Pressure profiles, preinfusion and the forgiveness factor

[HB]

In the Elektra A3 conclusion I wrote:

Plainly stated, dialing in the temperature and extraction was too darn easy. In retrospect, I realize that I carried an unstated assumption into the evaluation: Fast pressurization means poor preinfusion (wrong!). This bias originates in my experience with vibration pump E61-type machines, which pressurize much slower than a rotary pump's blink-of-an-eye 2-3 seconds.

At the time I didn't elaborate on my assumption that "fast pressurization means poor preinfusion", but thanks to Sean Lennon, I'm able to present some interesting data for your consideration. First a little background...

Yesterday morning the doorbell rang. When I opened the door, there was a large black pelican case with triple seals sitting on my porch, looking decidely serious. It was nice and sunny. I thought, "Hmm-m, what is this?" Sean had e-mailed me the day before telling me to expect a package. He didn't tell me what to expect. Whoa, an espresso machine Mr. Wizard tester kit! I'll leave the specific hardware description to Sean, the important thing is that with this gear, you can create combined pressure and temperature profiles with data points captured easily at 20 per second. That adds another dimension to an analysis, plus the capture rate is a lot better than my own setup plugging along at one data point per second.

Lino stopped by this afternoon to help me answer one question that puzzled me for months: The Elektra A3 appears to ramp up in pressure in seconds, so why isn't it "unforgiving"? Was there something different about its pressure increase? Was there an interval that should be rightly called preinfusion, however brief it might be? We took a stock thermofilter and tee'd in a transducer (translates pressure into an electrical signal). Now we could record the ramp up in pressure. Of course, it's not the same as a real coffee puck since dry coffee would act as a cushion, but it does give us a means of roughly comparing the machines' initial pressure characteristics.

Guess which one is the A3:


Comparisons of the pressure profile of three espresso machines (temperature omitted for clarity)

Keep in mind that these measurements are taken with a portafilter that's already full, so the ramp up is more rapid than normal (each vertical gridline is one second, each horizontal gridline is 10 PSI). Nonetheless, the A3's profile in the middle stands out as starkly different from the other two - the pressure rockets up in barely one second, which is consistent with the rapid onset of its extraction in actual use.

What I learned from this experiment is that pressure ramp up may contribute to the "forgiveness factor" we attribute to the E61, but clearly it isn't the whole story. Care to guess what the other two machines were? Or offer what you think might be other key contributors to the forgiveness factor?

[AndyS]

What I learned from this experiment is that pressure ramp up may contribute to the "forgiveness factor" we attribute to the E61, but clearly it isn't the whole story. Care to guess what the other two machines were? Or offer what you think might be other key contributors to the forgiveness factor?

Physical factors in the grouphead (headspace above and below the dispersion screen, water spray pattern, etc).

[skyryders90]

Nonetheless, the A3's profile in the middle stands out as starkly different from the other two - the pressure rockets up in barely one second, which is consistent with the rapid onset of its extraction in actual use.

Were the other 2 machines vibe pumps or rotary? I'm wondering if the differences you're talking about are pump-specific, or something particular to the A3 and the other machines.

Thoughts?

[HB]

They are both vibration pump machines.

[skyryders90]

They are both vibration pump machines.

That is what I thought. I know that with my Bricoletta (rotary pump) I see much faster initial beads than with vibe pump machines - 2 to 3 seconds instead of 5 to 6. Certainly, other factors play a role, but I think the pump is the largest one.

[1st-line]

What I learned from this experiment is that pressure ramp up may contribute to the "forgiveness factor" we attribute to the E61, but clearly it isn't the whole story. Care to guess what the other two machines were? Or offer what you think might be other key contributors to the forgiveness factor?

My guess is that first chart is the Brewtus and the third is the Valentina.

[Bob Barraza]

The results for the A3 seem counter-intuitive to me. I would have assumed that the sudden ramp up in pressure would create the dreaded 'water hammer' effect.

When I tested the A3 it was obvious how forgiving it was. My dosing and distribution were a bit less than consistent, yet the results were very good. Similar technique on my Livia 90 with a bottomless portafilter would have warranted the operator wearing safety goggles and a rubber apron at a minimum!

Is the pre-infusion more a factor of the flow rate of the pumps? As I recall the vibe pump starts at maximum flow rate which decreases as the back pressure builds.

Bob

[HB]

My guess is that first chart is the Brewtus and the third is the Valentina.

