Nice. I’d definitely keep collecting some weight data and see how it goes, consistency wise. It’s not something I’ve played with with our process and hadn’t seen any commercial roasters that do it, but I’m guessing it might actually be something that’s pretty awkward to do in a multi-kg drum roaster type of situation…
The readings see a couple of seconds between peaks and troughs, but I’d love to see the underlying measurement on a scope to see where the actual spectra of the noise is. I suspect you’d see a lot of smaller, higher frequency vibrations from the gearbox and then a fair bit of heaving at lower frequencies from the bean mass movement. The higher frequency stuff might get damped a little by the cork but the lower frequencies are where it’s hardest to deal with. All very interesting, anyway.
The decaf thing has been interesting. Part of our issue is that we’ve gotten used to using a 22g VST basket which gets dosed to around 20g, so pretty much a triple. That makes the difference between one, two or three coffees in a day a decent sized step in total caffeine.
We’ve not tried roasting any decaf yet. The supplier we use has 2 or 3 different options but we started buying whole beans from the local supermarket that are from a well known NZ roaster. It’s a weird contradiction because they’re a good blend and well roasted but from a supermarket so unknown roast date/age. Whatever the case, my fiance quite likes the flavour so I’m ok with sticking with that for a while but it’s definitely on the to-do list. Interesting that you noticed less chaff, that makes a lot of sense given that the processes seem like they’re just a huge extension of the washing process…
For the moment we’re gonna stick with the 2 grinder setup but I’d like to find a single dose, low retention grinder that has easy adjustment so we can just go back and forth as needed… I’m not sure if that’s a bit of a pipe dream, though. We’ll see. At this point I could also probably be convinced to finish up what we have and move to 100% decaf, given I’ve been enjoying it somewhat equally…
I didn’t think about that until you mentioned it, but I could probably take that planetary gear out of the mix as I am not sure that it is really needed for these small roasts. That does seem like it should take a lot of the vibration out. I might have to try that test again without that gearset. I just have blinders on related to the planetary gear. We have some cold and bad weather settling in tomorrow night, so might be a while before I can do that.
I might be able to do a decent test inside with 4oz of beans in the sifter and no heat gun and do a minute video with and without the gearset. That might give me some good comparison data. I bet that gearset is contributing a lot of the vibration to this. Also, I would run the motor at 30 rpm instead of the 150 rpm with the 5:1 planetary gear. That seems likely to lower the vibration of the motor more.
This would be just for testing purposes and an option to the design.
I could just spend some more money and buy a proper planetary gear built into the nema17. That said, I probably will not spend $40 to find out if it is any better.
I was looking at making another portable PS like my current one for the other roaster I am planning to build but then came across this one that seems simpler and less money. There are several different variations out there for different battery systems. This one is adjustable from 5v to 13v. I really only need 12v for my case.
I redid my wiring some yesterday to clean it up the spaghetti. I routed the 4 wires from the LCD to the Arduino through a 3d printed channel. The seam between the mating parts might be 0.5mm gap, so I put a piece of blue tape over that to keep chaff from getting in. I also moved the motor wire through where the 12v line comes in. That cleans up the outside wiring a lot and keeps the motor wire away from the LCD. I did not change anything for the thermocouple wiring. I did notice the temperature on the thermocouple being off by about 4 degrees Fahrenheit higher than my room thermometer, so I added an adjustment variable to the firmware.
While testing this, the drivetrain was starting to bind occasionally. I was thinking it was the bearing shims and was going to replace them with plastic shims, but then realized it was actually the shaft alignment off. I mostly just eye balled this before. To get this as best i could I created 2 height shims. One for the bearing post and one to go under the motor Hex coupler. It seems to be a lot smoother now. The screen shot shows a shim for the planetary gear coupler and motor coupler, but really only need one for the motor as I have no height adjustment for the gear system.
I believe this SS tube should put my thermocouple in the same spot for every roast. I put a SS M6 lock washer with a SS flange serrated nut on the end to keep it from pulling back into the 14mm OD SS tube. I clamped the wire for it down on top of the motor mount. I chose the 30-degree angle to get it as low as I could. Looks like snow here tonight, so probably will not be roasting for a little while.
Here’s a video showing my current setup. At this point, I hardly have to do anything beyond watching the beans roast. I started with 113.5 g of green beans and ended up with 93 g, which is an 18% weight loss—right in line with a dark roast.
I also received the Load cell but am sidetracked now by working on the Sliding Bean bouncer design. It is really only the electronics I need to wire up now.
