AssafL wrote: The article you quoted discusses fingering flows - it isn't channeling. As I understand it any flow through a packed medium is fingering flows.

Agreed, but that doesn't mean fingers are good. You would want them minimized because localized extraction starts in the fingers and they therefore become more extracted than the "rest" of the puck at the end of the shot.

I accept that the flow itself will be composed of many fingers meandering downward through the puck in concert, but I guess my hope (and interneral measure of success) is that they all hit the basket as closely as can be achieved for the above reasons. My earlier point of simply waiting for all the fingers to hit the basket before engagong the pump may help with extraction, but I doubt it. If finger #1 hits the basket first and starts dripping before fingers 2 through 17, those solids that have been stripped from the path of finger #1 will provide a preferential path for the pressurized water when the pump is turned on. It's not channeling, but it's also not good for the overall extraction. If all the fingers reach the bottom at nearly the same time and converge with one another, then pressurized flow through the puck should be more even and extraction should benefit.

AssafL wrote: [Edit: One concept we developed above is the compressed dry puck causing spritzers. Could finger flows be the cause of spritzers? PI ended too soon, few (if any) of the fingers made their way to the basket - and 9 bar is applied (too soon!!!). The only way for water to flow is the open finger. So it sprays water until a few more fingers open up and the flow is now directed across multiple fingers (which I assume converge eventually). A hypothesis...]

Possible. Since you rightly draw a line between fingers and channels, I'm not sure I buy that a flow finger should spurt, but perhaps a developed finger that has the puck collapse around it due to excessive pressure applied before saturation would turn into a garden hose nozzle... maybe that's all channeling really is? (I don't think so. Fractured pucks make channels and they are NOT just over active fingers...)

AssafL wrote: Chimera does more or less what the Decent does (as far as I understand what Decent does)

Once you get (maybe you have already) your flow profiling PID dialed in, I agree. One thing decent does is allow for a flow profile with a user defined peak pressure. Therefore one can say constant 2ml/sec for duration of shot, but don't exceed 8 bar, so the flow will drop off until the puck erodes, allowing pressure to creep back down...

AssafL wrote: Why should erosion be minimized? Isn't it the eroded stuff which gets emulsified or solubilized into the cup?

Yes. But if liquid is going into the cup, this is happening. I tend to think of the puck as a bed of resistors in parallel stacked in many layers in series. The wider the puck, the more resistors in parallel and the lower the resistance. The thicker the puck, the more resistors are in series (each "layer" added on top of the puck gives another parallel sea of resistors sitting atop the layer beneath) and the higher the resistance. The funny thing about puck resistors is that they shed resistance as their soluble compounds are stripped off. None if this is news, but here's why I think it's relevant to this conversation:

If there are individual fingered flows that start off first, their resistance starts dropping, and flow through the rest of the puck suffers. This may or may not present itself as spurts or gushers, but at the very least, it's localized regions of the puck that are more extracted than others. From the resistor analogy, if certain resistors in our web of parallel resistors begin shedding resistance before the rest of the web, the overall resistance of the web drops more quickly than if the flow had been even from the start. For this reason (uneven extraction leading to reduced puck resistance) given a grind and dose that yields your desired beverage amount in your desired shot time, the puck that erodes at the lowest rate is the puck that extracts most evenly. Do you agree?

AssafL wrote: I don't know what the fuss with nutating is...

I think that based on the above theory, the real-time feedback from decent has shown that nutating has the ability to reduce the rate of erosion for a given grind and dose. You still have to come up with a grind and dose that works, but we have data that suggests (I know, it's one data point...) that nutating is something that is worth further investigating.

Why?

I think that gentle and complete preinfusion changes the calculus. No longer are we left with the puck compression that we generated when prepping our basket. We now can let water trickle through the puck and expand it into a relaxed, yet compacted state. We don't allow for air to enter the puck, but it swells with water and reaches some sort of equilibrium state as the the puck wicks water into it and expands. This isn't some puck utopia, as a poorly distributed puck remains poorly distributed, but I think that anything we can do to improve distribution, including compacting the crap out of the puck by nutating, may have potential benefits that weren't realized without some sort of tool to objectively evaluate the performance of the puck. We're likely in the land of diminishing returns here, and only blind taste tests will tell the whole story. Nonetheless, I still think that techniques that measurably reduce the rate of puck erosion while producing the desired yield in the cup do so by decreasing variability of extraction within the puck.

AssafL wrote: BTW - If the ideal puck is uniform density, I fail to see how a nutating movement won't ruin it. I have a very accurate fitting tamper - so it doesn't necessarily do much damage like a small diameter tamper would. But why do it?