Ding! Ding! Ding! We have a winner! Apparently Jim is paying attention to my current inventory of espresso machines. The giveaway is the third chart showing La Valentina. It doesn't have an expansion chamber and you can see that the pressure increase is slightly steeper as a consequence. By the way, below is a snapshot of the setup for those who may be wondering:


Thermofilter with pressure transducer, Measurement Specialties Inc., model MSP 600

The portafilter has a Scace Thermofilter Temperature Device and the transducer I mentioned earlier. Temperature and pressure are recorded on a laptop. As neat as this rig may be, it's lightweight compared to Sean's own setup. Check out the Brewtus Epilogue page to see seven different variables plotted on the same chart. Needless to say, I asked him to simplify it for my use.

Is the pre-infusion more a factor of the flow rate of the pumps? As I recall the vibe pump starts at maximum flow rate which decreases as the back pressure builds.

Your question really goes to why I started this thread: Understanding the forgiveness factor is more complicated than I thought. The pressure ramp up certainly plays a role, as does the grouphead design. In the past I thought the importance of the former outweighed the latter. Now my thinking is that it is in fact reversed, that is, the pressure profile plays second fiddle to factors like Andy pointed out: headspace above and below the dispersion screen, water spray pattern, etc.

As a practical matter, vibration pumps are "slow" to move water compared to a rotary pump, even at no resistance and even slower once there is resistance. Rotary pumps have much much higher flow rates independent of the resistance that you see almost instantaneous pressurization.

[lino]

Hey Dan,

Just FYI. I've got the necessary quick release parts on the way. If you bring it on Friday, we can "spice it up".


Sean,
What's the temp compensation range of that transducer?

ciao

lino

[lennoncs]

Hi Lino,

Operating temperature range -40 to 100°C (125°C available,
consult factory)
Compensated temperature range -20 to 85°C (125°C available,
consult factory
Thermal error <±1% of FS (75-10,000 PSI)
Thermal error <±1.5% of FS (25-50 PSI)


This unit was built to 125C specs....I have significantly better transducers at home (-40 to 200°C ,<±.15% of FS (10-200 PSI)) but somebody is gonna give up a first born for a > $1,000 unit to test their machine with.

It is not a high dollar unit but I have used this sensor on parallel systems with both mechanical and other transducers and it has never flinched at 110C. If you are seeing issues with the readings, it may have been dropped or spiked in a prior life.


Cheers
sean

[swines]

I understand the fun of being able to quantify an item's performance. However, I'd like to know exactly how you think this information can and/or could be applied to making better espresso?

Can you relate the pressure ramp-up characteristics to better taste? If so, what or how does that mechanism work? Measurements for measurements sake are often interesting, but you can turn it into pseudo-science which then resembles the following for usefulness.


http://www.improb.com/airchives/p...9/v9i3/kansas.html

[DavidMLewis]

Keep in mind that these measurements are taken with a portafilter that's already full

This could bias the results, since a real puck has an unknown amount of air-filled head space. In the real puck's case, it could be seeing a slower pressure rise because of this, assuming you mean "already full" of water in the quote above.

Best,
David

[HB]

However, I'd like to know exactly how you think this information can and/or could be applied to making better espresso?

I am not asserting that a particular pressure profile is superior, only that they are different. Previously I had assumed that a slow pressure rise would be more "forgiving," but I'm no longer convinced such a blanket statement is justified. However, the pressure profile does appear to influence the best techniques for a given machine. For example, the uber fast pressure rise of the Elektra A3 will punish overdosing by increased channeling; this isn't true for other machines like the La Marzocco. My theory is that the grouphead design is behind these differences, not the pump pressurization speed.

The case for temperature measurements has more practical application. I principally use this information to learn the machine's performance characteristics more quickly. How much to flush if the machine has been idle for 5, 10, and 30 minutes? Generally the first shot or two will run sour if the group isn't heated sufficiently. So how many flushes are necessary to warm it up? How much does the group temperature change, if any, when steaming and pulling shots at the same time?

That's a sample of the questions this sort of data helps answer. Of course you have to confirm what the graphs say with the actual taste. I agree with you, these measurements are not good predictors of the quality of the espresso. They are helpful leading indicators of consistent performance, but I treat them as suspect unless there is weeks of actual use to support the assertion that a machine has a "good" temperature profile.

(Not that I'm convinced there is such an animal. Slant L or humped temperature profile? I'm unpersuaded either is superior for all machines and all blends; holistic comparison is really what matters, the rest is engineering data to support what your tastebuds already concluded.)

This could bias the results, since a real puck has an unknown amount of air-filled head space. In the real puck's case, it could be seeing a slower pressure rise because of this, assuming you mean "already full" of water in the quote above.

I absolutely agree. My original motivation for this experiment was to determine if the Elektra A3 had a preinfusion cycle comparable to the E61 expansion chamber or other means for slowing the pressure rise as some offline discussions had suggested. Based on watching how quickly the initial beads formed (around 3 seconds), I didn't think so, but couldn't resist a high-tech way of confirming. Apparently the Elektra designers didn't think a slow pressure build-up was crucial, and judging from the quality of the extractions, I believe they're right.

