Hello to all,

I made a prototype of a popcorn based roasting machine trying to overcome some problems of other DIY models I saw on the internet.

Here are the features of the project:
- fits all popcorn machines and modifications are simple.
- does not require a second power source
- doesn't cost much: $18 (+ $20 for the popcorn machine)
- temperature control by software PID
- tiny screen that displays the chosen program, time and temperature
- optionally, you can follow the time/temperature curve on a computer screen (via usb)
- 8 pre-set (and editable) programs and a fully manual mode (PID assisted)
- easy to use micro-controller if you want to modify the code (raspberry pico)
- (soon) a crack detector will allow a more precise approach of the desired program.
- open-source and (hopefully) collaborative




As you can see on the interface, the eight roasting programs are based on temperature. For now, I have set the code to complete a full-city in 12 minutes.The temperature increases a little faster from 196°. When the temperature is reached, the heater turns off and the fan continues to run for 5 minutes to cool the beans. At any time during a roast cycle, the user can change the roast program, take control of the temperature or stop the cycle.

I am fairly new to the roasting world and before publishing the project, I would have liked to have some feedback, especially on the roasting aspect for which I am fairly inexperienced.

If you are interested, do not hesitate to contact me. It's a very fun project and may introduce some other DIYers to the joys of roasting. It may not be the best model of this kind but it is very open and easily adaptable (programmed in python).

Have a good day,

Cool, you should probably take a look at the Artisan software and figure out how to integrate you popper with that. The adjustments required to get the repeatability and adjustability you probably want are beyond the capabilities of your box at the moment. Most of what you need is already in Artisan including a lot of support from it's user community. Having spent a number of years working on automating a Behmor, www.roasterthing.com, and finally giving up, if I were to try again, it would be using Artisan.

I don't want to diminish what you've done, it looks very nice, but if you're goal is to learn to roast coffee as well as possible, building your own automation and logging tools when something really good and open source is already there might make no sense.

If you want to sell hardware, likely making something to out Phiget Phiget would be the right approach.

Ira

Hello Ira, thank you for your advice and congratulations for your awesome work!

Indeed, compatibility with Artisan is a priority. I'm not trying to reinvent the wheel, on the contrary. For the moment my task has been mainly to take control of the popper from a microcontroller and to make things work with a few pre-recorded programs (based on curves found here and there).

Now that time, air and temperature are under control, I'm going to look at the protocol used by artisan.

Nikos ... cool project.

If you give us some background on you and your goals, people could give some good feedback. Are you just playing around as a hobby ... or are you developing a product?

When you say $38, is that including all the electronics?

Post a video of it in action! I'd love to see it at work.

A big problem with predefined roast profiles is that many beans respond differently to heat. I'm sure the coding is substantially more complex, but a training program would really be slick. The user would select a profile, then load beans and the roaster would see how the bean responds to its best guess. It would then generate a new heat profile that could be saved for that bean specifically. An iterative process would yield better and better curves for a specific bean.

I'm not a software engineer, so I have no idea of the computing power, memory, etc of what I just described, but I've used compact PIDs that do this for heating solvents (a much simpler process).

Number of problems with poppers

1st..... They have a thermal limit switch which usually limits the temperature to something too low to roast coffee with . When you exceed that temperature the switch cuts the electricity off .... So step one is to bypass the temperature limit switch

2bd..now they have enough heat to roast in 4 minutes usually. Or even less. And can scorch beans . So the control it you've got a modulate that heat. The fan is on the same power supplies so when you reduce the heat you're also going to reduce the fan speed and bean agitation. And the being agitation is extremely marginal anyway often requiring stirring by hand up until the being weight is reduced enough to begin agitating by itself.

So separating the power supply for the element and the fan is a first step for gaining more control. So I'm curious how you single power supply.

They also do laughably small amounts. About 80 g Max. So they quickly become a PITA. People quickly pursue larger roasters... It's just not worth fooling much with them after you use them to get your feet on ground in initial learning stage.... But they are great for that initial learning stage for beginners

The problem is if you drink two cups a day, you're roasting everyday. And if you let it rest a day, and many espresso drinkers let it rest a lot longer than that, you're going to do without coffee if you miss a day.

Finding 30 minutes every evening to roast coffee , or at best every other evening, is just a major time consumer for many. Yeah it seems like fun when it's all new to you..... Then it becomes a chore. I personally can't see roasting less than a pound at a time anymore.....lasts me a little over a week.

I'm actually drinking commercial coffee beans from a bag right now because I haven't been able to find time to roast in the last few days while my kids were here......too busy with other things. I started roasting, because there was no way to drive by the 1 decent coffee shop that sold decent fresh beans on the other side of town often enough......... Having to make a 30 minute detour on my way home from work 2 times per week sucked. Getting it mailed all the time just kind of pricey.

Hello and thank you for these answers!I have quite a bit of information to incorporate into this project, it's great!

Rickpatbrown: I am a new coffee lover. I've been transitioning from tobacco (planting, preparation, pipe smoking) to coffee for a few months now. This project allows me to learn about the basic parameters that will affect the tasting. I am not trying to sell a machine but to set up a project that will allow others to learn by doing. I will try to integrate the protocol you mention to train and adjust the basic pre-recorded programs. For now, they are editable text files that associate time and temperature. Something like this:
The pid interprets the program and tries to follow the temperature curve. With an integration in Artisan, it will normally be possible to create a real curve.

(yes, $38 is less or more the amount of money I put in the project)

Mbb: I think I solved all these problems except for the small amount of beans (80g). But for learning purposes, it's pretty handy.

I separated the two circuits (fan and resistors) and bypass most of the popper electronics. The heating resistor is directly powered in 220V (Europe) and the motor is powered in 24V via a step-down voltage and both are regulated by the micro-controller via an SSR interface. The raspberry is also powered from 220v through a step-down voltage (ACDC220-5v). So, in the end, I have a 220 plug that powers all this little world.

Jonk : Yes, the idea is to spend a good day setting up the project and then have something simple to use on a daily basis and simple to adjust. And if you need to go further, you can connect the computer and go digging into the curves.

Here is a rough idea of the set-up:

This is a very cool project - I am frustrated by the 'almost works but, really, truly horrible' software on my FreshRoast SR700 and would love to do it right. I've been toying with a similar idea where the PID is watching an actual 2d curve (X,Y values) and tries to keep up. I can do the curve part - with an ROR curve calculated on the fly and superimposed , but getting the info out to the hardware has been my sticking point so far.