I don't know the answer here, and I agree with your general point of view. One possible advantage is that nutating moves the grinds. Much like our wedges shuffle the grinds across the top of the puck and leverage weak spots in the distribution to incorporate the grinds on top into the puck, nutating disrupts the grounds (particularly at the edges, where my flows generally start) and collapses voids that flat tamps don't. With a quick shot profile, this would be annoying, and we would grind coarser to bring shot timing back into line. But we just wait for the puck to be saturated and fire away. I've been using my wedge and nutating and the results are improved in how the flow looks. But I've still got a long ways to go before I can claim to taste a major improvement in flavor. Remember, I don't even have temperature feedback at the group yet, so any claims I make to taste or consistency should be taken with a grain of salt

AssafL wrote: I am not sure I understand the question.

My mistake. The question was in relation to varying headspace in 3 double baskets resulting in vastly different outputs over the course of a 25 second shot. Basically the question is as follows:

Given a fixed time to pull a shot, would a larger headspace that takes a longer time to fill result in decreased volume versus a smaller headspace that fills quicker?

We know that the fully saturated puck will flow faster than the prematurely pressurized puck. The question becomes "Will it pour faster enough to make up for the time spent filling the larger headspace and catch up to the puck with less headspace?"

Given an unlimited time, the answer is of course "yes". Again my question, which was rhetorical in nature, was if it was differences in headspace that accounted for wildly varying volumes in a fixed dose, fixed grind, fixed time shot, was it the larger headspace that yielded the higher volume, or the lower volume? In particular, which of the 3 double baskets tested in the study had the larger headspace, and could headspace have been factored into the model and yielded predictable outcomes?

Cheers!

- Jake

Speaking of hubris,

Many of my assumptions were placed squarely on their heads by finding yet another of Schulman's research projects here. Frankly, I think this should be required reading for the entire membership body.

This, coupled with all the refractometer and EKspresso conversations, and everything else floating around got me to wondering what it is I'm really after. Is it time to throw away my SJ and get an EK43 with SSP burrs? What's my level of compromise? What is my goal ?

Then I remembered that I want my espresso to taste good.

The reading from Schulman found a fascinating relationship between %EY and puck thickness that said "Want more extraction? Drop the dose or increase the diameter of the bottom of your basket.". Basically, dose, in grams divided by the diameter of the hole area squared was a darned good indicator of %EY. The reading doesn't get into the why, but it does explain why under extracted coffee tastes sour and over extracted coffee tastes bitter, and what the barista can do directly to impact these flavors.

Wait, what? There's such a thing as over extracted? Well, I think for me the best thing to focus on is that sour compounds come out first, then the woody, grainy Maillard compounds before the sweeter ones finally come out and you want to get all the sweet ones you can. Keep going too long and the bitter ones come out. Not that higher %EY in itself makes coffee bad, but that pulling a shot long past blonding to try and chase the refractometer will not make better espresso. There is a reason 30 seconds is a good ballpark number. Long enough for the good bits to dissolve, short enough that the bad bits stay locked in the bean fragments.

So what is it that we're trying to do with all this fancy schmancy stuff? I think it comes down to redefining the relationship that Schulman reached in 2007 that says "If you want a more extracted (and good) espresso, you need a thinner puck", to say "If you want a more extracted (and good) espresso, you need to grind finer and find a way to still extract in 30-ish seconds", and also "Are you sure you really want a more extracted espresso?". From a technical standpoint, preinfusion accomplishes the former. Finding ways to limit the flow through the puck also helps. Dynamically changing the flow through the puck at various stages of the shot may further help. Levers achieve a marked flavor profile that is largely attributed to their declining pressure profile. A fancy pump driven machine can likely mimick this flavor profile. But what else is possible? What other secrets are in the bean? What does a rising pressure profile taste like? How about a decreasing flow profile?

These are questions that are only answered when folks get creative and try new things. That said, I trot into new territory (for me, anyway ) with the utmost respect and appreciation for those who've paved the way before me. Nothing learned through rigorous application of the scientific method is to be thrown aside because new and shiny ideas have arrived. We must accept established facts as facts, but understand the underlying framework on which those facts were predicated and ask if new technology, or better grinders, or anything else that's changed has provided an avenue to expand upon what has previously been established. This is an exciting time to be in this community, and I look forward to what we'll learn with decent, the mono flat max, and other advances in our quest.

Cheers!

- Jake

Recently I tried updosing an entire gram (whoa!).

To counter the limited headspace (which would fill up and the pressure would rise before Puck was throughly infused) I reduced the flow to about 40ml/min. That gave enough time for the puck to drink without the pressure climbing.

When PI was complete and coffee appeared under the basket I raised the pressure to 9bar. Flow was essentially the same as a standard 18gr dose.