From the thread Rotary pump cup clarity achievable with a mod ?

Dan, what is it about the A3 that made its "preinfusion" different from the others?

and this thread:

Or offer what you think might be other key contributors to the forgiveness factor?

Physical factors in the grouphead (headspace above and below the dispersion screen, water spray pattern, etc).

I recall that you added a preinfusion cycle to your heavily-modified Silvia. Based on these results, I wonder if modifying the grouphead itself to increase the space above the dispersion screen would improve Silvia's forgiveness factor? Interested in machining a few channels in the grouphead in the name of experimentation?

[AndyS]

Physical factors in the grouphead (headspace above and below the dispersion screen, water spray pattern, etc).

I recall that you added a preinfusion cycle to your heavily-modified Silvia. Based on these results, I wonder if modifying the grouphead itself to increase the space above the dispersion screen would improve Silvia's forgiveness factor? Interested in machining a few channels in the grouphead in the name of experimentation?

No problem sir! I'll turn her upside, fill the group with conc. nitric, and let her sit for a day.

In the name of experimentation, sir!

[HB]

Funny guy. OK, maybe that is a lot to ask. How about this: Would you explain how you modified Silvia to add preinfusion, and whether it improved her forgiveness factor? Both Mike and I were reminded of Silvia's fussiness during the PID'd Silvia versus Brewtus comparison and attributed Brewtus' better showing to the E61 expansion chamber. Now I'm thinking that's an overstated value.

[AndyS]

Would you explain how you modified Silvia to add preinfusion, and whether it improved her forgiveness factor? Both Mike and I were reminded of Silvia's fussiness during the PID'd Silvia versus Brewtus comparison and attributed Brewtus' better showing to the E61 expansion chamber. Now I'm thinking that's an overstated value.

Dan, as you know, the word "preinfusion" is tossed around a lot, but everyone uses it differently.

Years ago I put in a time delay relay that allowed the 3-way to open for a few seconds before the vibe pump came on. The coffee would get "preinfused" by the boiler vapor pressure. Damned if I know if that made her more forgiving. I didn't notice any difference at the time, so I took the relay out.

Sometime later I installed a rotary pump, and then a ways after that, a needle valve to regulate the time it took for the extraction to come up to full pressure. With the valve wide open, it would only take a second or two. With the valve tightened, you can control it so that that it takes 3 seconds, 5 seconds, or whatever. That way, I could get a "gradual infusion."

Some call this "preinfusion," others don't.

I think that the extractions are more forgiving when the rampup takes 4-6 seconds as opposed to 1-2 seconds. I think the 1-2 second extractions go blonde sooner. But I would caution that this applies to my machine, with its particular group head design. Other group head designs may perform differently.

And as I told you on the phone today, the REAL interesting part of this is that there's a significant difference in the grinder setting required to maintain 25-30 sec shot timing between the fast and the slow rampups. And the fast rampup requires a <gasp> COARSER setting, by about one notch on the Mazzer Mini collar. Presumably this is because the fast rampups slams the crap out of the coffee cake, compressing it like a 500lb gorilla tamp.

Again, this phenomenon occurs on my particular machine. Barry Jarrett could not duplicate this effect on his LM, but Rene confirmed it on a Reneka Techno.

[HB]

As a practical matter, vibration pumps are "slow" to move water compared to a rotary pump, even at no resistance and even slower once there is resistance. Rotary pumps have much much higher flow rates independent of the resistance that you see almost instantaneous pressurization.

I've never had two essentially identical machines except for the pump type to test - until now. Chris Nachtrieb agreed to loan us a Quickmill Anita for EspressoFest (the group, boiler and pump are the same between the Anita and Andreja). Putting the Anita and Vetrano under the close scrutiny of Sean Lennon's pressure profiling kit (thank you Sean!) produces the charts below:


Two E61 espresso machines, two different types of pumps

Anyone care to guess which one is the rotary pump E61 (Vetrano) and which is the vibratory pump E61 (Anita)?

[lennoncs]

Right=Anita

Sean

[HB]

Right=Anita

Sorry Sean, left = Anita, right = Vetrano. I was surprised too, but looking at the earlier profile of the Brewtus' vibratory pump, it's almost exactly the same curve. My guess is that the rotary's faster flow rate is producing the very distinct expansion chamber "knee."

[lennoncs]

Doohh!

Interesting, I would have thought the lower volume of the vibe would have shown a knee.

should have looked back at my own curves too...they show the same thing, now that I am actually looking at the slope.


Neat


Sean