos section as a great tool for diagnosing extractions. Kind of a visual glossary.

I could use a little more explanation of the "donut" in shot 3. I have watched it over and over because it goes so fast. Any chance we could get some more detail/definition? You state that you have a slow spot in the center. That's hard for me to see (it is first thing in the morning), at what second count is it most visible?

Very thankful for all the work you obviously put into this.

[cannonfodder]

In the previous session I learned that an overdosed basket and old beans make for a disastrous mess. We had lots of channeling, uneven flow and donut extractions.

Allow me to clarify the 'donut' phrase. If you watch the first 4 seconds of the extractions you will notice that the outside of the basket starts flowing first. That outer edge extraction then quickly closes in radiating from the outer edge to the center. You will also notice a very small spot in the center of the basket that has no flow. If you watch the streams, you will notice them dancing around the center of the basket but do not converge into a single centered stream until several seconds after the extraction starts. Those dancing streams are caused in part by the lack of flow from the center of the basket.

Once the streams do converge the cone still sways and moves around the bottom of the basket. That movement is caused by an uneven flow of coffee. As one location starts to channel, or dry up, the cone will migrate toward the area of faster flow. Even in the absence of obvious channeling jets, we know the extraction is uneven based on the behavior of the stream and cone. Also watch for the twister effect. That is where the striping of the flow twists as it flows down the cone forming a 'tornado'.

For the next session we have gone to fresh beans. The blend is the same but the beans are three days out of the roaster. They are still a little fresh and still degassing which will cause a large cone to form due to the excess carbon dioxide being emitted from the beans. Another tell tale sign of too fresh a bean is the crema in the cup. A shot that consists of 2oz (coffee and crema) will quickly settle and go flat. Your two ounce shot quickly becomes a 1.25 ounce shot as the 'false' crema falls. Keep in mind that a shot, even from properly rested beans, will settle over time. A two ounce shot will settle down to about 1.75-1.5 ounces if left to sit for a minute.

So taking what we have learned from the first series, we move on. I have switched to a fresh roast, same blend but only three days out of the roaster (I home roast). Knowing that fresh beans take a courser grind, I reset my grinder to 4.5 and put in my beans. Taking a lesson from the first series, I dose my basket to 18 grams and use the same distribution and tamp as the first set.

Once again let me point out that consistency is your friend. If you are new at espresso, and if you are reading this you probably are, you may want to invest $20 in a kitchen scale that does both grams and ounces. Then you can confirm your dose is consistent at the beginning of the learning process.

Listen closely and you can hear my wife playing with the dog in the background. I forgot to mute the audio.


The donut extraction is still there but what a difference fresh beans make. The flow quickly healed up and looked good until we get to the final 10 seconds. Then it goes down the tubes again. We have channeling and the dreaded dancing twister cone.

We have learned that even with fresh beans we still have extraction problems. The flow rate of the shot was close so changing our grind will not yield any better results. The next step is to modify the dose but that will have to wait for installment 3.

[cannonfodder]

Hmm, I wonder. Cellulose is cellulose. If the coffee is denser the cause is that it has less vacuoles (air pockets) in its cell structure. Less airspace means a thin layer may resist the flow just as much as a thicker layer of a less dense bean. So it could be dosing by weight provides a more consistent taste and grind than dosing by volume. I'm not sure about any of this, but it's something that needs testing (unless I missed reading about it)

I have had two different blends vary by a gram when comparing volumetric dosing to weight. Those heavier blends are usually a lighter roast and contain a lot of high grown hard bean like Yemen or Papua New Guinea peaberry. Testing your theory may yield some interesting results.

[cafeIKE]

My scale was jumping between 18 and 19 grams so this was a heavy 18 gram dose.

I have had two different blends vary by a gram when comparing volumetric dosing to weight...

Does your scale measure fractional grams?

Digital scales have an accuracy tolerance and a count error. A scale that indicates single grams probably has at best 1 gram accuracy and a 1 count error. This gives ±1g plus ±1 count for a total error range of 4 which is far too much for a ~18g total : ~20%.

If one wants to weigh espresso doses, a scale that measure tenths of grams is more appropriate. The error count is still 4, but the magnitude is reduced by 10 for a 0.4g error in ~18g : ~2%

Additionally, accuracy changes with time and temperature, so recalibrate frequently.

That being said, I never weigh doses. Coffee is organic, changing constantly. It's easier to adjust the grinder and / or up / down dose. Not as specific numerically, but just as effective in getting a good shot.

