Chapter 12: Profile what?

It's early November, 2017. I've been using a technique picked up from a post by HB member RegulatorJohnson for the last week where I run my right group with no restriction and open the solenoid to the left group to preinfuse at a reduced pressure.
(Until recent revelations, I was certain that the pressure applied to the puck was über low with this method. I now suffer from the painful realization that there is enough pressure using this method to lift the relief valve, so... thanks for ruining my fairy tail, AssafL )
While this method does not provide as low of a pressure as a simple LM style solenoid infusion at mains pressure, it requires zero modifications, is easy to do and visibly changes the way the shot progresses. Can I taste a difference? Maybe, but I can definitely taste the difference between a good shot and an underextracted one. This method has helped me big time with consistency, although it is entirely arguable that better technique is all that's really needed for that. Better technique and a gicluer. What a difference that made! For those of you that like the voltage/pressure and current/flow analogy, there are two models that work well:

Current divider: A constant voltage(pressure) will divide it's current(flow) between parallel resistors(gliceur, puck, relief valve, flowmeter, etc...) with the current(flow) favoring the lesser of the two resistances. In this model, the right group functions as a relatively low value resistor, placed in parallel with the puck and the bypass valve. In an electrical system, more current will flow through the lower resistance. As such, so will more of the available water flow through the path of least restriction, or the right group...

Voltage divider: A constant current(flow) will split its voltage (pressure)drop between resistors(gliceur, puck, relief valve, flowmeter, etc...) placed in series such that all the current (flow) must pass through each resistor with the larger voltage (pressure) drop occurring across the resistor with the higher value. Let's say there's 9 bar at the inlet of a gicleur that provides the exact same resistance to flow as the puck while extracting. In this example, each "resistor" would split the pressure drop evenly, with a 4.5 bar drop across the gicleur and a 4.5 bar drop across the puck...

Moving right along... I'm deciding to do some tasteful mods to my machine, so let's talk about what some of the other folks out there are doing and then get to what I plan on doing, which is a simple combination of two of the more common methods out there right now. I'm a fan of simple mechanical systems. My S20 will not be connected to an Apple watch, I won't have sub-second telemetry beamed to the cloud. It's simple. But I'm pretty sure it will be effective. So. What's going on out there now?

Pressure profiling:
This is sort of a misnomer, but whatever. There is maybe a pressure transducer in a "pressure profiling" machine, maybe not. One thing pretty much every pressure profiling machine has is a variable speed pump. In the case of a closed-loop system, the user selects a target pressure profile and the system varies the speed of the pump to hit the target pressure at multiple points throughout the extraction. The end result is that the pressure follows a profile. Other machines use a paddle style control, where all the way to the left is "wide open" and and maybe pointing straight at you is line pressure. Other folks just have a knob that turns the pump faster or slower. And some have an ap on a tablet... You get the idea. The reason I find this naming convention amusing is that espresso machine pumps are positive displacement devices. They don't make pressure, they move a fixed volume when they are turned at a certain speed. When you control the speed of a positive displacement pump, you control the flow directly. The pressure varies with the flow in response to the system... Not a big deal, but I think it's interesting.

Flow profiling:
Think Slayer. Precision needle valve. "That's not true pressure profiling, it's flow profiling 'cause they're using a needle valve." Boo hoo. When you look at the downstream side of a needle valve, they are a pretty effective device for controlling presure. Remember that the pressure drop across any orifice is proportional to the square of the flow through it. The two are joined at the hip. Truly, a needle valve induces a pressure drop across it based on the flow passing through. Once the needle valve is closed tight enough to cause a pressure drop large enough that the bypass valve opens, the flow then has a choice of whether to travel through the bypass, or through the needle valve. Even so, smaller valve openings result in lower pressures at the puck by virtue of the pressure dividing principles of constant flow through consecutive restrictions in flow. The combined restriction of the puck and the needle valve determine the total system pressure and if the bypass valve is opened, they also control how the flow is divided between the bypass and the cup. So, controlling pressure drop across a variable orifice in turn controls the flow out of the group...