During my roasts the motor stutters at each revolution of the planetary gear system. I thought those adjustment shims I used on the motor and bearing would get the shaft alignment correct but seems like that is still the problem. Watching the roast yesterday, I realized i could design an adjustment bracket for the gear. I designed last night and am printing the new bracket and modified planetary gear. It uses 4-M5x30mm screws with locknuts through slotted holes. M3 hardware seemed like it would be too small for that. I should have them printed later today. Here is what the design currently looks like.
Scratch this idea. It is too hard to adjust those 4 screws. I can just adjust the original design easier.
Edit
After watching a roast today, it appears to slow a little each time the wobble disc picks up some beans. The wobble disc is rotating about 30rpm. Maybe slower RPM would help some. The motor I am using is 0.9 step angle and rated at 1.5 A, 0.42 Nm. With the 1.13v Vref I set, I am probably at 75% the rated torque. Guess I could set the Vref higher and see if that buys me anything. I don’t like taking that case apart and might get the TMC2209 hotter, so probably will leave it for now unless someone has a better idea. It is probably not worth messing with anyway, since the roasts are coming well and it just slows a little every half turn.
ChatGPT thinks I am getting ~0.32 Nm at ~0.8 A RMS / 1.13 A peak with this setup.
If you’re having those kind of issues, it might be time to think back to the fundamentals of how that’s set up. Not really my area of expertise but I think you’ve kinda got 2 approaches:
Solid couplers like you’re using. These mean you need all the shafts to be co-linear, so that’s adjusted for to parallel by rotating in 2 axes and coincident by displacement in 2 axes. Advantages are simplicity, strength, no backlash, self-supporting. Disadvantages are the precision needed to avoid issues.
Flexible couplers like a bellow or jaw type. These give you the advantage that you don’t need to be perfect on the alignment, although different types are more suitable for different forms of misalignment, from memory. The downside here is that they shouldn’t be used to support the shafts themselves, so you’d ideally need some form of bushing/bearing on the downstream shafts.
So, my approach would be:
If you want to keep using the solid couplers, I would try to minimize the number of mechanical interfaces in the system to make it as rigid as possible and then give a method for fine adjustment. That might be making the motor mount directly attach to the gearbox or vice versa. If the printer is accurate enough, you could make solid alignment features like pins to locate things. If not, you could build in the ability to re-align stuff. I’d be thinking more along the lines of something repeatable like shims rather than slotted holes. The main issue with this approach isn’t necessarily going to be making it possible to get it aligned but instead figuring out if it’s aligned and, if not, what adjustments are needed.
If you don’t mind re-working stuff then I’d be looking at using something like a bellows coupler between each stage. This would likely mean you need to build in separate bearing support for the input and output shaft. As a first pass, this could probably just be a printed piece acting as a bushing, I suspect that with a bit of grease you’d get quite a few uses as a proof of concept before it got worn out.
If you want to go full weirdness, you could look at using something like a flex shaft to do the same thing but without needing everything to be aligned at all… That probably works best on the high speed/low torque side, although I don’t have a good feel for how much torque is even needed at the shaft of the wobble plate.
Edit: Actually if this were really me doing it, I’d probably use a windscreen wiper motor from a car wrecker hooked up via a bellows coupler and call it a day.
I don’t think it is the alignment after watching the last couple of roasts. It definitely slows down when the wobble disc has more beans on top of it and then settles down again until the next half revolution. From that observation, it does not look like an alignment issue like I first thought and what you are relating to with the flexible couplers.
It is not really much of a slowdown but is slightly noticeable. If you watch part of that last video, you can see what I am referring to. Watch it slightly slow down when the most coffee beans are on top of the disc. I will have to video roasting 6 oz instead of 4 oz next time. I don’t think I want to try 8oz yet though. It should still roast but should be more noticeable.
I still have hope for the sliding bean bouncer with the heat gun sliding back and forth under a rotisserie cage. I should have spent some time today soldering wires, since it got to 74F here.
I don’t see the flour sifter in that mix. I made mine a little less complicated with a purchased nema17 planetary gear stepper from zyltech. I need to check the actual reduction of it. Spec says 5.18:1 but seems closer to 3:1 which is still ok my use. After putting that motor on there, I realized my M5 shaft has become a little bent from the weight of that disc. It still works ok, but you can definitely see the bend in the shaft after I added the real planetary gear. The outer diameter of the disc hub is 14mm, so I do have enough diameter to drill it out for a M6 shaft or 1/4” shaft. I don’t think I have enough wall thickness to drill it out for a M8 shaft.
