gild that lily
1 Like
I was asking google AI some further questions and caught it in a contradiction. Now it looks like I just need to use SS lock washers and flat washers and not titanium ones.
1 Like
I might have solved all these lock washer, nut problems by just making the shaft part of the whole design. I made the shaft longer at one end which is like my current setup so I can get it into the sifter. I think my current setup I initially put all the hardware on there before mounting it to the motor bearing. I will 3d print it first to make sure it looks good. I might have to sand the ends down when I receive it as it might be a little over 6mm in diameter from the print process. I shrunk the hub down from 14mm to 10mm. Total weight should be about 4.4oz for this new part versus about 7oz for previous SS disc alone.
Personally, I’d be a little worried about bending the shaft and then scrapping the entire expensive part.
Is the weight of the SS piece a problem? I’d have thought that’d be relatively unimportant for an evenly balanced rotating assembly.
I am probably getting carried away with it and gilding the lily like Tuco said. It was definitely too much weight for a M5 shaft, but the M6 does seem to be holding up. It is probably around $175 for the cost after shipping and tariffs. It is an interesting idea to consider though. So will give it some more thought. I adjusted the design a little more for smoother transition. I could probably go to 8mm shaft with the cost going up to close to $200. That is a lot of money to consider.
My wife thanks you for bringing me back to reality. She says she can now get a couple extra massages. She should not have told me that because now I am still considering it, but only if the price comes down some. I did 3d print this design in plastic with a M8 size shaft and it feels pretty good and goes into the flour sifter easily. I did make it a fatter transition at the disc for better structure.
I printed a test stand for my 7 oz. SS disc to see if the M6 rod has bent any from the weight since the M5 did. I do not see any noticeable bend in it, so going fromM5 to M6 was a good idea. The only problem I see with the 3D printed disc is it stains from the coffee roasting, and it would not come off after soaking it in boiling water with the beer brewing cleaner I have always used for SS. It is probably not really a problem though. A SS sheet might not have this problem though as that is nice and shiny flat surface whereas the 3D printed disc is sand blasted finish.
I guess the only a couple of pros/cons of using the titanium would be:
Pros : does not retain the heat like the SS and a lot less weight
Cons: Higher cost and shaft possibly bending like @jono035 mentioned. Of course I could not make the shaft part of the disc like I did with 3D printed SS disc
Here is what the staining looks like.
2 Likes
Hahaha, great news? I guess you could also get the shaft machine as a custom piece if you wanted via PCBWay or similar, too.
I think the staining is probably just oils that are polymerizing onto it, it’s basically getting seasoned… I’ve seen people refer to this as being desirable in a drum roaster. My Kaffelogic has a similar thing going on, the inside of the Behmor is a nightmare and only gets cleaned every 6 months or more, etc.
I wouldn’t be too worried there.
I ran 4 - 4oz roasts using my new USB connection on the portable power supply. One less connection I have to connect to start a roast. If the display is correct, it is only drawing 0.05amps for the USB connection and it is running about 0.4 amps for the 12v and 5v. My 4 Ah Ryobi battery showed 2 bars out of 4 when I started and still had 2 bars after probably 50 minutes of usage. This USB connection was worth the time and little extra money. Now I am thinking of designing a shield to place between the nema17 and the electronics to keep any chaff from getting on the electronics. There is not much heat that far away, so plastic should be good.
I also decided against 3d printing that complete disc assembly in titanium but did buy a new coffee grinder that has titanium coated flat burrs since the price dropped this week on Amazon. I have only done 2 grinds so far. It is quieter than my previous cheap grinder and a lot less messy.
I have noticed some chaff getting into the electronics and the bottom of the heat gun, so I decided to add some chaff shielding for the heat gun and the electronics. The shield around the heat gun will also act as a wind shield and might make the roast hotter quicker since the heat will be more contained. I am not sure how hot that plastic will get around the heat gun until I try it. I can slide the U-shaped shield out easily so I can easily remove the heat gun and put the cooling fan there during cooling process. A Chaff collector at the top would be better if I could come up with a way for that.
My brother-in-law recently sent me some Black Rifle Silencer Smooth coffee that was labeled light roast which I thought was odd by the color of the beans. I thought they might relate the light roast based on weight loss, but the answer seems to be completely different than I expected. Here is what Google Gemini said about weight loss, color and why it was labeled light. This coffee was pretty good.
1 Like
I came across this recently and thought it might be a good alternative for the Arduino UNO R3 on my roaster but turns out probably would not especially since I am using a CNC hat for the 1 motor. PiZZa turns your spare Raspberry Pi Zero W and 2 W into an Arduino
That did get me looking at changing the display on this roaster from the 1602 LCD to a TFT 3.5” non-touch display. My biggest problem with the electronics on my roaster is the lag in updating the LCD display even when just updating a portion of it at a time. When I start the timer, it is a noticeable stutter with the motor as it updates the display every second. It doesn’t really affect the roast but is annoying.
After a chat with ChatGPT, seems like it should work better and give me more options of how to display the status. I am also going to update the Arduino UNO R4 minima while I am at it. I will get rid of my Profile button since after using this for a while, I really only use a couple of different RPM settings. A single array of several RPM settings should be sufficient. I could probably get rid of the CW/CCW motor button also as I never change that. That direction could always be changed in the firmware. The TFT screen I ordered is 480x320 pixels. I will have to move some pins around. I might be able to simplify the case also and put the display where I take the 2 buttons out. Now just to wait for the slow boat from China.
