GDM528 wrote: Unclear to me what "closed loop control" is... what does that mean?

No offense taken!

An open loop system doesn't measure the effects of its actions. Unless you have a servo controlled speaker system, or a scary modern one that laughs at your jokes, audio systems are open loop. Signal goes down the chain and sound comes out. The system doesn't check to see WHAT sound came out.

Actually, the Home is a closed loop system for the intake temperature. A PID compares the measured temperature to the loaded profile curve and computes the correct wattage for the next second or millisecond, or however frequently it updates. Without the measurement, it's open loop - send so many watts at such and such a time, consequences be damned.

The only difference is that the Pro measures exhaust temperature, compares *it* to its profile, and sends a new temperature value to the intake loop, which is then compared to the intake sensor, which figures out the wattage. So it's TWO closed loops. The next loop is the human, who tries the coffee and makes adjustments to the profile accordingly. If the human doesn't change the profile no matter how the coffee tasted, that would be an open loop.

Clear as mud...

mtbizzle wrote:I'm trying to understand the functionality of the current (v2) Home with Android. Their website makes it sound like there is no support/app control right now unless you have iOS.

I gather that, currently, with (out of date) Android Ikawa home app + v2 home roaster, you can fully edit profiles? Is that right?
Let's say Android app gets updated, on par with current iOS app. What profiling choices does the Ikawa home offer? (It's still not entirely clear to me). I understand you can choose from a set of parameters (light/med-light/med/med-dark/dark) (dev time) ... Is there a profile library to pull from? Can you download custom profiles?

My new Ikawa Home (V2) works just fine with the "old" Android version of the app - and I ain't never gonna update it! It appears like the Graph Editor offered with the new iOS app is a back-rev to the prior code that allowed editing - this is based on looking at screen-shots peeps have posted.

The Android app does currently allow full profile editing... BUT the latest iOS app can also edit profiles, WITHOUT paying for the Graph Editor. Just download a "legacy" profile (from an older versions of the app), and it will unlock editing (of just that profile) with the free iOS app - yay! You can find legacy profiles here: IKAWA Home - profiles

Ikawa may burn this bridge at some point, but it may sacrifice the goodwill they've extended to the owners of their previous models.

mgrayson wrote:No offense taken!

An open loop system doesn't measure the effects of its actions. Unless you have a servo controlled speaker system, or a scary modern one that laughs at your jokes, audio systems are open loop. Signal goes down the chain and sound comes out. The system doesn't check to see WHAT sound came out.

Actually, the Home is a closed loop system for the intake temperature. A PID compares the measured temperature to the loaded profile curve and computes the correct wattage for the next second or millisecond, or however frequently it updates. Without the measurement, it's open loop - send so many watts at such and such a time, consequences be damned.

The only difference is that the Pro measures exhaust temperature, compares *it* to its profile, and sends a new temperature value to the intake loop, which is then compared to the intake sensor, which figures out the wattage. So it's TWO closed loops. The next loop is the human, who tries the coffee and makes adjustments to the profile accordingly. If the human doesn't change the profile no matter how the coffee tasted, that would be an open loop.

Clear as mud...

"This coffee tastes like dirt... What do expect? It was ground this morning." [Men in Black 3]

Has anyone posted an Ikawa teardown - or will I end up being the first? That should determine exactly where/what/how the temperatures are controlled.

Based just on my performance testing of the Home (100g), the rapid settling time implies a very tight control loop. By "tight", I mean minimal delay time between stimulus (the heating element) and response (the temperature sensor). Longer delays between heat-in and temperature-out can destabilize the control loop, and the usual response is to slow down the control loop accordingly. My thermal measurements indicate that adding beans to the chamber will slow down the response from 10 seconds to about 60 seconds - yikes. Just guessing, but I'm thinking the exhaust temperature is used for a very gentle adjustment to the input temperature settings, lest it scorch the beans by overreacting. If it can remember previous runs, it could use that data to respond faster.

