Category Archives: Sensors

Ordered parts this weekend for roaster controller.

By | | , , , , ,
1 Comment

I’ve ordered some parts from Maxim/Dallas as well as from Microchip this past weekend. The stuff from Maxim is supposed to be coming US Postal Service so I don’t really know when they will arrive. I forgot to add a temperature sensor (ambient) to the order though so I will need to come back to that later. I need to compare it with a similar chip offered by Microchip to see which one I want to use. I believe the chips by Microchip are actually smaller and more accurate but I will need to confirm later.

According to Microchip’s website / my invoice the IO board, graphics board and LCD touch screen, Prototype cards, and a 9V Power Supply will ship sometime in the next couple days. They are scheduled for arrival on the 17th. The Network Card and PIC32 Starter Kit is not due to be available for 2 weeks and is supposed to arrive on the 30th.

Building a roast controller system.

By | | , , , ,
Leave a comment

As alluded to earlier in the blog I am intending on building a roaster controller system.  I’m probably a bit of an over-achiever in terms of my projects because every time I create something on the internet for one of my websites, as part of a building project in the yard, or whatever else I’m creating I always over-analyze it.  In my career one of the areas I’ve been responsible for was “Disaster Recovery” planning for computer systems.  In other words my job was to over-analyze all the things that could go wrong and figure out a plan to deal with them.  At the same time I also have to plan all the ways for it to go right and have a plan so anyone can do it even without strong computer skills.

Focusing on a roast controller as mentioned I’ve already purchased some testing equipment for measuring temperature using a K type thermocouple. I am now finalizing the equipment that I’m going to use to build this and test it.  I’ve focused on using the PIC32 microprocessor due to its abilities to work easily with Ethernet, LCDs, and having many input/output circuits.  In addition it seems to be a robust platform for RTOS systems allowing a lot to be going on all at the same time.

To ease development I am choosing to embrace the development tools provided by the manufacturer.  This of course adds to the startup cost of the system but allows me to “try new things” in the future too using the development platform.  Once I finalize the project I would then order the individual components and manufacture my own circuit boards.  To start off I expect to need the following.

The Starting Parts

  1. DM320003-2 – PIC32 USB Starter Kit II – $55.00

    • Main processor (MCU) test board.
    • Contains PIC32MX795F512L processor.
    • 80 MHz with 512K Flash, 128K RAM. (this is actually a lot for a simple coffee roaster but is necessary for drawing graphical LCD, developing complex roasting profiles, and loading Ethernet overhead.)
    • Allows USB, Ethernet, LCD, I2C and SPI communications. 
  2. DM320002 – PIC32 I/O Expansion Board – $72.00

    • Connector to access pins and interface auxiliary boards.
  3. AC164132 – Fast 100Mbps Ethernet PICtail Plus Daughter Board –$49.99

    • Physical wired ethernet board for interface to DM320002.
  4. AC164126 – Prototype PICtail Plus Daughter Board – $20.00
    • Prototype boards to solder additional parts and extensions such as temperature probe chips, time devices, relays, etc.
  5. AC164127-3 – Graphics PICtail Plus Daughter Board with 3.2 Display Kit – $154.99

    • SSD1926 Graphics chip (kind of like a video card)
    • 3.2 QVGA (240X320) TFT LCD with 18-bit parallel RGB interface
    • Touch Screen

The Later Parts

The above parts allow me to start testing User Interface functionality without worrying too much about the circuit boards except for the occasional prototype board communicating with a sensor or two.  I’m doing that to determine the final touch screen configuration and “what is possible” on the screen real estate to determine if I want to pursue additional sensors and features.  It will also allow me to begin testing response time of various sensors on the prototype boards.  Some parts I expect to need:

  1. MAX6676 Thermocouple Amplifier
  2. DS3232 Real Time Clock
  3. Lots of resistors
  4. Lots of capcitors
  5. Various Transistors
  6. Lots of diodes
  7. Photosensitive PCBs
  8. PCB etch Chemicals and developer
  9. SMD soldering and hot air rework station
  10. Infrared circuit board pre-heater station
  11. Small Drill Press w/ bits for drilling PCBs
  12. Oscilloscope for dimmer testing
  13. Several relays and/or some dimmer circuits that need to be designed still.
  14. Probably a video camera at some point to demonstrate progress/results.

Roast temperature probe sweet spot in Fresh Roast

By | | , , ,
Leave a comment

As I mentioned previously there appears to be a “donut” shaped sweet spot in the roast.  I had been testing my SR500 with a K type thermocouple in various spots.  Most of the time people run their roasters with the thermocouple shoved straight down the middle in a lot of the photos I’ve seen.  With the SR500 I’ve found this region to be very sparse with beans due to the high flow of air up the middle.

As shown in this diagram the center portion at the bottom shoots up the middle in the air column grabbing beans on the edges of the two green zones.  They are then tossed outward in the orange circular zone represented on top.  The beans that sit out on the purple areas on the bottom tend to burn due to low circulation.  Ultimately I would expect some sort of triangular slanted chute sealed to the edge would allow beans to be directed away from this burn zone and closer to an area where movement would be possible will help improve this.