Line pressure preinfusion:
From the simple "middle position" with a plumbed E61 to the 6 o-clock position on the paddle of an LM linea or Synesso, using line pressure to preinfuse is a technique that's been around for ages. Pressure profiling? Not quite, but it does offer a level of preinfusion that is a bit more complete than what the Faema and KvdW designs can provide. One issue here is that rotary pumps generally require that any pressure regulator placed in front of them be set to a higher pressure than would be optimum for pre-infusing in order to prevent cavitating the pump. The other issue is that the flow is not metered well. The primary restriction to the flow is the puck, so variations in coffee preparation yields inconsistencies in the preinfusion flow rate.

Ok, so we've got three different methods that seem to be readily available on the market that result in either staged preset pressures applied to the puck at different times, or super slow preinfusion, or full control of the pump output, or some combination thereof. It was through reading Slayer's patent and user's manual that I got a good sense of what I aim to do with my S20. Remember that I've invested no more than $350 in this machine so far, and my intent is to keep my modifications simple, effective, reversible and cheap. That said, Slayer presents a simple goal in their patent that I aim to accomplish. I'm paraphrasing here, but the basic intent is to fully wet the coffee grounds at a low enough pressure and flow that the fines are not migrated through the puck, causing damming of the flow. By accomplishing this, the hypothesis is that the fines will be suspended in the saturated puck as the coffee blooms. Suspended fines will not provided the same resistance to flow as they typically do when they migrate and dam the flow. The grinder is set significantly finer such that the macro particles themselves provide the resistance to flow that the dam caused by migrated fines typically does. These smaller macro particles have greater surface area and consequently allow a more complete extraction than a typical grind in a typical machine can accomplish.

Slayer accomplishes this with a dedicated flow circuit containing a precision needle valve that is set to an extremely low flow rate of 40-60 grams in 30 seconds with the pump running... See details of this explained rather well by Shadowfax here. My modified preinfusion method delivers 247g without the puck in place. I suspect the increased resistance of the puck drops this down significantly with the right group open, as it takes 10-15 seconds to saturate the puck with my current method, but even so, it is nowhere near as gentle a flow as what Slayer delivers.

So, what to do, what to do? This post is long enough, so I'll finish my plans in the next chapter...

Stay tuned!

Jake_G wrote:
Current divider: A constant voltage(pressure) will divide it's current(flow) between parallel resistors(gliceur, puck, relief valve, flowmeter, etc...) with the current(flow) favoring the lesser of the two resistances. In this model, the right group functions as a relatively low value resistor, placed in parallel with the puck and the bypass valve. In an electrical system, more current will flow through the lower resistance. As such, so will more of the available water flow through the path of least restriction, or the right group...

Voltage divider: A constant current(flow) will split its voltage (pressure)drop between resistors(gliceur, puck, relief valve, flowmeter, etc...) placed in series such that all the current (flow) must pass through each resistor with the larger voltage (pressure) drop occurring across the resistor with the higher value. Let's say there's 9 bar at the inlet of a gicleur that provides the exact same resistance to flow as the puck while extracting. In this example, each "resistor" would split the pressure drop evenly, with a 4.5 bar drop across the gicleur and a 4.5 bar drop across the puck...

Minor corrections here...

A constant pressure (voltage) will deliver the same flow in each leg as if the other leg did not exist. So assuming you have a very large motor and a pump that can deliver a constant pressure at the output port (irrespective of bypass and irrespective of flow - i.e. unlimited compliance flow) - and there is indeed 9 bar at the output, the flow through the left and right groups will not influence each other. Just like your 110v powers your bathroom heater, TV and espresso machine all at once....

A constant flow (current) however, will split the flow between the two alternative (in parallel) paths. Just like a river island splits the flow of a river into two streams. The flow rates of the two streams add up to the flow rate of the river.

If we look at a recirculating rotary pump, it does it's best to keep the output pressure at 9 bar. So in theory, opening one group or two shouldn't matter. However, the Italians seem pretty frugal with their components. So more than likely, the maximum compliance of the pump (i.e. what is the maximum flow it can sustain at 9bar output pressure - obviously the recirculating valve would fully shut at this point...) is limited, and the flow may be reduced a bit. Hence the importance of gicleurs for multi group machines. Likewise, while a gear pump is volumetric it also has compliance as well - at maximum pressures it will start to "slip"; either the magnets will relent or water will force their way back through the gears - possibly damaging the gear surfaces.

