Tricked-out Silvia's brew water temperature measured via WBC protocol

[AndyS]

Here is a chart of the temperatures retrieved by performing a WBC Protocol test on my Tricked-out Silvia. Sometime soon I'll get smart (like Ken did) and get a real datalogger. Meanwhile, I snapped photos at one second intervals of the Fluke display (monitoring a Scace device) and a digital timer, then entered the data into Excel by hand.

The WBC Protocol doesn't allow you to do any custom flushing procedures. It's pretty much a one size fits all approach; you do a two second flush before and after every shot. The data could easily have been made better with extra flushing before the first two shots, but that's not allowed.

This chart shows 14 WBC-protocol shots; the interval between them varies from 10 minutes to 10 seconds:

Considering the fact that there are 14 shots buried in that graph, the temp stability is pretty damn good (almost as good as a Cimbali Junior <g>).

Of course, the whole point of this is to try and get temperature under control in the service of better coffee. Five years ago when I started dabbling in this madness, I wanted to test Schomer's theory that small (eg, <1F) temperature differences made a noticeable difference in the quality of the espresso. Despite various claims that have been made over the years, I don't believe anyone has presented good blind tasting evidence in support of this (please correct me if I'm wrong). I guess we're getting to the point where the hardware is nearly good enough to run the tests. (Maybe Versalab got there a couple years ago, but we have no data on that machine).

Now I just need a lot of coffee, time, a lot of patience, and some good volunteer tasters....

[AndyS]

Here's the raw data:

[erics]

Hi Andy -

Firstly, much thanks for posting this data on our beloved lady. To me, it clearly shows that she, in the hands of a worthwhile operator, can compete with just about anyone, provided she is intelligently modified. Surely the Rancilio Police are watching.

Don't you believe it would be quite a task to test temperature differences in the 1.0 F range and have the ONLY difference being brew temp? Like - what temperature are we talking about and where would one measure? Did you use 17.5 grams or was it 18.5 grams ? Did you pack to 30 lbs or was it 25 lbs? Was that pump at 9.00 bar or was it 9.05? Was it a 2.0 ounce shot or 1.75? etc. etc. etc.

I assume this is Silvia with boiler PID, GH heater PID, preheated feedwater, custom OPV & rotary pump. Perhaps some new readers would be interested in knowing just how modified the ol' girl is. Again, much thanks for posting this info.

Eric S.

PS: 25 adaptors sold - lots of $ to CoffeeKids but not one to a Silvia owner.

[miKe mcKoffee]

Andy,

My hat's off to ya for the effort put in to capture and manually log all that data. Both the work and results impressive to say the least. If someone not familiar with just how unstable Silvia is stock, running same WBC protocol comparison would be shockingly revealing to say the least.

The stability you've attained with your Missy almost makes me wish I hadn't upgraded to a Bric', almost. Of course, calling your machine a Silvia is tantamount to calling a Top Fuel Funny Car with a 1960 style Vwee Bug body a Volkswagen.

[AndyS]

Don't you believe it would be quite a task to test temperature differences in the 1.0 F range and have the ONLY difference being brew temp? Like - what temperature are we talking about and where would one measure? Did you use 17.5 grams or was it 18.5 grams ? Did you pack to 30 lbs or was it 25 lbs? Was that pump at 9.00 bar or was it 9.05? Was it a 2.0 ounce shot or 1.75? etc. etc. etc.

A LOT of work. To do it well, you might have to do it over several times, because you always realize the mistakes you're making when you're two thirds of the way through.

I guess you'd want to start by picking one dose, one extraction pressure, one shot volume, etc. Then you go up and down the temperature scale as many times as necessary for your tasters to identify consistent flavor trends. The better you can reproduce all the other variables, of course, the fewer test runs you'll have to make. No small task.

I suppose it might help to use a coffee that's reputed to be very temperature sensitive, such as Schomer's.

[HB]

I suppose it might help to use a coffee that's reputed to be very temperature sensitive, such as Schomer's.

How helpful that would be is proportional to how representative such reputedly temperature sensitive blends are to typical espresso blends, would it not? Intentionally choosing a "niche" blend that amplifies the importance of temperature reproducibility seems more academically interesting than practical. Well, unless you're one of the people who loves said "niche" blend and no other.

[AndyS]

Intentionally choosing a "niche" blend that amplifies the importance of temperature reproducibility seems more academically interesting than practical. Well, unless you're one of the people who loves said "niche" blend and no other.

Certainly, you're correct. Probably a reasonable (and time-consuming) approach would be to try several different coffees, some with a reputation for sensitivity, some not.