Nice write up on the start-up.

[cannonfodder]

The scale is a single gram so the error margin is high and not ideal for exact measurements, but for the purpose of this dialing in process, it is adequate. Regardless of whether the scale displays 17 grams but actually contains 19.5 grams of coffee, as long as it consistently reads 17 grams we are OK. The point is not to precisely measure the dose, but to provide a reference point.

Normally I take a more holistic approach with my dosing. I know when it is right because it feels right. Unfortunately, that does not provide much of a reference point for someone that is trying to dial in their first espresso machine. With time and practice, the need for a scale or even a thermometer will disappear. When starting, having a reference point to provide a repeatable variable helps with initial tuning.

[cafeIKE]

The scale is a single gram so the error margin is high and not ideal for exact measurements, but for the purpose of this dialing in process, it is adequate. Regardless of whether the scale displays 17 grams but actually contains 19.5 grams of coffee, as long as it consistently reads 17 grams we are OK. The point is not to precisely measure the dose, but to provide a reference point.

Agreed the idea of a consistent reference is fine, but consistency is rarely attained. I have a couple of 5kg scales and a 100g reference weight to calibrate my 0.1g 200g scale. When set upon the 0.1g scale it has always measured 100.0 or 100.1. On the 1g scales it varies from 98 to 102 on one and 99 to 101 on the other.

Scales are more accurate near the mid point of their range. Typical range for 1g scales is 5kilos. Accuracy at sub 20g is marginal at best.

It's possible for a 1g scale to read 17 for 16.5 to 17.5 and 18 for 17.5 to 18.5, ignoring count and accuracy errors. Ergo, both a 17g and an 18g reading could actually both be 17.5 misleading the user to think they made a change when in fact they did not.

The sticking point is that some will use the posted numbers with a more / less accurate scale as a reference and be no farther ahead in getting a consistent / optimum dose.

If we are going to use numbers, we should be accurate.
If we are going to use a scale, it should be as well.

IMO, the time spent futzing with a scale is better spent learning how to adjust the dose manually by tapping the PF and how it affects the tamp and shot. Once Joe Noob has a handle on that, he's well on his way to mastering the Mano M.

If Joe Noob wants to use a scale, then get an accurate one and burn through several kilos of supermarket coffee until he can dose to ±0.25g repeatedly.


Still a great piece on getting started.
Thanks for the write up.

[cannonfodder]

Point well taken. I normally use this scale (5 pound max) to weigh out greens for roasting by the pound and ounce so super accurate measurements have not been a big issue for me. As you stated, learning to work without the safety net of a scale is not that difficult. After a few weeks even Joe Noob will grow past the scales usefulness.

Still a great piece on getting started.
Thanks for the write up.

Thank You

The next installment goes up tonight.

[Dogshot]

Wow! Congrats on the new machine - what a beauty. This thread is a great way to introduce us to the A3 in a different way, so thanks.

Re: digital scale accuracy. I dose by weight, and my technique is similar to Cannonfodder's; I add beans to 17gm, then add 4 more. I have found a level of intershot reliability doing this method that is inconsistent with the following statement about 1gm digital scale accuracy:

Digital scales have an accuracy tolerance and a count error. A scale that indicates single grams probably has at best 1 gram accuracy and a 1 count error. This gives ±1g plus ±1 count for a total error range of 4 which is far too much for a ~18g total : ~20%.

Luckily, this hypothesis is fairly easily tested. I turned on and tared my scale and added beans until the scale just tipped 17gm. Then I put those beans in another container, and re-did the same thing 5 times. Every measure was exactly the same, except for 1, which required a single extra bean to tip 17gm. I stopped testing at that point, because such low variability after 6 tries, and the level of consistency I had been getting from my shots using this method were enough to convince me that my 1gm scale achieves much better than 20% consistency between measures. Any idea why this test would not agree with the above quoted, well-informed statement? (other than that my scale is a wunder-kind?)

Mark

[dsc]

Hi cannonfodder

Excellent idea for a thread!! I'm a newbie and your posts gave me a lot of information on changing the variables to make good espresso. I just bought a few days old coffee, so I might just try out your advice, use a scale and see how it goes.

Cheers,
dsc.