Pressure profiling doesn't need a variable speed (actually volumetric) pumps. The E61 type system (and its clones) as well as levers pressure profile the flow. They profile the pressure after the headroom fills up and pressure starts building. Slayer, by virtue of keeping the gear pump speed constant profile the pressure (it decays as the pull progresses - albeit the pump is at constant speed). The funny thing about the slayer is that they use the volumetric pump in the least "controlled" manner possible. Just like the Ron Popeil delightfully awful Rotisserie ("set it and forget it"). Except that the pressure does decay - and by over 10%...

The circulating pump however, is of sufficent compliance so that the pressure would not decay. They temper that down with the gicleur so at least headroom fills slower than the pump could deliver. As long as there is headroom it is flow limited ("profiled") - not pressure; the pressure being 0 on the puck. A slayer and any gicleur machine profile the flow; once pressure builds up the valves and gicleurs no longer affect the flow nor pressure. By coincidence - it is what Slayer does with the needle valve - as if it were a tiny gicleur; except that the valve is adjustable - and perhaps more important - if it ever clogs just play with it until the deposits crumble...

AssafL wrote:Minor corrections here...
... The circulating pump however, is of sufficent compliance so that the pressure would not decay. They temper that down with the gicleur so at least headroom fills slower than the pump could deliver. As long as there is headroom it is flow limited ("profiled") - not pressure; the pressure being 0 on the puck. A slayer and any gicleur machine profile the flow; once pressure builds up the valves and gicleurs no longer affect the flow nor pressure. By coincidence - it is what Slayer does with the needle valve - as if it were a tiny gicleur; except that the valve is adjustable - and perhaps more important - if it ever clogs just play with it until the deposits crumble...

I agree with everything else you brought up. emphasis added above

What you describe above is as if saying that a resistor only effects the inrush current when a motor is started but once it's going at steady state the resistor has no voltage drop across it and ceases to limit the current. While it's true that when filling the headspace the effect is greater due to the time factor of how long it takes to purge the air from the system, remember that all flow requires a pressure gradient to occur. It's just that this pressure drop is not linear with flow as voltage is with current. It is proportional to the square of the flow, so if you cut the flow in half, you lose 75% of the pressure drop. So valves and gicleurs do have less of an effect on pressure and flow once the pressure ramps up.

All the same, your point is taken, as the pressure in a slayer does ramp up very nearly to full pressure once the extraction begins, even in pre-brew through the needle valve. But remember that switching back to the pre-brew at the tail end of the shot drops the pressure in a Slayer. This is because the flow through the puck becomes less restricted as the shot progresses, so the flow would naturally increase in a normal machine as it does through the 0.7mm orifice brew path in a slayer with constant-ish pressure at the puck. When switching back to the needle valve, this increased flow causes a larger pressure drop across the valve than the restricted flow at the beginning of the extraction did. As a consequence, the pressure gauge at the puck shows a trailing pressure profile when switching from brew to pre-brew at the latter half of the shot. The flow falls off, too, as the flow and pressure drop through a system must lie on the system curve just as the flow and pressure out of a pump must fall on the pump curve. Once they reach a new state of equilibrium, you see the tail end of a slayer shot arrives at a much lower final pressure at the puck when in pre-brew than you do just going through the gicleur in brew mode.

This is not a function of headspace causing a flow-profile, but rather a function of any flow through an orifice that creates a corresponding pressure drop across it as part of any hydraulic system. I do totally agree, however that when purging squishy air out of a hydraulic system these effects are magnified to the point that one could argue that steady-state flow renders them nearly negligible. But extractions are not steady state. They start slow, and ramp up in rate. To your point, a big enough pump does do it's best to maintain constant pressure. Most pressure gauges on machines are reading pressure upstream of the gicleur in the group, so they only pick up on the variations in pressure caused by the "spurious" gliceurs located in the rest of the system... The rest of this fun stuff happens quietly behind the scenes. But the puck sees it all...

My last jab at your reply is that while you are absolutely right that E61 machines and the like provide a profile to the pressure and flow, I don't think a manufacturer would ever get away with marketing that design nor a lever as a "pressure profiling" machine... but as always, you're spot on with the details.

Cheers!

