Does anyone know about this newcomer? Looks interesting and fair/good value for the money? Any input is welcome, was planning on a ikawa pro but the price of this make me reconsider

https://dailycoffeenews.com/2023/03/06/ ... to-the-us/

Howard Bryman wrote:Built upon a foundation of hardware and technology laid by the recently launched Kaffelogic Nano 7 home roaster, the Link adds a variety of features designed to appeal to green coffee buyers wanting to roast samples while traveling. The Link maintains a capacity of 50-100 grams.

"So much weight is placed on the success of the roast in the assessment and purchasing model," Sam Corra of Nucleus Coffee Tools recently told DCN. "It's a pivotal point in the process that can make or break a deal as it dictates how the buyer perceives the quality of the sample. Too often it is the quality of the roast that causes so many great coffees to be left un-purchased on cupping tables."

They had a few of these at MICE last year and I spoke to Sam about it. The article basically summarises it. Sam played around with a kaffelogic, took a bunch of notes, made a tonne of kaffelogic profiles and came up with basically a massive table where you can measure the bulk density of the green, pick the processing method and grab a corresponding profile that Sam thinks is likely to be a pretty good educated guess. I mean I assume there are probably also different profiles for different weights eg. 50g v 100g. Then you just roast it fully automatically; you don't even decide when to stop the roast; you just rely on the kaffelogic profile to do it for you. Sam has also had the kaffelogic at some of the world brewers' cup championships and used it to roast coffee that the competitors have used. At MICE, they plugged through roasting samples of the top 10 lots in the 3 categories for best of panama and apparently that was all 100% lookup table, corresponding automated profiles.

The "pitch" for the link is basically that it makes it easy for coffee professionals to roast their samples. Which is actually a huge problem. I've been helping a guy who splits coffee for home roasters to evaluate importer samples for a while (I don't make money from it, but I get to steer him towards green that I want to buy so I can buy tiny quantities of it) and it's a bit of a nightmare. If you get roasted samples from importers, most of them are not well roasted. We've had heaps of too dark or baked samples where we have concluded that the green might well be great, but we can't be sure, so it doesn't get bought. Because importers have like 100 offerings and they're roasting like a 50g sample for you in an ikawa and just taking a stab themselves. Or they send you maybe 100g green if you're lucky and you've got to work out some way to roast it. At the moment I'm bumbling around trying to make some inlet temp profiles on ROEST and ignoring BT reading, since I don't think it's reliable on any roaster at 50g doses. But at least if you get a 100g green sample, you can have two shots at it. So if the link system works, it'd be great for green buyers.

The elephant in the room of course is how this relates to kaffelogic, since you could buy the kaffelogic now, and for less money. And it seems to lead to this bizarre position where kaffelogic are sort of marketing their cheaper machine as the thing with programmability and flexibility. But they are marketing more towards home roasters, who you'd kind of think probably don't have much skill and experience. And then link are marketing their thing to industry ... but based on ease of use and automation ... but you'd think that of home users and industry people, industry people ought to have more skill and experience to want the programmability. So I'd kind of think that the link kind of intuitively feels like it should be marketed to consumers more. I suppose it's worth observing here that the roast profiles are likely to be for cupping and filter, so I suppose home roasters wanting to have some set and forget profiles for espresso might be disappointed, but, who knows, maybe they do have some espresso profiles.

Sam mentioned an idea for a hardware thing that they might add to the link, but I don't know if that ever materialised. I mean, I'd expect that they would be pointing it out if it had. Anyway, I'm going to assume that that's something I'm not supposed to be talking about. I guess if the board on the link stores the 41 profiles, it's probably custom, since I don't think the stock KL can store that many profiles on it.

So I guess the question is "how good are the roasts"? And I suppose the answer to that is "relative to what?" I've had lots of roasts that suck from kaffelogic/link, ikawa and roest and I've had good roasts from all of them. How easily or repeatedly you can deliver roasts on any of these relative to the other, I make no comment about.

Compared with Ikawa, the KL/link has an alleged bean temperature probe, but doesn't have an inlet temp probe. I have no idea how the KL/link might perform with 50g samples. Compared with Ikawa home, the link isn't trying to position you to buy a curated green selection service that you might not want, or a subscription model to unlock more flexibility to program profiles (I'm assuming the link retains the normal KL studio software). The Ikawas obviously have a whole cyclone setup that the KL/link lack, and the old chaff collection assembly from the KL was pretty annoying, but hopefully the new assembly is a bit better. Probably won't be as convenient as emptying an ikawa chaff jar.