This might also explain some thermal data mathof posted for an older model Ikawa: it had been programmed to slew temperature in 30 seconds, but seems to have overridden the program and slowed the ramp down to 60 seconds.

I suppose by measuring exhaust temperature and comparing to input temperature, you can infer something about how much heat energy the beans absorbed, which would be a significant determinant in the roast level. Did I suppose that right?

GDM528 wrote: I suppose by measuring exhaust temperature and comparing to input temperature, you can infer something about how much heat energy the beans absorbed, which would be a significant determinant in the roast level. Did I suppose that right?

That's the idea, anyway. Actually, given how quickly the air gets from the beans to the sensor, I bet it's very close to a bean mass probe. Come to think of it, the Pro will also take an input temperature curve, so we could compare. I have the Blue Ayarza. Dark+++ may get smoky, but it's for science!

I wonder when someone will make a synthetic coffee bean with internal temperature sensors and bluetooth so that we can REALLY know what's happening during the roast and not just measuring the temperature of the environment, whether embedded in the bean mass or just downwind.

Conceptually if you knew the thermal properties of a coffee bean throughout the roast, you could model an air temp probe in the bean mass to estimate bean temp. Or even just wrap a bit of Kapton tape around the probe till it matches the rate beans gain temperature. Though I'd guess beans gain temperature at a non-linear rate due to drying and cracking so the tape trick might be a huge fail and it will need to be modeled in software.

Ira

mgrayson wrote: I have the Blue Ayarza. Dark+++ may get smoky, but it's for science!

Some (including me) have complained that Ikawa's recipes don't go dark enough. Their Blue Ayarza changed my mind. Ikawa's Dark+++ recipe produced a Full City Plus level roast. After a few days post-roast all the beans were lightly oily - "comfort" espresso. You've been warned!

mgrayson wrote:I wonder when someone will make a synthetic coffee bean with internal temperature sensors and bluetooth so that we can REALLY know what's happening during the roast and not just measuring the temperature of the environment, whether embedded in the bean mass or just downwind.

Alas, it's way too hot in that chamber for any semiconductor technology to operate. An alternative approach would be a thermally reactive passive RF resonator, similar to the anti-theft RF security tags used in department stores. Just need a few million in VC funding and I'm off to the races. My experience says they'd shut me down when they hear how small the market would be.

Brilliant. That's why I'm a Mathematician and not an Engineer.
But I thought coffee was a multi-billion dollar business. All that coffee has to be roasted....

Oh, and the Guatemalan roasted without smoke and looks/smells very good. I wouldn't want it any darker.

ira wrote:Conceptually if you knew the thermal properties of a coffee bean throughout the roast, you could model an air temp probe in the bean mass to estimate bean temp. Or even just wrap a bit of Kapton tape around the probe till it matches the rate beans gain temperature. Though I'd guess beans gain temperature at a non-linear rate due to drying and cracking so the tape trick might be a huge fail and it will need to be modeled in software.

Ira

The thermodynamics of coffee roasting has been studied (by a Brazilian university, of course) and the thermal conductivity and diffusivity is known. So yeah, if you know the temperature just outside the bean you can calculate the temperature inside the bean, and the most appropriate unit for that delay would be seconds, not minutes.

Sorry for the bad pun, but measuring the internal temperature of the bean may not be as important as it's cracked up to be. De riguer for cooking the perfect steak involves soaking in a sous vide bath to a uniform temperature, then maillarding (just made that word up) the exterior very briefly under high heat. So there's two distinct levels the steak is cooked to, to create the desired blend of flavors. That seems analogous to the browning and development phases of coffee roasting. The Ikawa is so efficient at transferring heat into the beans, that I assert you can park them at whatever internal temperature you want in about 60 seconds, then brown the exterior to your liking.

mgrayson wrote: But I thought coffee was a multi-billion dollar business. All that coffee has to be roasted....

Certainly! But not 75g at a time in a fancy popcorn popper.