www.klatchroasting.com: USBC champion, voted 2009 'best micro-roaster'

Temperature study of Alex Duetto

Postby JimG on Fri Sep 26, 2008 11:36 pm

Chris' Coffee was kind enough to lend a new Alex Duetto, and over the past week I finished a short series of temperature tests. My test machine is a 115V version as sold in the US. It has a Gicar PID with 1F temperature resolution.

I instrumented the machine with a couple of additional sensors: 1) a K thermocouple fastened to the top of the brew boiler (see photo below); and 2) a K thermocouple inserted into the grouphead using one of Eric's E-61 adapters.

The stock machine uses a NTC thermistor, inserted into a thermowell in the brew boiler, for PID control of brew boiler temperatures. The steam boiler is controlled using a traditional pressurestat.

Fresh water is fed into the brew boiler through a heat exchanger that passes through the steam boiler. The fresh water supply to the brew boiler enters through the top. When the steam boiler is on, the supply water to the brew boiler is preheated by the HX. But the steam boiler can be switched on, or off, independently of the main machine switch, allowing operation either with, or without, preheat of the brew boiler supply water.

The objectives of this study were:
  • Determine optimum (or nearly optimum) tuning parameters for the Gicar PID
  • Examine intrashot temperature behavior
  • Examine recovery and intershot stability with a series of closely spaced shots
  • Compare behavior with steam boiler on (preheat) vs steam boiler off (no preheat)

Data were collected on three calibrated thermometers:
  • Omega HH506RA connected via RS-232 to laptop (for realtime plotting during PID tuning)
  • A pair of Fluke 54 II thermometers (data uploaded using Flukeview Forms)

The brew boiler thermocouple sensor consists of a 24 AWG type K bead probe soldered on to a thin brass plate. The plate was punched to allow the M4 stud from the safety thermostat to pass through. The safety thermostat was removed, thermal paste spread thinly on the mating surfaces, the new plate was placed against the boiler surface, and the safety thermostat was threaded back into the boiler, clamping the sensor tightly to the boiler. (I have had poor success using small bare bead probes in similar situations on other projects, and this plate assembly worked much better).

Image
JimG
 
Posts: 657
Joined: Jun 11, 2006
Location: Lexington, KY

Postby JimG on Fri Sep 26, 2008 11:54 pm

Test Protocol

I allowed the machine to warm up for several hours, with PF locked in, before logging any data. Each series of tests consisted of the following steps:

Time     Activity
0:00     remove PF, flush 3 oz water through grouphead
0:45     lock PF in place, begin data logging
1:00     start Shot 1
1:25     stop shot, remove PF
2:45     lock PF in place
3:00     start Shot 2
3:25     stop shot, remove PF
4:45     lock PF in place
5:00     start Shot 3
5:25     stop shot, remove PF
6:45     lock PF in place
7:00     start Shot 4
7:25     stop shot, remove PF


Brew water temperature estimates were recorded using a Scace thermofilter with type T thermocouple and a Fluke 54 II thermometer. Grouphead temperatures and boiler temperatures were recorded using type K thermocouples (previously described) and a second Fluke 54 II. Data were recorded at 1 second intervals.

For each series of shots, three plots were created:
  • combined plot of boiler temperature and grouphead temperature vs time
  • continuous plot of temperatures recorded on the Scace device vs time
  • overlaid plots of 4 individual 25-second duration shots

Next: temperatures logged with steam boiler OFF (no preheat)
JimG
 
Posts: 657
Joined: Jun 11, 2006
Location: Lexington, KY

Postby JimG on Sat Sep 27, 2008 12:13 am

Temperature Tests with Steam Boiler OFF (no preheat)

The Gicar PID was set to 220F. The steady average temperature reported by the thermocouple sensor added to the top of the boiler was approximately 217.6F. This difference may be due to any number of factors. But since this offset (~2.4F) remained relatively constant throughout testing with the steam boiler off, the readings from the thermocouple give a good indication of when the boiler temperature has stabilized.

In fact, the PID display generally showed a return to steady temperatures more than 30 seconds prior to the sensor on the top of the boiler stabilizing.