- Jake

While I agree that marketers will find it difficult to market a PI chamber as a "pressure profiler", from my technical POV it is exactly that. Perhaps it isn't a customizable PP machine (like a Strada EP or MP would be) - albeit, in theory one could replace the spring with one of a different k - but it does profile the pressure buildup.

Perhaps a more appropriate way to look at PP and FP is as mnemonics to better understand and memorize the rule of thirds; in any case it is far more complex than a simple "ignore the gicleur" or "don't ignore the gicleur" over-simplication.

But that said, these approximations seem to work out well in the field. Here is a Joe's shot I pulled this AM. Nicely balanced, somewhat older coffee (~60 days) so lower, lighter crema.


You can see the pressure taper from ~9.1 bar to about 7.8 bar due to the puck becoming more permeable to the liquid - the pressure is profiled by the erosion of the puck. Flow, conversely, increases in the same period. Since the needle valve opens at 15 seconds (after the pressure ramps up at the needle flow rate) there is no jump in the pressure. One might be tempted to not open the needle at all....

Regarding the E61 PI, I totally agree with your assessment of whats happening. How about "user-controlled dynamic pressure profiling"? Once the needle valve at the top of the original group design was replaced with a gicleur, the pressure ramp profile is now out of the Barista's hands, but clearly still a fundamental function of the intended design.

Thanks for sharing that profile. Do you think the pressure is dropping ahead of the gicleur and this is slip on the gear pump or simply over-running the pump's capacity is the drop in pressure only visible between the gicleur and the puck. My understanding is that Slayer is reading pressure at the puck...

I thought the same thing regarding "why not just leave the the needle in the loop?". Please post a shot profile doing just that if you don't mind. I suspect the erosion will be far more pronounced with the needle in place, but I'm super happy to be dead wrong!

Part of my silly assumption is that if slayer could have left the needle valve in place and totally simplified their design, they would have?

Cheers!

- Jake

This isn't a Slayer - it masquerades as one so there may indeed be some difference (but I don't believe in magical thinking so I wouldn't believe there is a big difference).

The pressure reading is at the boiler - so after the flowmeter gicleur and before the group gicleur.

I will do a test as you suggest later.

First of all, I just wanted to chime in and say I really appreciate all of the feedback on your journey, as it is very insightful.

Not to hijack, but AssafL, where is this screenshot from?

AssafL wrote:While I agree that marketers will find it difficult to market a PI chamber as a "pressure profiler", from my technical POV it is exactly that. Perhaps it isn't a customizable PP machine (like a Strada EP or MP would be) - albeit, in theory one could replace the spring with one of a different k - but it does profile the pressure buildup.

Perhaps a more appropriate way to look at PP and FP is as mnemonics to better understand and memorize the rule of thirds; in any case it is far more complex than a simple "ignore the gicleur" or "don't ignore the gicleur" over-simplication.

But that said, these approximations seem to work out well in the field. Here is a Joe's shot I pulled this AM. Nicely balanced, somewhat older coffee (~60 days) so lower, lighter crema.
<image>

You can see the pressure taper from ~9.1 bar to about 7.8 bar due to the puck becoming more permeable to the liquid - the pressure is profiled by the erosion of the puck. Flow, conversely, increases in the same period. Since the needle valve opens at 15 seconds (after the pressure ramps up at the needle flow rate) there is no jump in the pressure. One might be tempted to not open the needle at all....

Not hijacking at all!

I had the same question...

uid0 wrote:Not to hijack, but AssafL, where is this screenshot from?

For a complete description, click here.

For a brief description and some screenshots of my somewhat simpler version, click here. The big differences are 1) no needle valve (I do Slayer shots by reducing the input water pressure regulator to 1.5 BAR and turning off the motor), 2) no brew ratio calculation (I do have auto cutoff when the target weight is reached), 3) no ability to store/replay manual profiles (I've not found any advantage to pressure profiling over Slayer shots), and 4) no built-in touch-sensitive graphics display (I use an Android device as a front-end to get those features.)

Thanks Dick!

Those reads made me happy The stuff you GS3 modders have pulled off is nothing short of astonishing...

I'm still finishing up my reply to Assaf. No matter how I read through it and restructure and paraphrase, I feel like it seems I'm arguing with him when in fact I'm thoroughly enjoying our discrouse. Thanks for keeping my ramblings interesting (to me at least ).

-Jake