Could the reason updosing needs a coarser grind because of the headspace? And not the flow dynamics?

AssafL wrote:Recently I tried updosing an entire gram (whoa!).

You rebel!

AssafL wrote: Could the reason updosing needs a coarser grind because of the headspace? And not the flow dynamics?

I think yes and no.

In a fixed PI machine, the answer appears to be yes, but really puck height I think is what matters. Since a fixed PI machine has a PI duration that is tied to headspace, it makes sense that the issue of grinding coarser is because of the reduced headspace. But it's compounded because a larger dose needs more PI time because of the thicker puck, but the machine gives it less because of the reduced headspace. As you mentioned earlier, fixed PI machines should naturally have a sweet spot [for a given basket] that optimizes head space to puck height and yields more complete preinfusion. Tying it to basket geometry is my tweak on your idea. A triple basket should be fine on an E61 as long as there is enough headspace to match the thickness of the puck...

Your experiment suggests exactly what you have postulated here and elsewhere, which is (paraphrased) that as long as the puck finishes drinking before the full pressure ramp, you should be able to grind finer and have an excellent extraction. I buy into this big time. With machines capable of gradually saturating the puck, you eliminate the boundary conditions between wet and dry layers in the puck that cause hydraulic collapse of the dry puck and require coarser grinding. Complete puck saturation before pressurizing gives you "the VST effect" of flow dynamics that mimick a much lower dose. My personal (and unfounded) thought on VST is that they do this by having a thinner puck than other baskets because their bases are much wider. Since their pucks are thinner, and additional headspace isn't necessary with a thinner puck, their baskets can be shorter. (The basket shootout shows that their baskets don't hold any more than their old school counterparts when loaded to the brim, so I don't think they offer more headspace than the competition to get their increased flow...) I bring up VST because their claim of improving extraction appears (to me, at least) to be related to the dose-over-surface area relationship, which is why grind has been tied so closely to dose and triples have long been under extracted as a consequence.

Jake_G wrote:I think it comes down to redefining the relationship that Schulman reached in 2007 that says "If you want a more extracted (and good) espresso, you need a thinner puck", to say "If you want a more extracted (and good) espresso, you need to grind finer and find a way to still extract in 30-ish seconds", and also "Are you sure you really want a more extracted espresso?".

I think we are bending the rules and exploring new new territory by thinking about things in terms of complete preinfusion. Call it the lever effect, or the slayer effect, but methinks this is what it's doing. I think we will find that with the correct preinfusion, we can set the grind wherever we want it to optimize extraction for flavor and dose up to whatever we can fit in the basket to obtain our desired yield before blonding and we should have consistent and consistently great shots. You've proven you can dose higher, change your preinfusion rate to compensate, and achieve a very similar shot as what you had before.

Interestingly, my preinfusion is currently "fixed" in that I hit the puck with line pressure pretty quickly, but I control how long it lasts and I can actually hear when the headspace fills. It will be very interesting to see how things change when I'm actually dynamically decreasing the water debit and allowing the gradual increase to 9 bar (or 5, or anywhere in between ). I still have a step change in puck pressure right now, but the saturated puck doesn't seem to care all that much. It's darn near impossible to choke the machine. Even if it takes 45 seconds to wet the puck, the shot progresses fairly ok... Exciting times lay ahead for me

Quick parting thought experiment:

If we dose 18g and we believe the puck absorbs 36g during complete preinfusion, when we raise the pressure and pull a 36g shot, we are effectively pushing our entire shot out of the saturated puck. That is, none (or very little) of the pressurized water actually makes it into the cup. The water going into the puck during preinfusion is effectively scrubbed from the puck and pushed into the cup.

This is of course an over-simplification, as a good portion of the water is absorbed by the puck on the molecular level, and this absorbed water does little if anything to influence extraction, and is not likely to be exchanged with fresh water when the pump is turned on, but even so, I think it's safe to say that 50-75% of what ends up in our cup is put into the puck during complete preinfusion...

Cheers!

- Jake

Do you expect to get more - or - less crema with full PI?

I seem to get more, but I feel like I have no business expecting it...
My pulls seem to be close to 100% crema after the first drops. I suppose if the coffee is blooming while becoming saturated, that would release plenty of CO2 for some heavy crema...

Other side note:
I found a legitimate use for nutating while trying to down-dose for a few days. My baskets are tapered shortly after the ridge, and dosing less than 17g results in a puck that lives entirely in the taper when tamped. As such, my tamper won't fit deep enough in the basket to tamp level. Solution? Nutate. The rolling motion of the tamper reaches further into the basket and does a nice job sealing the puck the basket wall. Even though there was a thinner puck on the edges than the center, and the puck height was well below the beginning of the taper, flow development was consistently as good as my ability to nutate evenly, and the pucks grew enough to stick to the shower screen when I pulled the portafilter after the shot. Nothing a short flush doesn't immediately take care of, but interesting all the same.