Jonk wrote: [*]A glass jar for measuring density..

FWIW, I thought the "density tube" was metal. I mean it's built for travel, so you'd hope so. You'd also hope the machine would be dual voltage, but no mention of that. I wonder if they got a 110V unit to Australia, ran it off a transformer and checked/tailored the profiles.

After using the Ikawa quite a bit, it is still troublesome to try to nail a roast on the first go of a new sample. I don't know too much about the Link or KL but Luca mentions that they do not have an inlet temp probe. I've found myself using inlet temp profiles the majority of the time and then stopping the roast when I get the development I'm shooting for. That is the only way to repeatedly get a nice steadily declining RoR. Having the software adjust the heat based on the exhaust temperature can introduce strange behavior and has never been as smooth as a steady heat input. Since it doesn't seem that the Link or KL have direct inlet temp profile control then I question whether they can hit "textbook" profiles as nicely as the Ikawa. I can see the Roest being intrinsically more thermally stable since it uses a drum and is a larger machine.

An example for those to see the difference in the control schemes:

Exhaust control (feedback loop) showing an example of a wonky roast. The feedback software got into some weird rhythm. Not fun behavior to have during a critical sample roast. This was using one of Royal Coffee's profiles, so presumably a dialed-in profile. Please note this is an extreme example, but none approach the smooth behavior on the inlet profile.



Inlet profile (follows the heat profile, no feedback) showing how it smoothly follows the temperature profile. It is up to the coffee to react as it would with a prescribed inlet temperature. Different coffees behave and heat differently but none are exposed to wild swings.

Milligan wrote:After using the Ikawa quite a bit, it is still troublesome to try to nail a roast on the first go of a new sample. I don't know too much about the Link or KL but Luca mentions that they do not have an inlet temp probe. I've found myself using inlet temp profiles the majority of the time and then stopping the roast when I get the development I'm shooting for. That is the only way to repeatedly get a nice steadily declining RoR. Having the software adjust the heat based on the exhaust temperature can introduce strange behavior and has never been as smooth as a steady heat input. Since it doesn't seem that the Link or KL have direct inlet temp profile control then I question whether they can hit "textbook" profiles as nicely as the Ikawa. I can see the Roest being intrinsically more thermally stable since it uses a drum and is a larger machine.

Mmmm ... this seems like some pretty selective reasoning to me, and, if anything, the correct reasoning is exactly the opposite! If you are talking about a "textbook" profile, then the relevant RoR (not a dig at anyone, but I really hate that expression; why on earth we can't just call it the derivative, like anyone does in maths I don't know; generally I hate there being new words made up where we have a perfectly good existing word) is the bean temperature ROR. It looks like the RORs that you are talking about are the derivative of the inlet temperature. The Ikawa does not have a BT probe. The Kaffelogic's only probe is allegedly a bean temperature probe, at least at 100g + doses, in part of the bean mass. You can go into the studio software and program a "BT" profile that has a perfect "textbook" BT ROR. So, if anything, it's the exact opposite and the KL is the machine of the two that is best for delivering a "textbook" BT ROR profile.

I don't have a particular opinion on either of these roasters or what is best. My personal view is that I'm deeply sceptical of the value of BT probes in these systems. I doubt anything can get a good BT reading at 50g doses. I question whether the high airflow renders the readings inaccurate. In the KL, the temperature at the bottom of the chamber is much higher than it is at the top of the chamber, yet if you put a probe anywhere in it, people would call it a "BT" probe, since it is in the bean mass.

WRT the roest, they have programmed a between batch protocol, but not yet a preheat protocol. Some of the models have a stationary drum temp probe. I find that starting with a heated roaster at a constant drum temp seems to lead to fairly repeatable roasts, at least in the sense that the first roast doesn't seem to end up slower than the subsequent roasts. I'm working on developing a 50g profile for it, based on the new inlet temp sensor that Scott and I asked them to make and that they gave me for free (disclosure). One thing that I kind of wonder is if the stationary drum temp sensor might function as an incredibly delayed BT sensor in the roest, or if the 50g bean mass isn't really sufficient to affect it. I'm sure that there's some clever maths that could be done comparing the results for a 50g load, a 200g load and an empty load. Certainly the drum temp shows a drop when 50g are loaded, so it's at least somewhat affected by bean load. Maybe drum temp ROR could be used to give some insight as to 50g roasts? Who knows!

