[anonymized] wrote:But replacing the PID in the small roaster with a more advanced one (by Fuji) was successful and worked very precise after I let the PID to an auto-tune.

Which roaster do you have? Was it difficult/costly to add Fuji? Is there a link with descriptions/instructions? Thank you

[anonymized] wrote:I also added an extra PID to the bigger roaster and it allows me, using the 'alarms' functionality in the Artisan software, to repeat roasts very precisely.

Sorry for an amateur question, but my understanding is that you set up all the alarms before the roast, and yes, it adds consistency. But I do not understand how PID can "cooperate" with the PID control. Can you add alarms on the fly?

Maybe it is a naive question, but I don't know much about the subject..

Thank you!

Here is a good thread from someone that built a business on the Artisan 3E and then made the switch to drum. He gives some great insights about the differences. Long story short, his beans tasted the same after dialing in the drum roaster.

My "new" 5kg Turkish roaster has arrived

Why you see so many people say they roast manually is because there isn't a lot of choice in automation in the price range and capacity you are looking at. Most of it is hacking together a PID and testing for days. I own an Ikawa Pro100 and it is good at following a profile but I've learned a very valuable lesson. The beans will still guide the roast. It doesn't matter what you make the curve, it is nearly impossible to completely force a profile on the beans. It is still somewhat of a trial and error to dial it in. I've got Ikawa profiles that look amazing on paper but in practice I get a flick and crash, so I have to adjust. I find that "automation" is just a play-back button of a roast that worked successfully. Lots of tinkering still take place.

There is a ton of choice if you look at automation as a playback button, but you have to pay. Mill City has their 2kg machine with full automation. Diedrich has the DR3 which is fully automated. US Roast Corp offers full automation on their 1lb machine. There is the Bell Weather.

Honestly, the big take away from Almico's thread is that the fast changes from fluid bed aren't very useful. Coffee doesn't like quick changes in temp. It likes a steady environment with fluid progression of roast profile so that the chemical reactions can properly take place. You get that innately with traditional roasting and have to somewhat fight fluid bed to get that same smoothness.

The benefit of fluid bed is faster roast times since it is 100% convection. However, newer drum roasters are around 90% convection and at the high end becoming more of a hybrid setup to get the benefits of both (US Roast Corp Revelation, Loring.)

I'd seriously look at the Aillio Bullet for your needs and price range. I'd forget about the "allure" of quick temp changes.

Lastly, manual roasting doesn't take a lot of input changes once you get your "recipe." You may only interact with the machine and make 3-4 adjustments the entire time. I'd venture to say that most people that get deep enough into roasting to spend $1000's, spend weeks learning, and dedicate time to the craft enjoy the few minutes a week that they get in front of the machine. Now production roasting is a different beast.

Milligan wrote: It doesn't matter what you make the curve, it is nearly impossible to completely force a profile on the beans. It is still somewhat of a trial and error to dial it in. I've got Ikawa profiles that look amazing on paper but in practice I get a flick and crash, so I have to adjust. I find that "automation" is just a play-back button of a roast that worked successfully. Lots of tinkering still take place.

But in #2 and #3 I posted videos where the PID was quite successful, right? So, you are saying that no one can guarantee a similar success for any reasonable profile produced by the designer?

My frustration comes from the fact that I did a dozen or more roasts with Hottop during last couple of weeks (220 g loads) and I was never able to get a good RoR curve. So, I am hopeful that an algorithm would do better.
Thank you!

I think your frustrations in the Hottop are very normal and a sign that you have outgrown that machine's capability and want more control/stability to your roasts. That is a good thing. The Hottop and such are consumer products for someone that wants to brown some beans and them not taste bad. Exceptional will be fleeting and hard fought for on that machine. I'm not so sure fluid bed "auto" profiling is the utopia you may believe it is. I enjoy my Ikawa Pro100 but I wouldn't buy it purely to roast great coffee. It is great at what it is meant to be, a sample roaster. To taste greens for green buying choices and cupping. I've much enjoyed it for that but if I was a home hobbyist buying for consumption for myself, family and friends then I'd be disappointed compared to other things out there in the same price point (I know people do this, strictly just my opinion with different goals.) Perfecting a roast (production profiling) on the Ikawa is fiddly, takes time, and just like every other roaster out there, it changes with the specific beans you put in it. You can't just choose a "Ethiopian Natural" standard profile, load up your beans, and expect the best roast for those. It takes hours of tinkering, tasting and wasted batches to get something dialed in with great technical knowledge of the roast process.

