Measurement of espresso machine performance - Page 3

#21: Post by **MOSFET** » November 16th, 2005, 7:50 pm

terryz wrote:Actually Keith we have a huge underground espresso lab here in the great Northwest. Most of us are so damn busy tasting the coffee we forget to share the info

The great part about this machine and the company developing it is that they have asked for opinions. It is a rare opportunity to get in on that kind of action, so heck yeah, we are all working together to insure that when this baby goes to market, it is the best it can be for the job it is being asked to perform.

The Scace device is a great tool, that I use everyday. It allows me to bench machines with a consistency that was never before available. Use
of the device on the GS3 was simply for verification purposes. I'm kind of old school with regards to all of the technical aspects of espresso machine testing and tend to spend more time on functionality and taste of the coffee. All the statistics in the world will not change the results in the cup.

The "Holy Grail" of temperature stability has been found in the last two years, so now where do we go? Sean L. seems to think it is in the pump / pressure, and maybe he is right, all I know for sure is that allot has changed in the last two years and it is all good stuff. Statistics aside, I find shear joy and happiness in the simple fact, that someone is paying attention and doing something to raise the bar commercially.

Well send some LOVE out here to the northEAST. I'm cold and lonely. If the Scace device saturates I can guarantee you more precision with my LM34 silicon temperature sensor and seven-digit 1978 model Fluke DMM.

Anyway, hope you all are having fun. Wish I could visit. I'll go back to the bat cave and continue modding my Linea.

Keith

#22: Post by **lennoncs** » November 16th, 2005, 8:30 pm

MOSFET wrote:Ok. Temperature, pressure, and, if you insist, time. But I took time as a given:

Pre-infuse pressure? PRESSURE
Pre-Infuse Time? TIME
Ramp to extraction pressure? TIME/PRESSURE
Time at pressure? TIME/PRESSURE
Pressure decay or increase during shot? TIME/PRESSURE
Flow rate of the shot? TIME/PRESSURE
Pressure/ Time relationships? SIGH...
Grind? TIME/PRESSURE
Distribution? TIME/PRESSURE
Roast? BEEF? MARINATED? SOUNDS GOOD.
Etc. TIME/PRESSURE/TEMPERATURE

sorry, I'm in a silly mood. Blame it on black label.

Keith

toss out what you don't like

it is all time, temp and pressure, it is just the amount of each that is so difficult.

November 16th, 2005, 9:05 pm

AndyS wrote:
It's not fair, it's not fair.

Use a straight line profile and an actuator made of physical matter, rather than the divine stuff -- a lot cheaper, same effect, since last time I looked, infinite rates fo change were still illegal in this universe.

#24: Post by **lennoncs** » November 16th, 2005, 9:19 pm

terryz wrote:The "Holy Grail" of temperature stability has been found in the last two years, so now where do we go? Sean L. seems to think it is in the pump / pressure, and maybe he is right, all I know for sure is that allot has changed in the last two years and it is all good stuff.

Actually, I wish I knew.

I believe it is an insanely complex process that we have only rudimentary understanding of and even less data to support our theories.

My contribution to better understanding this process is to help develop some tools such as a servo pump that can be programmed to deliver a specific pressure profile repeatedly or help support the collection of high quality data by making available systems and sensors. if we want to stream 16 channels of temperature data...no problem, how about a couple channels of temperature, a pressure channel and perhaps a flow sensor and some inputs indicating when switches are toggled to start a shot...no problem...that is what I bring to the party.

Cheers,
Sean

#25: Post by **lennoncs** » November 16th, 2005, 9:45 pm

another_jim wrote:Use a straight line profile and an actuator made of physical matter, rather than the divine stuff -- a lot cheaper, same effect, since last time I looked, infinite rates of change were still illegal in this universe.

Ha!

Actually I was referring to how I represent the user input to specify a profile...think of the user interface to specifying how you want a pressure profile to look.

The problem is that this "pump" is a servo piston machine with a 1/2Hp servo motor doing the work. if you say "go from 0psi to 30psi in 1/2 second and hold that pressure for 20 seconds, after which proceed to 135psi in 1 second and hold that pressure for 26 seconds and then back to 0...it does that just fine except that the transitions or "corners" are very abrupt....I am working with specifying the "corner rounding" so to speak.

an interesting side effect of the machine is that you can also specify volume..for instance; hold 135psi for 26 seconds OR 1.5oz, then go to 0. or perhaps you want to specify an exact volume for a flush...not a problem, torque control-position control...your choice or both.

Sean

November 16th, 2005, 11:39 pm

lennoncs wrote:
Ha!

Actually I was referring to how I represent the user input to specify a profile...think of the user interface to specifying how you want a pressure profile to look.

The problem is that this "pump" is a servo piston machine with a 1/2Hp servo motor doing the work. if you say "go from 0psi to 30psi in 1/2 second and hold that pressure for 20 seconds, after which proceed to 135psi in 1 second and hold that pressure for 26 seconds and then back to 0...it does that just fine except that the transitions or "corners" are very abrupt....I am working with specifying the "corner rounding" so to speak.

an interesting side effect of the machine is that you can also specify volume..for instance; hold 135psi for 26 seconds OR 1.5oz, then go to 0. or perhaps you want to specify an exact volume for a flush...not a problem, torque control-position control...your choice or both.

Sean

Your're right of course, I was just being envious -- I rather detest that I can't get my roaster's PID to just french curve along the time/temperatue points in the ramp, rather than running in a straight line. it would cut down on the instabilities tremendously. I have a bunch of nonparametric curve fitters in my stats software that work great - basically these use a straight polynomial interpolation algorithms but weight the distant point downward, yielding "fractional power" interpolations.

#27: Post by **lennoncs** » November 17th, 2005, 5:31 am

another_jim wrote:I have a bunch of nonparametric curve fitters in my stats software that work great - basically these use a straight polynomial interpolation algorithms but weight the distant point downward, yielding "fractional power" interpolations.

Very nice, what kind of output do they generate? plottable data?

Robots often use cubic B-splines for blended motion, they are quite nice...right up to the point where I tried coding one on a small embedded control...I became quite happy with the jerky transitions I had and kept my brain cells intact.

sean

November 17th, 2005, 6:08 pm

lennoncs wrote:
Very nice, what kind of output do they generate? plottable data?

Mostly. They work as part of a non-parametric regression package, so they also put out estimates for the best parametric models. I think they can be set up to also estimate the best order for b-spline interpolators like loess; but I'd have to dig out all the manuals. The regression package is part of a standard math subroutine package, that's probably free by now, email me if you want me to dig out the references.