Jonk wrote:Thanks for the great write-up Luca (and correction ).

No worries. Thanks for the info on the 50g profiles. I wonder if you'd have better luck with a power profile. Only problem is the KL is so ridiculously limited in how you can power profile, plus power profiles are just the added power and can't account for ambient air temperature. This is one of the reasons why I wanted an inlet temp sensor instead of using power profiles - the PID controls the temperature at the inlet, after the heater and before the roast chamber, so the PID accounts for hotter or colder ambient air. This might not be a big deal in a temperature controlled green coffee importer's office, but I don't know that that's always the use case.

luca wrote:Mmmm ... this seems like some pretty selective reasoning to me, and, if anything, the correct reasoning is exactly the opposite! If you are talking about a "textbook" profile, then the relevant RoR (not a dig at anyone, but I really hate that expression; why on earth we can't just call it the derivative, like anyone does in maths I don't know; generally I hate there being new words made up where we have a perfectly good existing word) is the bean temperature ROR. It looks like the RORs that you are talking about are the derivative of the inlet temperature. The Ikawa does not have a BT probe. The Kaffelogic's only probe is allegedly a bean temperature probe, at least at 100g + doses, in part of the bean mass. You can go into the studio software and program a "BT" profile that has a perfect "textbook" BT ROR. So, if anything, it's the exact opposite and the KL is the machine of the two that is best for delivering a "textbook" BT ROR profile.

The RoR is measured from the exhaust temp on the Ikawa. The exhaust temp is closely related to the BT (and confirmed from tests done here with a probe down the chute), but not a pure BT measurement (if there is even such a thing.) So it is not a measurement derived from the input, but of the output.

I don't have a particular opinion on either of these roasters or what is best. My personal view is that I'm deeply sceptical of the value of BT probes in these systems. I doubt anything can get a good BT reading at 50g doses. I question whether the high airflow renders the readings inaccurate. In the KL, the temperature at the bottom of the chamber is much higher than it is at the top of the chamber, yet if you put a probe anywhere in it, people would call it a "BT" probe, since it is in the bean mass.

Which is why Ikawa chose exhaust temp which closely follows BT instead of trying to read from the bean mass. Ikawa gets bottom chamber temp and top chamber temp. Bottom is input, top is output. Heat transfer is highly efficient so the BT is very close to the exhaust temperature. There was a lot of discussion about this in the Ikawa thread awhile ago.

WRT the roest, they have programmed a between batch protocol, but not yet a preheat protocol. Some of the models have a stationary drum temp probe. I find that starting with a heated roaster at a constant drum temp seems to lead to fairly repeatable roasts, at least in the sense that the first roast doesn't seem to end up slower than the subsequent roasts. I'm working on developing a 50g profile for it, based on the new inlet temp sensor that Scott and I asked them to make and that they gave me for free (disclosure). One thing that I kind of wonder is if the stationary drum temp sensor might function as an incredibly delayed BT sensor in the roest, or if the 50g bean mass isn't really sufficient to affect it. I'm sure that there's some clever maths that could be done comparing the results for a 50g load, a 200g load and an empty load. Certainly the drum temp shows a drop when 50g are loaded, so it's at least somewhat affected by bean load. Maybe drum temp ROR could be used to give some insight as to 50g roasts? Who knows!

That is one thing I haven't had to deal with too much on the Ikawa. The exhaust temp profiles (which use a feedback loop) preheat to a prescribed level before alerting the user to charge. When using an inlet temp profile then there is no such alert as far as I'm away. I usually charge at the same exhaust temp as my BBP.

Jonk wrote: I've never had issues anyway near as bad as in Milligan's first picture, but I agree that inlet temperature sounds like a better way to control a roast. Maybe not for samples, but at least if you're doing repeated runs to improve the profile. On a KL you never really know how the PID will react.

I cherry picked that wonky result to illustrate the issue with feedback loops on the Ikawa and likely in general on small air roasters. I don't find myself sweeting over what to do with a 50g sample often. Most importers give 250g+ samples so I like the inlet profile because I'll have a few goes at it to get it just right. I usually don't have much fuss out of Ikawa's own profiles and some gentle modifications of those. I've heard claims that steady RoR isn't as important on air roasters, but I seem to get the better tasting cups out of a roast that doesn't have an RoR that looks like a roller coaster.