It doesn't make sense to ignore a "niche" blend. After all, I've always believed the old adage, "if you've got a niche, scratch it."

Bada boom.

[Ken Fox]

Here is a chart of the temperatures retrieved by performing a WBC Protocol test on my Tricked-out Silvia. Sometime soon I'll get smart (like Ken did) and get a real datalogger. Meanwhile, I snapped photos at one second intervals of the Fluke display (monitoring a Scace device) and a digital timer, then entered the data into Excel by hand.

The WBC Protocol doesn't allow you to do any custom flushing procedures. It's pretty much a one size fits all approach; you do a two second flush before and after every shot. The data could easily have been made better with extra flushing before the first two shots, but that's not allowed.

This chart shows 14 WBC-protocol shots; the interval between them varies from 10 minutes to 10 seconds:

image: http://i17.photobucket.com/albums/b62/shekk/wbc-2.png

Considering the fact that there are 14 shots buried in that graph, the temp stability is pretty damn good (almost as good as a Cimbali Junior <g>).

Of course, the whole point of this is to try and get temperature under control in the service of better coffee. Five years ago when I started dabbling in this madness, I wanted to test Schomer's theory that small (eg, <1F) temperature differences made a noticeable difference in the quality of the espresso. Despite various claims that have been made over the years, I don't believe anyone has presented good blind tasting evidence in support of this (please correct me if I'm wrong). I guess we're getting to the point where the hardware is nearly good enough to run the tests. (Maybe Versalab got there a couple years ago, but we have no data on that machine).

Now I just need a lot of coffee, time, a lot of patience, and some good volunteer tasters....

Andy,

Firstly, thanks for doing this; it shows a lot of work and for that I commend you.

I must add, however, that you cheated by using a low initial boiler temp to produce a low initial shot temp. I never attempted this series with an initial shot temp of 198F. Had I done so I think the graphs I produced would have been much prettier, as my machines only dropped down to a bit below this range having started out several degrees above it. I believe there is a problem with the amperage one can get out of a typical 120v home circuit that makes producing shots (at say) 203F a lot harder on the WBC protocol, especially towards the end of it, very difficult. I never tried to get 198F shots consistently so I'd bet if I'd tried that it would have been "easier."

I'm not going to suggest you repeat this because I fully appreciate how much work it was, but I would suggest that had you started with a brew temp of 203F or so you would have had very different results.

Best,

ken

[AndyS]

you cheated by using a low initial boiler temp to produce a low initial shot temp. I never attempted this series with an initial shot temp of 198F. Had I done so I think the graphs I produced would have been much prettier, as my machines only dropped down to a bit below this range having started out several degrees above it. I believe there is a problem with the amperage one can get out of a typical 120v home circuit that makes producing shots (at say) 203F a lot harder on the WBC protocol, especially towards the end of it, very difficult. I never tried to get 198F shots consistently so I'd bet if I'd tried that it would have been "easier."

I'm not going to suggest you repeat this because I fully appreciate how much work it was, but I would suggest that had you started with a brew temp of 203F or so you would have had very different results.

Huh, I'm surprised to hear you say that. The temp I used was simply what I'd been using in the morning to pull shots. It wasn't a "low" setting selected to give some advantage. I almost never pull shots at 203F anyway. Do you?

As far as I know there's no problem with getting the power out of the wall. Heating water to 203F from (say) 60F takes only 3.6% more power than heating the same water to 198F. (203-60)/(198-60). My PIDs were certainly not going full out, or close to it.

Curiously, I don't see a particular temperature standard in the WBC protocol. I guess it's left up to the organizers.

If there was steaming involved too, then 115v/20a would certainly be a limitation. But in this WBC test, you don't do steaming. That's a good thing, because it would be tough on Silvia!

Another thing is, I ran the tests out to 30 secs per shot. You only have to go to 25 secs for the WBC standard. If I had cut each shot at 29 sec instead of 30, that in itself would have made up almost all the energy difference you cite.

I'm not particularly anxious to run the series again (unless you lend me the datalogger), but I'll at least run the last (high energy) part of it to see if keeping up with the demand is problematic.

[miKe mcKoffee]

I believe there is a problem with the amperage one can get out of a typical 120v home circuit that makes producing shots (at say) 203F a lot harder on the WBC protocol, especially towards the end of it, very difficult. I never tried to get 198F shots consistently so I'd bet if I'd tried that it would have been "easier."

This intrigued me and my gut didn't think it would hold up. Seemed to me 120v should be enough juice but who knows. So I just ran the WBC protocol on my rotary Bricoletta. Boiler is set at 1-1.2bar toggle. I manually recorded temps at 5, 10, 15, 20 & 25 sec (end of shot) points. The anticipated fade towards the end with that 10sec idle time series finish was a surprise.

