HomeRoast Digest


Topic: Screening coffee, Some first results. (14 msgs / 286 lines)
1) From: R.N.Kyle
The salton was certainly a bean basher wasn't it?!
You can't beat my Sultan up anymore then I already have. It was never =
very good, even when it was new. It was sold by QVC as a Cooks =
Essential. 
I thought any test needed a variety from the best to the worst, I new =
the Sultan flat burr would be at the bottom.
Ron Kyle
Anderson SC
rnkyle

2) From: jim gundlach
Today was my 60th birthday and one of the things I got was the time to 
do a few things I wanted to.  One thing I did was make a mess of the 
living room by running two samples of coffee grounds through the set of 
sizing screens I put together.  The first one was a sample from my 
Rocky set at notch nine, a little coarser than the six I use for 
espresso.  The second one was from Ron Kyle's worn out flat burr Sulton.
The way I conducted the screening was to take the sample, they were not 
the same size, and run them through each screen starting with the 
coarsest,  which has 12 wires per inch, and is shown on the graph 
available at the URL below as "12".  If all the sample passed through a 
screen that sample was scored a zero for that screen.  If any of the 
sample was retained by the screen, it was measured by the following 
makeshift method.  I poured the sample into a clear drinking straw and 
measured the height of the part of the sample caught by that screen in 
millimeters.  I continued down through screen size 60.  The last line 
was the dust that went through screen 60.  My plan was to construct an 
bar graph of one straw for each of the screens and take a picture of it 
but that plan failed because the Rocky sample required more than one 
straw to hold the coffee caught by screen 42.  I also discovered that 
the batteries were dead on my camera again so I just used one straw to 
measure with.  The next time I am in a place where I can find some I 
will buy some measuring flasks that will allow me to measure volume in 
a standard measure rather than this makeshift  millimeter of a clear 
drinking straw of an unknown diameter.  Also, I was having to take care 
of my four year-old granddaughter while doing it and I think that in 
trying to do too many things at once I may have mixed up two of the 
measures from the Sulton.  To a large degree this was a learning 
experience.  For those with some statistical background, the kurtosis 
for the Rocky was 63 and the Sulton .03.
Next, I would like for some of the others who got the sets of screens 
to get theirs put together and we could then go about replicating tests 
of different grinders.  I would like to get a couple of samples from a 
Mazzer to see how it compares with the Rocky.  By the way, Ron believed 
his Sulton was doing a bad job and hoped the data would demonstrate the 
need for a new grinder.  I think it does.
Anyway, you can see the results, and guess where the data may be mixed 
up,  at:
   http://homepage.mac.com/jimgundlach/PhotoAlbum4.htmlThe page has two pictures, click on the right one to see the graph.
Jim Gundlach
Roasting over pecan wood fires
in La Place, Alabama
homeroast mailing listhttp://lists.sweetmarias.com/mailman/listinfo/homeroast

3) From: Jim Schulman
Terrific work!
Looks like the technique is a success. Exact 
weights or volumes aren't really necessary, as 
long as each grind sample is measured 
consistently. 
Obviously, the Rocky performs better than a worn 
out Salton; but it looks like the same test could 
show any difference between grinders that are 
regarded as good to excellent, i.e. the Solis, 
Innova, Rocky and Mazer. Obviously, someone should 
send you one of each, gratis, for testing ;)
Maybe I'm missing something, but wouldn't standard 
deviation rather than kurtosis be the proper 
single quantity measure of grinder quality (or 
rather lack of it).
Jim Schulman
On 15 Dec 2002 at 20:32, jim gundlach wrote:
<Snip>
homeroast mailing listhttp://lists.sweetmarias.com/mailman/listinfo/homeroast

4) From: Dan Bollinger
Great job!  I'm curious how the other grinders will perform, especially my
MM.  Feel free to mail me the screenings, I have a gram scale good to 0.1
gram.  See the 'bell' curve is very informative.  Dan