In the end I don't think down-dosing is a particularly useful tactic when pre-infusing, or at least not with my grinder. I'm already grinding close enough to the zero point that the grinder decides every few months to choke and just twirl the beans around in the grinding chamber instead of actually putting grinds in the doser. This may be a consequence of single-dosing, but I'm not so sure, as the burrs touch just three ticks away from my normal espresso grind range...

Truth be told, I followed your lead this morning and up-dosed a gram and head a beautiful shot. After 16s of preinfusion ,I had 34g in the cup in 26 seconds with a nice, steady flow. No change in grind from my normal doses.

Get a narrower tamper. I have one for single dosing (43 or 49mm). Or stick with doubles. I don't really use my single dose tamper. Last used it a few years ago.

Speaking of grinders: nothing in my espresso gear remains stock. So in the process of getting the grinder to do what I want - I dropped the speed to 370 or so RPM. The reason is that distribution is easier. The higher the RPM the more centrifugal force you have and the puck becomes donut like. It is harder to then redistribute.

But the down side is what I assume to be some stratification. WDT helped. Since WDT only helps a sick grinder, I tripled the number of normalization vanes (called wipers in the VL world). Cured the grinder.

So slower burrs (conical + flat), & triple vanes & Tap & wedge make a perfect puck.

I think the conical burrs of the VL really help when you are down in the finer grinds. It takes longer as I assume the beans bounce around, get shattered by the conical, and are then small enough to fit in the flats.

Feedback Wanted!

Here is a sneak peak at my proposed design for the paddle action of my profiling mod. The idea is pretty simple. The paddle controls the water debit of the group by directly turning a miniature needle valve installed in place of my gicleur. A normally closed microswitch (not shown) is opened when the paddle is placed in the rightmost position, which turns the pump and brew solenoid off. Once the paddle is lifted and the microswitch is closed, the pump and brew solenoid are engaged. The needle valve will be at its smallest operating opening at this point. My intent is to have the water debit be set in the neighborhood of 30mL per minute or less at this position, which should be low enough to effectively limit the peak brew brew pressure to 4 bar or so. It's adjustable, so I'll set it where it needs to be to achieve as low of a pressure at I desire... Swinging the paddle further to the left progressively opens the needle valve. The intent is that fully to the left would mimic the 0.5mm gicleur I currently have in place, which gives me a water debit of 750mL per minute with the pump engaged. If this leaves me with too narrow of a profiling range on the right side to be useful, I have a second needle valve chosen with a tighter taper, but that valve limits my peak water debit to around 250mL per minute, which may very well be preferable, but I think I'll start with the ability to match what I've got now. The prominent central "arch" is something I'm thinking about 3D printing to house the puck pressure gauge and give the paddle a shrouded look to it, providing a location to put flow graduation tick marks. The flow profile of the selected needle valve(s) is largely linear in the operating range, so the ticks should provide a repeatable setpoint for manual pressure profiling setpoints. This is effectively the same principle behind the Dalla Corte Mina, but 100% manual with visual feedback via the puck pressure gauge.

What do you think?

Cheers!

- Jake

Have you considered having 2 microswitches? One to open the 3 way and the other to turn on the pump? That way you can extend the range of the paddle. The pump switch should happen after the headspace is full so it won't cause a step response.

Another potential problem I foresee is needle valve hysteresis. I really don't understand it as needle valves are based on a stainless needle in a stainless orifice so in theory it shouldn't happen. But the fine needles tend to have a different outputs going clockwise than counterclockwise. I just accept it as I set it once and leave it. Maybe it is some tiny play in the threads?

Also - is the needle valve between the 3 way and the puck? Or is it in the input to the 3 way. The latter is okay. The former would mean coffee grinds would be pulled back when backflushing and whenever the 3 way opens. A needle valve there is likely to clog.

On the subject of PI (which doesn't seem to get less complex as the discussion progresses) - I am observing more and noticed something interesting. Pucks, as you stated, drink about 2x water. Pretty consistent. So an 18gr dose would drink 36 grams of water.

The odd thing is when do the first drops appear at the bottom of the puck (which we identify as end of PI - and which mean the puck would no longer crush and seal). It can happen from about 1x to almost 2x water. At a very low water debit and fine grind it can almost take 2x. At higher water debit and somewhat coarser grind -1x.

I suspect it may influence EY.

Edit: why that last sentence? Because of in cup results. The brew ratios can stay constant. Puck water content stays constant. But percolation time extends from 1x/debit to 2x/debit. Soak the puck and push the water out vs. wash the puck.