1- 214.9 213.8 213.0 212.8 212.6
2- 215.2 214.9 214.0 213.7 213.5
3- 216.7 216.0 214.7 214.2 214.0
4- 216.9 216.1 215.0 214.4 214.2
5- 216.9 215.9 214.9 214.5 214.3
6- 217.4 215.8 214.9 214.5 214.2
7- 216.7 215.5 214.5 214.1 213.9
8- 216.5 215.8 214.8 214.1 214.1
9- 216.6 215.6 214.6 214.1 213.9
10- 215.7 215.0 214.3 214.1 213.9
11- 215.7 215.1 214.3 213.9 213.5
12- 216.2 215.2 214.3 213.7 213.3
13- 216.0 215.1 214.4 214.1 213.9
14- 215.9 215.7 214.7 214.3 214.0

Hmmm fade didn't seem to happen, guess the Bricoletta's most powerful heater in it's class does kick butt and lives up to it's commercial rating.

Obviously shots at those temps would suck, but we all know the WBC protocol does not allow for surf flushes, just those fixed 2sec before and after shot flushes. I always felt the Bric' could be tuned for a fairly reliable set target shot temp after initial HX cooling E61 group heating flush and did have it way down at 0.8bar at one time for a nominal surf. But steaming was too weak so bumped it back up to 1bar. Back to the idea of learning more about the Bricoletta's particular HX mechanics and how to tune the HX itself down about 10f. Or better yet tune the HX down even further and bump the boiler up a bit more.

[Ken Fox]

Huh, I'm surprised to hear you say that. The temp I used was simply what I'd been using in the morning to pull shots. It wasn't a "low" setting selected to give some advantage. I almost never pull shots at 203F anyway. Do you?

As far as I know there's no problem with getting the power out of the wall. Heating water to 203F from (say) 60F takes only 3.6% more power than heating the same water to 198F. (203-60)/(198-60). My PIDs were certainly not going full out, or close to it.

Curiously, I don't see a particular temperature standard in the WBC protocol. I guess it's left up to the organizers.

If there was steaming involved too, then 115v/20a would certainly be a limitation. But in this WBC test, you don't do steaming. That's a good thing, because it would be tough on Silvia!

Another thing is, I ran the tests out to 30 secs per shot. You only have to go to 25 secs for the WBC standard. If I had cut each shot at 29 sec instead of 30, that in itself would have made up almost all the energy difference you cite.

I'm not particularly anxious to run the series again (unless you lend me the datalogger), but I'll at least run the last (high energy) part of it to see if keeping up with the demand is problematic.

Actually, I think 208.7 is the sweet spot for espresso brewing

The last time these issues of 110v and the WBC came up, I seem to recall Greg saying that in fact there was supposed to be a longer time period in between shots than the procedure seems to indicate; I forget exactly what he said. My interpretation of the procedure AS WRITTEN, is that you pull a shot, flush a little, put the PF back in, wait the 10 minutes or 10 seconds, the remove the PF for 15 seconds with a small flush at the end and then lock the Scace and pull the next shot. I believe Greg said there was another 15 seconds or more in there.

Exactly how did you do this test, Andy? I can tell you that neither of my machines are capable of putting out 75ml + maybe 5 mls on each side as flushes, every 10 seconds plus 15 seconds plus 30 seconds (works out to be 85 mls every 55 seconds for the last number of shots) without steadily dropping in temperature. This volume output is well beyond the rated shotmaking capacity of the machines on spec sheets, as well.

ken
p.s another issue which I haven't seen discussed is where this 75ml standard comes from that Scace is calibrated for. Pulling the sort of short shots that good cafes do, does this require anything like THAT MUCH water?

[HB]

p.s another issue which I haven't seen discussed is where this 75ml standard comes from that Scace is calibrated for. Pulling the sort of short shots that good cafes do, does this require anything like THAT MUCH water?

Have you weighed your pucks lately? IIRC, they absorb around 30-35ml of water.

[erics]

The 75 ml in 25 ± 5 seconds comes from the WBC procedure for measuring brew water temp. Copied here for convenience:

4.3 Adjustment of Brew Water Flow Rate:
Specification: The flowrate of water through the measurement portafilter shall be measured by a graduated beaker and shall be adjusted so that 75 ml of water is collected in an elapsed time of 25 seconds ±5 seconds. The allowable volumetric tolerance of the beaker graduations shall be ±5% or less.