5) From: jim gundlach
On Sunday, December 15, 2002, at 08:54 PM, Jim Schulman wrote:
<Snip>
You are right, the standard deviation is the traditional measure of 
variation and a more consistent grind would lead to a smaller standard 
deviation.  Kurtosis is a measure of peakedness of a distribution which 
corresponds to the amount of variation but is not preferred because the 
values of kurtosis do not have the interpretive value that the standard 
deviation does. Rather it tells us how a distribution deviates from a 
normal distribution.  Values greater than one are more peaked and thus 
have less variation than a normal distribution, and values less than 
one are flatter and have more variation than a normal distribution.  
However,  when you have a variable where the levels are not continuous, 
as in this case where the screen is measured as number of wires per 
inch and the wires have different diameters, the numeric values of the 
standard deviation are not comparable when the means have different 
locations.   In this case, the hole size generally gets smaller as 
number of wires get larger.  However, the screen with 42 wires actually 
has larger holes than the screen with 40 wires.  That is why they are 
reversed on the graph.  Before I conduct the next analysis I will 
calculate the size of the holes and use those as the values for the 
screens.  But, I have to go back to McMaster-Carr's catalog and get the 
size of the wires of a couple of the screens before I can calculate all 
of them.  Once I get hole size instead of number of wires as the screen 
label I can compare the standard deviations.  In general you are right, 
the standard deviation is the preferred measure and once I get the data 
in shape, I can use it.
Jim Gundlach
homeroast mailing listhttp://lists.sweetmarias.com/mailman/listinfo/homeroast

6) From: Jim Schulman
Thanks for the clarification. I didn't realize 
that the screen sizes aren't in a neat linear 
series. 
If the conventional ideas of grind quality are 
correct, a grinder that produces an uneven grind 
evenly spread of over three contiguous screen 
sizes is probably better than one that produces 
say 75% at one screen size, and 12.5% as boulders, 
and the other 12.5% as dust. So a measure that 
penalizes extremes even more than a standard 
deviation is probably required. A non-parametric 
version of Kurtosis may be the ticket.
Jim Schulman
On 15 Dec 2002 at 21:21, jim gundlach wrote:
<Snip>
homeroast mailing listhttp://lists.sweetmarias.com/mailman/listinfo/homeroast

7) From: Simpson
Jim, very cool. I am surprised on the Rocky front since I had the notion
from the Illy book that an espresso grind was a bimodal distribution where
the fines were important for hydraulic control in the puck. I would have
expected another, smaller peak at a small grain size. Hmm.
The salton was certainly a bean basher wasn't it?!
Ted
*********** REPLY SEPARATOR  ***********
On 12/15/2002 at 8:32 PM jim gundlach wrote:
<Snip>
snip
homeroast mailing listhttp://lists.sweetmarias.com/mailman/listinfo/homeroast

8) From: jim gundlach
On Sunday, December 15, 2002, at 10:02 PM, Simpson wrote:
<Snip>
I'm sure it is possible, but right now I'm having difficulty imagining 
how you would make a grinder create a bimodal distribution.  It may be 
the Illy blends the output of two grinders to produce their espresso.  
Guess I will have to get some to find out if it is bimodal.
  Jim Gundlach
homeroast mailing listhttp://lists.sweetmarias.com/mailman/listinfo/homeroast

9) From: Rick Farris
On 15 Dec 2002 at 20:32, jim gundlach wrote:
<Snip>
It's very interesting that the right hand tail on each is virtually
identical.  I wonder if that is an effect of the crushing action of the
grinder.  But wait...doesn't that indicate that they both have the same
amount of coffee dust?
-- Rick
homeroast mailing listhttp://lists.sweetmarias.com/mailman/listinfo/homeroast

10) From: Dan Bollinger
Jim,  You are right, the hole size changes and sometimes not in the
direction you expect.  You can calculate the hole size if you have a
micrometer or caliper to measure the wire diameter. Dan
<Snip>
homeroast mailing listhttp://lists.sweetmarias.com/mailman/listinfo/homeroast

11) From: rev mark gilstrap
<Snip>
An optical comparator is the tool of choice for measuring
hole diameters and web (wire) thickness (the wires wear 
with time when sifting material more abrasive than coffee).  
Inexpensive handheld ones are available from Edmund 
Scientific.  At work we use a Nikon scope with projection 
onto ~ a 20'' diameter ground glass field, but that is a Fortune 
500 solution.
homeroast mailing listhttp://lists.sweetmarias.com/mailman/listinfo/homeroast

12) From: jim gundlach
Both produced very little dust.  It may be worth noting that both 
grinders use flat burrs.  I should have noted that Ron sent two 
samples, a drip and an espresso grind.  This is his espresso grind.  
When I built this set of screens I though the bigger screens were 
probably unnecessary.  I don't think so now.
    Jim Gundlach
On Sunday, December 15, 2002, at 11:37 PM, Rick Farris wrote:
<Snip>
homeroast mailing listhttp://lists.sweetmarias.com/mailman/listinfo/homeroast

13) From: Ben Treichel
R.N.Kyle wrote:
<Snip>
With pieces of the burr at the bottom it gives a whole new meaning to 
chunks ;-)
<Snip>

14) From: susan oppenheim
HAPPY 60TH JIM
jim gundlach wrote:
<Snip>
homeroast mailing listhttp://lists.sweetmarias.com/mailman/listinfo/homeroast


HomeRoast Digest