Image

Plot showing boiler temperature recovery between shots (above)

Image

Plot showing continuous record of Scace device temperatures during test session (above)

Image

Plot of instrashot profiles for 4 shots closely spaced (above) [Note: typo noted in legend for Shot 1. Correct average shot temp is 200.24]

The instrashot profiles had a very slight hump at the start, and then remained almost dead flat. The average shot temperatures (calculated by averaging 16 readings, from 10 seconds to 25 seconds) were as follows:
  • Shot 1 = 200.2F [EDIT: typo corrected 9/28/08]
  • Shot 2 = 200.1F
  • Shot 3 = 200.0F
  • Shot 4 = 199.7F

Next: temperature measurements with steam boiler ON (with preheat)
JimG
 
Posts: 657
Joined: Jun 11, 2006
Location: Lexington, KY

Postby JimG on Sat Sep 27, 2008 12:33 am

Temperature Tests with Steam Boiler ON (with preheat)

Following the conclusion of the "no-preheat" tests, I switched on the steam boiler and waited over an hour to let everything stabilize. An identical series of shots was pulled with the steam boiler ON.

Comparing the plot below to the corresponding plot for the no-preheat sequence immediately shows the effects of preheat provided by the steam boiler. The drop in brew boiler temperature during a shot is greatly reduced, and recovery to stable brew boiler temperature is considerably faster.

Image

Plot showing boiler temperature recovery between shots (above)

Image

Plot showing continuous record of Scace device temperatures during test session (above)

Image

Plot of instrashot profiles for 4 shots closely spaced (above)

Again, the instrashot profiles had a very slight hump at the start, and then remained almost dead flat. The average shot temperatures (calculated by averaging 16 readings, from 10 seconds to 25 seconds) were as follows:

* Shot 1 = 201.7F
* Shot 2 = 202.0F
* Shot 3 = 202.0F
* Shot 4 = 202.1F

For the same PID SV = 220F, shots made with preheat averaged approx. 2F hotter than shots made without preheat. A portion of this rise can be explained by a corresponding slight rise in the boiler temperature recorded by the brew boiler thermocouple and by the rise recorded in group temperature (from ~197F to ~199F). Possibly the remainder of the 2F rise is explained by the warmer supply water entering the brew boiler during the shot.

Comments and questions about the information posted here are welcome. Thanks for reading.

Jim
JimG
 
Posts: 657
Joined: Jun 11, 2006
Location: Lexington, KY

Postby CoffeeOwl on Sat Sep 27, 2008 8:54 am

It is interesting* to see there's quite no difference in temp profiles from steam boiler on to steam boiler off.


EDIT
* changed into 'interesting' from 'really nice'... I made this comment without reading the discussion in Duetto thread :oops: sorry everyone
'a a ha sha sa ma!


LMWDP #199
CoffeeOwl
 
Posts: 1071
Joined: Sep 30, 2007
Location: Lodz Poland

Postby Marshall on Sat Sep 27, 2008 10:13 am

jggall01 wrote:Test Protocol

I allowed the machine to warm up for several hours, with PF locked in, before logging any data. Each series of tests consisted of the following steps:

Time     Activity
0:00     remove PF, flush 3 oz water through grouphead
0:45     lock PF in place, begin data logging
1:00     start Shot 1
1:25     stop shot, remove PF
2:45     lock PF in place
3:00     start Shot 2
3:25     stop shot, remove PF
4:45     lock PF in place
5:00     start Shot 3
5:25     stop shot, remove PF
6:45     lock PF in place
7:00     start Shot 4
7:25     stop shot, remove PF


On another thread we have been speculating on the effects of ambient temperature changes on brew temperature. I also think most people turn their machines off at night and do not have the luxury of a several-hours warmup in the morning. Would it be possible to run tests that reflect some common usage patterns such as:

1. A first-shot-in the-morning in a cold kitchen after a 15 min. warmup.
2. Same after a 30 minute warmup (for those with a timer).
3. A warm mid-day kitchen after a 30-min warmup.
4. A cool evening kitchen after a 30-min warmup.

I'm asking this, because I think temperature offset stability is also important, and that the ability to pump out several identical shots in a row for guests may be less critical than coping with variations in room temperature and less-than-ideal warmup times. In other words, will my mid-afternoon espresso taste the same as my wake-up espresso? (This may be less of an issue for us in Southern California than for people who run the house thermostat low in frostier climes.)
Marshall
Los Angeles
User avatar
Marshall
 
Posts: 2643
Joined: May 13, 2005
Location: Los Angeles, California

Postby JimG on Sat Sep 27, 2008 12:14 pm

Marshall -

I don't have enough data to answer precisely. Instead, I can provide information regarding the likely temperature of the grouphead after varying amounts of warmup time (see plot below).

After 15 minutes warmup, the grouphead temperature is still 16F low. Very roughly, you might expect this to result in a shot that is ~8F low. So I would say the machine is not ready to go after 15 minutes.