I have not written much lately because I’m currently not at home using my roaster.  Prior to leaving I did some testing of the sweet spot ultimately getting pretty close to the middle of that sweet spot.  No matter which beans I roasted and the length the beans were being roasted I always noticed, while adjusting fan and temperature settings on the controls, that the beans always had first crack at the expected temperature +/- 1 degree Fahrenheit.    The same with second crack seemed to hold true as well.

Based on those readings and the reproducibility of that it is my opinion that the ideal thermocouple position in the SR500 and probably the SR300 is as I described earlier.  The height seems to be just below the surface of where 120 grams of beans fills the roasting chamber to for the tip.  This keeps it from the bottom where the heat is somewhat higher and allows it to be centered in contact with the main mass of beans once they expand slightly and begin to float more easily in the air column above his position.

Regarding the coffee’s I’ve roasted.  I previously mentioned some by name and certain characteristics of the roast as I was roasting them.  I had decided until I can track the actual temperatures and accurate times to share it was probably best not to continue posting the resulting curves as I roasted them.

Experimenting with a thermocouple.

By | | , , ,
Leave a comment

So now that I have a faster responding thermocouple I’m finding that the  sensor works better closer to the edges of the roaster.   When thinking of the bean mass on the SR300 and SR500 you probably need to consider the bean mass as if it is a donut at the bottom of the chamber.  You want to get the tip of the probe centered in the middle of the donut and not the middle of the chamber.

Most modifications people have shown put the probe drilled straight down the middle.  This has the advantage of clearing the “chaff grate” on the sides down a natural hole in the roaster.  The problem with this is that the very center is actually a sort of “fountain” of beans and hot air.   The probe has very little contact with any of the roasted beans and is mostly picking up the air put out by the roaster.

Many people who have sensored this middle area report temperatures in the 420 to 425 degree range.  At the very end all of the beans are very fluid and filling almost the entire chamber area moving freely.  Once this finally happens the temperature readings then reach the 440-450 range showing a typical second crack area that coincides with actual second crack occuring.  Often second crack ACTUALLY happens while the probe is showing 400 degrees when placed in the middle.  This is likely due to the bean mass thermic reaction being hotter than the air rushing up from below and only a few minutes later does the air column catch up.

I’m starting to get low on beans from the samplers  with only a pound or two left of all the varieties.  I’m feeling that the probe placement I used initially in the middle helped stall a few roasts due to inaccurate temperatures of the bean mass that I responded to.  I’m testing a roast where the probe was moved closer to the “donut” range to see what sort of tastes develop but this one is pretty close to a standard roast.

My feelings are there is a “donut” sweet spot.  In the diagram below there is a Orange region [1] that is the normal plume of bean being pulled from the middle area on the bottom and [2] tossed up on top to the left and right.  As you descend into that area you enter part of the sweet spot [Green].  In the far left and right corners [Purple] you have a “burn” zone due to very low circulation of the beans.  This is caused by the more central area of the donut [3] falling towards the middle and then being jetted to the top [1] repeating the cycle.  Only when the beans have sufficiently roasted does the bean mass begin to rise above the initial [Yellow] fill line.  As the mass makes it above the silver metalic band there is enough movement to cause beans to leave the “Burn Zone” and make it easily into the column.  This initial burn area is where the stray first cracks are likely to originate and the uneven roast occurs.

Freshroast "Sweet Spot" created in roasting chamber.

I would thus interpret the space where the D on the left donut and the t on the right donut would be the approximate ideal locations to sample bean mass temperatures due to the piling on of beans from the fountain [1] above it.

Freshroast SR500 apparent “safety” mechanism. Adding new thermocouple.

By | | , , , ,
Leave a comment

Due to the microprocessor installed and the sensor mounted at the top of the roaster base where you install the roasting chamber.

NTC Thermistor mounting

It appears that when you turn the roaster on to High and lower the fan knob to the lowest setting the roaster will reach (unloaded) 430 degrees.  At that point the fan automatically revs up slightly (possibly due to the heater being off)  and the heat stops climbing at 433 degrees.  Occasionally it drops sooner but always between 430 and 433.  It then drops from 430ish degrees down to approximately 420 degrees and then resumes climbing again.  The heat appears to always stop at 420 and then goes back up.  The highest it stops dropping at is 422.

Temperatures were measured with a new K type thermocouple mounted approximately where the center of the bean mass would be.  I obtained this at OSH (Orchard Supply Hardware) since I needed some sort of multimeter for the moment.  They had one inexpensive “Craftsman” model with the ranges I need for measurements with the Digital Multimeter side with an included K type thermocouple mode.  I’m expecting the thermocouple to not last very long due to the temperature rating they put on it.  I will likely get one from eBay or something that is compatible with the K thermocouple sensors in the meter.  I’m guessing the protective covering they put on the tip end insulation area to keep the insulation from unraveling is not able to withstand the high temperature for very long but the insulation looks like it can take a lot more.   I’m going to probably get some high temperature silicone and put a little dab right at the end above the tip to hold the insulation better so it won’t fray.