Technical rationale: The specified water volume was determined by direct volumetric measurement, using the Specialty Coffee Association of America (SCAA) standard espresso volume of 60 to 74 ml as the target volume of a double espresso after brewing. An espresso machine was calibrated to produce double espresso shots of 60ml in volume. The amount of water used to produce these shots was measured by brewing a series of "blank shots" into a graduated beaker. The measured volume was approximately 75 ml. The following sections account for the difference in water produced by the blank shots, compared to the brewed espresso.

Water absorption by the coffee cake: A series of measurements were performed, in which a dry portafilter was dosed with coffee and weighed. The portafilter was reweighed after espresso was brewed from the coffee. The measurements show that a dry coffee cake absorbs approximately its own weight in water during the brewing process, about 18ml for a double espresso shot.

Volumetric measurement of crema: A series of 60 ml double espressos were brewed into a graduated beaker, using coffee that had rested for 5 days after roasting. The coffees were allowed to stand for 5 minutes, allowing the majority of the crema to dissipate. The volume reduction due to crema dissipation was approximately 20 ml.

Residual water: Not all of the water that is pumped through the flowmeter percolates through the coffee. During brewing, water fills the space immediately above the coffee, and plumbing within the group. This water is ejected to the drain when pressure is relieved after brewing.

As far as the amount of water absorbed by the coffee cake, I measured several times (whereas the standard writers probably did it several dozen times) a year ago and it was always easily within their findings, i.e. the saturated coffee cake doubled in weight.

I, for one, applaud this standard because it, in whole or in part, allows for easy comparisons between machines A, B, & C with users D, E, & F.


Eric S.

[miKe mcKoffee]

The last time these issues of 110v and the WBC came up, I seem to recall Greg saying that in fact there was supposed to be a longer time period in between shots than the procedure seems to indicate; I forget exactly what he said. My interpretation of the procedure AS WRITTEN, is that you pull a shot, flush a little, put the PF back in, wait the 10 minutes or 10 seconds, the remove the PF for 15 seconds with a small flush at the end and then lock the Scace and pull the next shot. I believe Greg said there was another 15 seconds or more in there.

As I understood and ran the protocol there's one more 10sec step, the puck dump following the shot, which as I interpreted the directions included the post shot 2sec flush in the same 10sec puck dump time frame before starting next idle period.

[AndyS]

The last time these issues of 110v and the WBC came up, I seem to recall Greg saying that in fact there was supposed to be a longer time period in between shots than the procedure seems to indicate; I forget exactly what he said. My interpretation of the procedure AS WRITTEN, is that you pull a shot, flush a little, put the PF back in, wait the 10 minutes or 10 seconds, the remove the PF for 15 seconds with a small flush at the end and then lock the Scace and pull the next shot. I believe Greg said there was another 15 seconds or more in there.

My interpretation of the protocol is this, which is approximately what I did for the test:



It comes out to be about 59 seconds plus the idle time between shots. So in the last phase (with 10 sec idle times) it's 69 secs between shots.

I can tell you that neither of my machines are capable of putting out 75ml + maybe 5 mls on each side as flushes, every 10 seconds plus 15 seconds plus 30 seconds (works out to be 85 mls every 55 seconds for the last number of shots) without steadily dropping in temperature. This volume output is well beyond the rated shotmaking capacity of the machines on spec sheets, as well.

Really? Below is a summary of the power requirements to keep up with the 69 sec shot rate. It assumes a 50F incoming water temperature, flushes of 12.5ml each, 90% machine efficiency, among other things:



As you see, your 115v wall outlet should easily be able to handle the load. You must have other issues. Perhaps your PID tuning isn't getting the job done. Or perhaps your heat exchanger can't transfer enough heat to keep up with this shot rate.

p.s another issue which I haven't seen discussed is where this 75ml standard comes from that Scace is calibrated for. Pulling the sort of short shots that good cafes do, does this require anything like THAT MUCH water?

Back when I had the flowmeter installed, my machine expended a total of about 55 ml of water in pulling a double shot (filling the pipes, wetting the cake, extracting the espresso, etc). So 75ml would be a bit high for my technique, but YMMV.

[Ken Fox]

Ken Fox wrote:

The last time these issues of 110v and the WBC came up, I seem to recall Greg saying that in fact there was supposed to be a longer time period in between shots than the procedure seems to indicate; I forget exactly what he said. My interpretation of the procedure AS WRITTEN, is that you pull a shot, flush a little, put the PF back in, wait the 10 minutes or 10 seconds, the remove the PF for 15 seconds with a small flush at the end and then lock the Scace and pull the next shot. I believe Greg said there was another 15 seconds or more in there.