Coffee goes through a few "state" changes that greatly affect its ability to absorb energy and that is where you get swings in temperatures regardless of what you set a graph to. Automation is a consistency and labor saving feature, not a replacement for great knowledge in the art and science of roasting.

Milligan wrote:Automation is a consistency and labor saving feature, not a replacement for great knowledge in the art and science of roasting.

Yes, and I would like to get some recommendation on home Roasters that would not be terribly expensive, and would be electric, as I cannot have a gas roaster even in my basement, I guess.

Seeing that the Hottop that you use is over $1000, I'm guessing you'd be okay spending more than that to upgrade. The natural progression from a Hottop without getting into gas would likely be an Aillio Bullet. It has the automation features you are interested in, offers a lot of control and has a larger capacity (1 kilo) than your current machine. The price seems to keep going up on them but I think you can still get one for around $3k. Most fluid bed roasters with automation in that price range will have significantly less capacity (Ikawa 100g and Kafelogic 100-120g.) I think an Artisan 3e would be overkill and require a lot of tinkering to get auto-profiling. I know folks have added PID control to Artisan 3Es but I believe they did that more for production than home roasting. Behmor is coming out with the Jake which is much more advanced than the 2000ab but haven't heard much more about it recently.

I appreciate the kind words on my video! I roasted on a 2kg commercial drum before I switched to Coffee Crafters Artisan roasters. Forget any concerns you have about the coffee having less body or being in ANY way inferior to a drum roast. A lot of that outdated commentary comes from a time when PID control wasn't quite as common as it is now. You can produce world-class coffee from a fluid bed roaster. You can also produce cardboard from it. The same is true of a commercial drum roaster. Which is to say that a well-designed roaster is still only a tool... a tool that is only as good as the person using it. I would recommend joining the Fluid Bed coffee roasters group on Facebook. There, you can ask questions like this to a large community of users who know fluid beds quite well. We have several members who produce tens of thousands of pounds of coffee out of a single Artisan Xe roaster each year.

If you're asking for my admittedly biased advice, I would say the Artisan 3e wins out in a heartbeat. It's a workhorse of a machine that can follow a PID curve very accurately at batch sizes 2.5 lbs and under. (With a full 3lb load, it's pretty much running at 100% heat, so PID can't really do anything.)

To Milligan's point about coffee beans not really liking fast changes in profiles, I would completely agree up to a point. But I still think that fluid bed PLUS PID automation is tough to beat. Fluid bed roasters have the ability to change temps very quickly, so manually controlling them to match a curve is tricky. The heat controls can be sensitive, and changes to loft WILL impact your heat delivery. So there's certainly some learning curve and finesse that has to be learned to get it just right. However, add PID to the mix, and you now have a computer that is making tiny, granular changes to heat 1-2 times per second to keep the temp right on the prescribed curve. In essence, the PID is REMOVING the rapid changes in RoR that manual control inherently introduces (because we're using our slow, clumsy hands and eyes to control something that is much more easily seen at the data level by the computer hardware, e.g. RoR data). Fluid beds also have the advantage for really "crashy" beans. I have a Sumatra that crashes HARD at first crack. I know this, so I sometimes take over the controls to start ramping up heat just prior to 1C... the fluid bed roaster can nearly instantaneously respond with a big blast of heat to help mitigate the crash.

Of course, if you don't need something as big as the 3e, you can check out my latest teardown and review of the little 300g Ali Express fluid bed roaster, which has the potential to be a great little machine for the money!

Finally, I'm actively working on finalizing a more-or-less plug-and-play automation kit for these fluid bed roasters so that more non-technical roasters can reap the advantages of PID combined with fluid bed.

Hope this helps a bit.

-Jason

Here's the link to my full teardown and review of that 300g roaster. It's nearly half an hour long, but hopefully you'll find SOMETHING useful in it.