[cafeIKE]

Luckily, this hypothesis is fairly easily tested. I turned on and tared my scale and added beans until the scale just tipped 17gm. Then I put those beans in another container, and re-did the same thing 5 times. Every measure was exactly the same, except for 1, which required a single extra bean to tip 17gm. I stopped testing at that point, because such low variability after 6 tries, and the level of consistency I had been getting from my shots using this method were enough to convince me that my 1gm scale achieves much better than 20% consistency between measures. Any idea why this test would not agree with the above quoted, well-informed statement? (other than that my scale is a wunder-kind?)

Mark

One minor flaw in your method is you used different beans on each test. To test the accuracy of the scale the same beans should be loaded for each test. The variablility of beans probably guarantees on some tests your count would vary ±1 bean.

There are also a time constant and acceleration effects. When I weigh 225g for roasting, if I just dribble the beans in until 225g is displayed, the value will probably hold. If I remove a few beans until it reads 224, I may have to add fewer or more beans to return to 225. If I take the same beans and dump them in as a lot, the reading will probably hold on 226. In practise, I aim for ±2, but am certainly not going to pick out a few beans if the load is 228

Of course, this is way OTT in terms of any affect on the shot or roast, but if one seeks to mitigate variables by mechanical means, the minimum possible error should be obtained.

Over the short term, a scale left on, not subject to larger loads and loaded a bean at a time will be fairly consistent. Over the longer term with time, temperature, pressure and battery voltage variations, the repeatability will diminish.

If a scale were at all required to make great espresso, I'd switch to TEA!!!

One danger of fixating on a numerical value for a dose is that it is of limited utility for a very narrow range of criteria : ONE blend at ONE roast at ONE grind on ONE day. The barista is better served learning the range of adjustment that can be effected by adjusting dose and grind and knowing when and how to adjust them in concert.

Who knows how many noobs tear the top of the puck negating all grind and tamp adjustments, forever running in an endless loop changing grind and tamp.

Perhaps the best advice for anyone with a new machine and / or grinder is to plan on binning 2 kilos of FRESH coffee over the weekend to dial in the machine[s]. If they've never made espresso before, plan on the same routine for a month. It's important to have a quantity of the SAME coffee. It's crazy to sink a grand or two into hardware and worry about a few bucks worth of coffee.

Don't even pull a shot until you can dose the PF with consistent clearance to the shower screen. Once you can do that repeatedly, pull a shot and see what it looks like. If it's in the 25~30sec range, give it a taste. If not, keep adjusting grind until it is. Then experiment by backing the grind off a notch and updosing by tapping the PF on the counter. Then go the other way and tighten up the grind until the machine chokes. Can you tamp lighter, still clear the shower screen and pull a shot? If not then that's the end stop of grind for that coffee.

Something that is seldom discussed is grinder detritus. ALL grinders retain a varying amount. If you grind a dose, pull a shot, come back in 15 minutes and grind another dose, you have a load of stale coffee in the bottom of the PF. Either grind per shot and clear the grinder by what ever means before the next dose or grind for a couple of seconds to clear the grinder.

[cannonfodder]

From the previous update we learned that an 18 gram dose with fresh beans does produce a better shot but does not resolve our slow center extraction and end of shot channeling. Looking at our previous tests we see that a reduction in the dose had a marked improvement in the extraction. So the next logical step would be to further reduce the dose and see what happens.

As I stated at the beginning of the post, grind/dose and tamp are all part of the same formula. So when we desire a different result, we must change the equation. When we change one variable we often have to compensate in another. When we reduce our dose we have to fine up the grind to compensate for the thinner puck. So keeping that in mind, I drop to a 15 gram dose and drop the grinder from 4.5 to 4.0 and retain the same tamp.


Isn't that interesting. We are getting a more even extraction in the start, the coffee flow starts relatively even across the surface of the basket. The flow was too fast and we still had some channeling at the end of the shot.

In keeping with our systematic process, the next step is to reduce the dose to 14 grams while retaining the same grind and tamp. In keeping with our formula, if we keep the grind and the tamp the same but reduce our dose, we should get an even faster flow.
That faster extraction should lead to faster blonding and more channeling. So let's see what happens.


Not as dramatic of a transition as I thought it would be. There was still more channeling than the 15 gram shot. The flow was slightly faster then the 15 gram dose and the puck was a little loose after the shot.

Now we need to slow down the extraction. We will repeat the dosing but adjust the grinder one click finer, from 4 to 3.75.

First up is the 14 gram dose, same tamp, and the grinder adjusted to 3.75


The flow has slowed as expected but we are still getting channeling at the end of the shot. The flow is slightly uneven which could be attributed to a slightly off distribution in the basket.