My interpretation of the protocol is this, which is approximately what I did for the test:



It comes out to be about 59 seconds plus the idle time between shots. In the last phase it's 69 secs between shots.

Ken Fox wrote:

I can tell you that neither of my machines are capable of putting out 75ml + maybe 5 mls on each side as flushes, every 10 seconds plus 15 seconds plus 30 seconds (works out to be 85 mls every 55 seconds for the last number of shots) without steadily dropping in temperature. This volume output is well beyond the rated shotmaking capacity of the machines on spec sheets, as well.


Really? Below is a summary of the power requirements to keep up with the 69 sec shot rate. It assumes a 50F incoming water temperature, flushes of 12.5ml each, 90% machine efficiency, among other things:



As you see, your 115v wall outlet should easily be able to handle the load. You must have other issues. Perhaps your PID tuning isn't getting the job done. Or perhaps your heat exchanger can't transfer enough heat to keep up with this shot rate.

Ken Fox wrote:

p.s another issue which I haven't seen discussed is where this 75ml standard comes from that Scace is calibrated for. Pulling the sort of short shots that good cafes do, does this require anything like THAT MUCH water?


Back when I had the flowmeter installed, I used about 55 ml in preparing a shot.
_________________
-AndyS

I apologize for how this is going to look when posted but I don't have the energy to edit it with HTML and Dan's software quotes only the last sentence. Hence, I pasted all this crap in for context but only some of it shows up.

Thanks for clarifying all this, Andy.

Long ago I gave up on the WBC test as irrelevant to my usage pattern and equipment. Nonetheless, the curves you post are incredibly good ones, way beyond the capability of my equipment, which just to restate for those unfamiliar with them, are PID'd heat exchanger machines whose only parameter controlled by the PID is the boiler temperature. I think you should be proud of what you have accomplished, probably the most stable curves anyone has produced on any piece of equipment out there, stock or hacked. Congratulations.

Getting back to my own equipment (as self-centered as I am), my own interest at the moment is shifting. While I can think of several ways that the temperature stability of my machines might be improved, I'm haunted by the fact that I seem to prefer shots made on my old vibe machine when compared to my newer rotary machine. Since the pressure ramp ups on the two machines are more similar than not, after all the hacks, I'm left with the curve shape as being the most likely thing that makes me prefer shots from the vibe over shots from the rotary. The rotary machine now produces more or less flat espresso extraction curves and the old vibe retains the "heat exchanger hump." Probably I could duplicate the hump on the rotary by increasing boiler temperature and perhaps also flush volume. Since in the end I can't drink temperature stability but am reduced to drinking espresso shots, to me right now the money is in trying to find out whether the humped curve is in fact what I like in an espresso shot more than absolute temperature stability. The problem with this is that it ends up being a matter of taste and without some blind tasting data the data I might produce would be worthless, since I'm going to know what I'm drinking in advance unless I can get Jim Schulman back for some more taste testing. So, as easy as graphing is, tasting and evaluating on curve shapes will be much harder.

But maybe I'll just do it for my own fun and amusement and then it won't matter.

ken

[AndyS]

The rotary machine now produces more or less flat espresso extraction curves and the rotary retains the "heat exchanger hump."

You've got a typo in there.

[Ken Fox]

You've got a typo in there.

Yeah, just fixed it. Chalk it up to brain asphixia from having to breath LA smog for 2 days during the past weekend's festivities.

ken

[HB]

I apologize for how this is going to look when posted but I don't have the energy to edit it with HTML and Dan's software quotes only the last sentence. Hence, I pasted all this crap in for context but only some of it shows up.

Sorry, it's an imperfect response to the alties who add "Yes, I agree" beneath two pages of quoted text when said post is directly above their own.

See this button for an alternative solution:

[AndyS]

Long ago I gave up on the WBC test as irrelevant to my usage pattern and equipment.

I hear you, but what's valuable about the WBC test is that it reveals machine behavior at varying duty cycles. That is important in real world usage, even home real world usage.

Since in the end I can't drink temperature stability but am reduced to drinking espresso shots, to me right now the money is in trying to find out whether the humped curve is in fact what I like in an espresso shot more than absolute temperature stability.

Agree 100%. What's really unfortunate about the WBC test is that it rewards a certain machine behavior ("flat temperature profile"). Meanwhile, nobody's demonstrated that the flat profile is really optimum for any particular coffee.

Teahan says, that the Italians say, it's NOT optimum. Of course, what do they know, anyway? :-0

I'm still hoping on my Silvia that I can get more of a hump at the beginning by running the grouphead heater a little hot. Unfortunately, then it becomes more and more like a heat exchanger machine, with all the attendant duty-cycle instabilities.