Ultimately I’ll source out a thermocouple that can do more properly as I get further into this.   I’m going to switch to using the previous thermometer as a timer (was a secondary feature… you get one or the other on screen at any point in time) and the thermocouple as the new temperature sensor since it actually responds faster on temperature changes.

Modified Thermometer Test w/ Rwanda Gkongoro Nyarusiza

By | | , , , , , ,
Leave a comment

So as mentioned previously I took the thermometer probe and used one of those pipe cutters with the “wheel” for cutting things like copper pipe and sliced off part of the probe.  Due to the fact that this sort of cutter “crushes” slightly while it cuts this ended up making it so I could not slide the inner electronics out without fear of not getting it back in or damaging it (without having a soldering iron currently).  I instead slid all of the formerly connected probe upwards and taped it to the probe’s plug using electrical tape.  I then crimped the remaining probe in place at the top against the braided cover.

My initial test seems to have yielded a technically proper heat ramp based on what was happening.  Additionally the cooling cycle appears to have been pretty normal too.  It appears that when the cool cycle halts the beans are approximately 158 degrees and linger there while in the roasting chamber even when removed from the roaster.  They drop to around 130 degrees after being dumped out into another container within about 20 seconds and then they hang around 130-120 degrees for several minutes if undisturbed.

Here is my heat readings for a standard roast.  In this case I was roasting the Rwanda Gkongoro Nyarusiza.

Markers are A= Rolling First Crack, B= Second Crack.  Fan speed is shown as 100% for the knob being set all the way to the right.  50% is fan knob set in the middle position (straight up).

I was not able to keep track of any additional details due to manually tracking the roast on paper.  Cooling with fan speed increased to 100% starting at 6.1 was as follows:

Bean mass started at 120g and reduced to 102g for a loss of 18g or 15%

Problems with thermometer – YAY!

By | | , , ,
Leave a comment

So I finally got a chance to try a roast with the thermometer.  At approximately 3.5 minutes into the roast in the reading it crossed 482 and would no longer read.  This is OBVIOUSLY not a correct reading.  I expected some imperfection in the readings due to this probe expecting liquid/solid meat as opposed to air/dry readings but everything is significantly off.  First crack came as a reading of 449 which puts me a good 50 degrees above where I expect to be and follows about 50 degrees over everything I expected during the roast for the color changes etc.

My theory is that the probe is so long that it has to be inserted through the grate on the chaff collector on top.  The probe is then picking up heat transferred by the chaff collector that it is touching and particularly the metallic disc embedded in the lid being transferred through the plastic.

The cooling cycle lasts approximately 2.5(ish) minutes and it ends at an unusually high number.  With the probe and chaff covered removed and sitting 5+ minutes it still shows a temperature in the 90s so this is certainly picking up the temperature from the cover.  At this point I will need to attempt to remove the probe in some way and see what else I can do to “shrink” the size of the probe and find an alternate place to put it.

[easychart height=”300″ type=”line” title=”Heat Results” groupnames=”Temperature F” valuenames=”0.0,0.5,1.0,1.5,2.0,2.5,3.0,3.5,4.0,4.5,5.0,5.5,6.0″ group1values=”78,222,366,402,413,428,460,482,482,482,482,482″ group1markers=”6.1″ hidechartdata=”true” ]

[easychart height=”300″ type=”line” title=”Cool Results” groupnames=”Temperature F” valuenames=”0.0,0.1,0.2,0.3,0.4,0.5,0.6,0.7,0.8,0.9,1.0,1.1,1.2,1.3,1.4,1.5,1.6,1.7,1.8,1.9,2.0,2.1,2.2,2.3,2.4,2.5″ group1values=”482,482,474,402,320,262,240,212,201,183,172,167,158,152,147,141,140,138,138,132,129,127,127,127″  hidechartdata=”true” ]

While the HEATING is flawed with the metal disc and the cooling is most certainly flawed at the end as well it is interesting to witness the significant drop in heat from the cooling cycle.  The beans (by touch) clearly release heat faster than the metal mass does and they are quite cool in the end.  In about 30 seconds the heat has dropped  to around 250 with the metal heat skewing the readings.  It is my guess that all cooling readings are approximately 30-40 degrees higher than actual in the bean mass meaning the real “roasting” has stopped before 30 seconds of cooling are up.

Lesson Learned: The chaff collector gets significantly hot enough that you need to avoid ANY contact to any probe devices with the top and try to steer clear of the metal disc at all costs.

Update: I have used a copper pipe cutter to slowly slice a ring around the metal casing of the thermometer allowing me to shorten the probe.    I googled around and found this article http://brettbeauregard.com/blog/2009/09/anatomy-of-a-digital-thermometer-probe/ which shows the inner guts of a typical probe.  As you will see this is mostly the braided wire until the very end where the actual sensor is.  I’ve sliced off several inches of the casing and slid it up the wire towards the socket.  Once I determine if this seems to work I will try to figure out a way to slide this off without mucking things up too much and permanently removing it and crimping the end of the probe onto the braid.  New article once I get to that stage if it works out.