After 30 minutes warmup, the grouphead is a little less than 3F low. Again very roughly, this might produce a shot 2F low, or maybe even better. So I would say that shots pulled after 30 min warmup would still be very drinkable, but not optimum.

On the Duetto, it took 35 minutes for the grouphead to be 99% warmed up, and 46 minutes to be 99.5% warmed up. I guess we could then consider 35 minutes to be a "good" warmup, and 46 minutes to be a "best" warmup.

I know this is not a full answer to your question. If time allows, I will measure some shot temperatures after 15 minutes and after 30 minutes and see how close my rough guesses were.

Jim

Image

Duetto warmup with steam boiler OFF, room ambient temp ~77F
JimG
 
Posts: 657
Joined: Jun 11, 2006
Location: Lexington, KY

Postby cafeIKE on Sun Sep 28, 2008 3:20 pm

Why 15s between lock in and pull?

I'd like to see a series with 4, 8 and 16 minute intervals, a 1oz. cleaning flush after the 'shot' and the PF left in the group until 30s before the shot.
User avatar
cafeIKE
 
Posts: 3083
Joined: Jun 27, 2006
Location: Woodland Hills, CA

Postby JimG on Sun Sep 28, 2008 4:46 pm

cafeIKE wrote:Why 15s between lock in and pull?

Primarily for the benefit of the tester (so he wouldn't mess up the intervals). I think you are implying that 15s is a long time. In hindsight, I probably could have shortened this from 15s to 5s or less and still stayed on track.

But examination of the data suggests this might not have had an important effect on the results:
  • Sequence no.; Scace temp at start of shot; average shot temp
  • 1; 157.2; 200.2
  • 2; 163.8; 200.1
  • 3; 168.9; 200.0
  • 4; 166.8; 199.7
I suggest that there would be better correlation between the start temp and the average temp if the lock-in timing were important. Instead, the difference in average shot temp is only 0.2F at the two extremes of the beginning Scace temperature (157.2F, 168.9F).

cafeIKE wrote:I'd like to see a series with 4, 8 and 16 minute intervals, a 1oz. cleaning flush after the 'shot' and the PF left in the group until 30s before the shot.

Good point. Those intervals would tell us more about the thermosyphon loop and its ability to maintain the grouphead temperature. My testing was slanted towards getting the PID tuned. If time allows, I'll pick one of those intervals (probably 8 minutes) and test as you suggested.

Jim
JimG
 
Posts: 657
Joined: Jun 11, 2006
Location: Lexington, KY

Postby Mole on Sun Sep 28, 2008 10:00 pm

Thanks Jim! This information is fascinating!

Actually, without wanting to push it, I'd be interested in two or three additional things as well...

1. Do a similar test to the HB review of the Alexia with PID, that is, 2 minutes recovery between shots, rather than 2 minutes from start of shot to start of shot. I would hope that the preheat off recovery would be improved significantly by the additional 25s recovery time and preheat on would be even more stable shot to shot.

2. A walk-up shot with no flush (both preheat off and on). That is, do we need to flush at all?

3. A vastly reduced initial flush: 3 oz seems way too much for a dual boiler, particularly with preheat off, and then to only give the boiler and thermosyphon less than 1 minute to recover before the first shot seems harsh. Where did 3oz come from?

4. A 1oz (3 second) flush sometime between shots (time to be determined!), to simulate "cleaning the group of grounds and coffee residue."

Marshall wrote:On another thread we have been speculating on the effects of ambient temperature changes on brew temperature.

Marshall, I agree entirely and would like to see some data on this. Given that you owned a PID'd Zaffiro for many months, have you any data to share on this? Did you find the need to keep tinkering with brew temps throughout a variable day? What was the warmup time like for you? I cannot imagine these issues would be vastly different.

Thinking about offset variability due to ambient changes: With a 20F offset from boiler to shot temp, in 70F ambient, at boiler temp of 220F, we have 20F offset for 150F temp delta from ambient to boiler. Would we therefore not expect no more than 1F delta in shot temp for each 7.5F delta in ambient temp? It would be interesting to test this. However, if this is the case, small tweaks can be predicted (if you know the ambient temp). I've yet to see whether such predictions come true, since the ambient in our house has been pretty consistent (18 to 20C) for the last few weeks. Perhaps in the coming winter months I will find out more?

Or, of course, use the PID control as a way to (resonably accurately) adjust by taste!

All the best,

Martin
Mole
 
Posts: 24
Joined: Dec 13, 2007
Location: London, UK