Now for the 15 gram shot, 3.75 grind and same tamp.


Close, so very close. The timing was better but just a little long, the flow looked better but we still had some minor channeling at the end of the shot. At this point the blemish in the shot could simply be caused by a slightly off distribution. I am not accustomed to dosing down so far so what you may be seeing is a flaw in my technique. That is something only time and practice will resolve. The cup looks and tastes pretty darn good.


We have discovered that the Elektra does not like being over dosed, the sweet spot appears to be 15 grams in the stock double basket with this blend and tamp. We have explored 20, 18, 17, 15 and 14 gram dosing and gotten the best shots from the 15 gram dose.

One thing that does bother me is that at 15 grams I am still getting a loose puck. The coffee is not setting against the shower screen. I also notice that at 17 grams I get a fairly heavy impression of the shower screen on the surface of the puck after the shot. With the 15 gram dose I am not getting any impression from the shower screen and my shots are experiencing a problem at the end of the extraction. Missing from this series is 16 grams. Could that be the magic bullet I am looking for? We have gone this far, we might as well be thorough and explore that 16 gram dose. However, that will have to wait until the next installment.

For now I have to get to my Buffalo wings and football game, go Colts.

[cannonfodder]

Perhaps the best advice for anyone with a new machine and / or grinder is to plan on binning 2 kilos of FRESH coffee over the weekend to dial in the machine[s]. If they've never made espresso before, plan on the same routine for a month. It's important to have a quantity of the SAME coffee. It's crazy to sink a grand or two into hardware and worry about a few bucks worth of coffee.

Exactly why I say one of the variables is the blend and why I am using the same blend throughout the process. Changing the blend is a major change in one of the key variables. The other is time. As we all know coffee grind changes (I hinted at that earlier by pointing out fresh beans take a courser grind) as beans age. So once you do find that magic combination that works for your particular machine, grinder, and blend, the grind will slowly shift as the bean ages.

Don't get fixated on numbers. Please, don't get fixated on the numbers. The grind and dose are for my blend on my grinder on my machine. Your settings will surely vary, even if you have an A3 and Cimbali Jr. grinder. Settings vary from machine to machine so my 4 may be your 4.75, my 15 gram dose may be 16.5 on your scale. The entire point of this post is to emphasize the need to make small adjustments, one variable at a time. Do not try to adjust your dose, tamp and grind all at the same time. Also keep in mind that distribution makes an enormous difference. Even if your dose and grind are exact, if your distribution is off your shot will not be any good. I actually find getting the dose and grind the easy part. Learning to get an even distribution and fine tuning your technique is the hard part.

[Dogshot]

One minor flaw in your method is you used different beans on each test. To test the accuracy of the scale the same beans should be loaded for each test. The variablility of beans probably guarantees on some tests your count would vary ±1 bean.

My explanation was not very clear, but I used the exact same beans for each test (with the exception of the single bean that had to be added for test #3). I agree whole-heartedly with your main points, and my primary purpose in weighing my dose is to waste as little as possible. I use the WDT, and find it a great way ensure consistency in dosing. For me, after stirring the grounds in the basket, sheering off the excess is a consistent under-dose, a single tap is the ideal dose, and 2 taps is an over-dose.

Mark

[Abe Carmeli]

Steve,

What grinder are you using? If you use the WDT on these shots and these are fresh beans, the chanelling is likely a tamping or a grinder issue.

[cafeIKE]

I actually find getting the dose and grind the easy part. Learning to get an even distribution and fine tuning your technique is the hard part.

It's important to have a tamper that allows you to judge level and head space EASILY while tamping. If your tamper piston stands several mm above the edge of the basket its much harder to get a good feel for headspace and level than if the piston edge is just a mm or so above the rim. If the showerscreen is curved, a slight curve on the piston may help. I have to pay much more attention with my flat 57mm generic than I do with my 58.25mm Reg Barber American Curve, mostly for the reasons stated.

The WDT and the M4 are two significant improvements I've made to making espresso. Another, adding an Isomac Gran Macinino as a press & drip grinder. Having an ESPRESSO ONLY grinder is something many overlook. Even if you never recover the cost of the second grinder in coffee, it's well worth it for the reduced frustration. Also clearing the previous shot detritus. These seemingly minor details could mislead one into thinking there is a problem in one's technique.

The Devil is in the Details and there are details a'plenty!

Again, a great write up.
Well worth reading by all.
Thanks for all the effort.