Yeah, weight loss shouldn’t be used to judge anything other than consistency within a single batch. The comment above about initial water content is accurate, as far as I’m aware. I check it for each batch but mostly just to see that nothing went wrong, not as a measure of anything. I’ve even heard of people deliberately re-hydrating beans before roasting, although I’m not entirely sure I follow the logic behind that. Consider 2 cases, one with a very wet starting bean and one with a dryer one. The wet starting bean will require more energy input during the first phase due to water likely having a higher specific heat capacity than dry bean mass as well as the energy being absorbed to flash the water to steam. This will cause it to have a slower rate-of-rise with the same input power or need more input power to achieve the same RoR. A dry starting bean will require less energy so lower power or higher RoR. At the same final temperature, it’s likely the beans weigh the same so the wetter starting point will have lost more weight, relatively, but also may have spent less time at higher temperatures so may actually be less developed, so a ‘lighter’ roast. It wouldn’t surprise me if that also drives a difference in internal vs external temperatures, too.
The other thing is that I believe the colour under the same profile will vary based on the varietal and processing method, so again there’s no ‘absolute’ value that can be placed on anything here. A super light coloured natural process bean under the same profile and amount of development may end up looking quite different to a darker washed process bean, despite technically being roasted to the same level. Again, I think that the fancy colour testers etc. are looking for batch-to-batch consistency rather than any one true absolute finishing point. Checking exterior colour vs interior would also indicate consistency of roast.
There are a few weird things going on that I remember Scott Rao’s book talking about. One is that during the roasting process the interior temperature lags behind the exterior temperature due to slow conduction into the center of the bean. Once first-crack happens, the claim was that this process is exothermic and will pull the internal temperature up above the external until eventually it all evens out. Then again, I’ve seen other comments saying that the first crack process isn’t sufficiently exothermic and that the ‘flick’ seen on temperature sensors isn’t a real temperature change and is some side effect of the steam/CO2 buildup being released as the bean ‘puffs’.
I think surface oils can also be misleading as that depends significantly on the age of the bean and variety/growing conditions etc.
Ultimately, I think I came to the conclusion that trying to use descriptors like that is fine, but it’s a relative scale, like calling something ‘sweet’ or ‘salty’… It all depends on what you’re used to. The comparison between commercial vs specialty kinda illustrates that. Neither of those are ‘wrong’, they’re just scaled differently. I’ve got no personal experience with it, but it looks to me like even the stuff that use to be called ‘light’ in a specialty sense is now being driven up by stuff on the ultra-light scale. I saw a discussion somewhere, perhaps the kaffelogic discord, about classifying ultra-light roasts and whether that inherently needed to be dropped before first-crack to be considered ultra-light. That idea is pretty wild to me, but I guess I should just try it at some point!
At the end of the day, if the flavours you’re tasting match a specific expectation for a development level, it’s probably fine to call it that. Maybe one coffee gets dropped at 215C and ends up bitter with lots of spice/wood/cocoa notes. I’d call that dark, because it’s really the flavour that is important at the end of the day and that’s what we’re describing.
Edit: Also, to be clear, this is just the random waffling of a dude who enjoys reading about this stuff but couldn’t pick up a flavour note in a cupping session to save his life, lol.
1 Like
I use a Pi on mine, and just use my laptop as the display via a webUI. I’m not doing stepper control as of yet however, but it is controlling the SSR, load cells (3) and temp probes (2).
UI has changed since then (dark mode is unreadable outside)
2 Likes
You jumped way ahead of me on your design. Did AI generate that code for you? I never did anything with the load cell I bought. It is still sitting on my desk. I probably do not have enough pins for it now since I am adding the TFT display. I have a Pi 4 using mainsail with that BTT board on the sliding roaster that I have not done anything in a few months with. Klipper does not do well with changing motor speeds on the fly, but on my PBB roaster I have not changed speed during the roast in a while since I found the speed I like. My new TFT display should be here from China in the next few days.
I just bought 14.5lbs of coffee from https://burmancoffee.com/ since they had a couple of origin coffees I had not tried yet, Vietnam and Hawaiian Maui not to be confused with Kona coffee. I roasted the Vietnam coffee yesterday to 15.3% water loss and went to a little higher temperature than I usually go, 265-268f versus 245-250f because it took a while to get thru 1st crack. It smells really good and my wife that does not really drink coffee says she might have to try some. Some of the Maui beans are probably half the size of other beans and they warn about using them in some drum roasters with holes they might go through. That is not a problem for the flour sifter thought. I got a sample pack of 3 different Maui beans.
It was a big Claude Code experiment, yes
The above is the current state of things. The main features I’m actually gaining from is the SSR temperature control and creating/printing labels for the bags, and some history tracking.
The load cells are as flaky as someone above predicted, but it seemed cheap and easy enough to play with, so I did it, no regrets, but I wouldn’t do load cells again.
I have temp probes but am too timid to drill holes in my flour sifter screen, mostly since I fear the wobble disk will get stuck on the bean temp probe (where the probe is actually in the beans, I’d be cutting a slot in the disk). Right now it rarely gets stuck, pretty reliable. If I got the probes actually measuring bean temp I could integrate it with Artisan (which mostly works, but it wants the temp data)
You may notice a TMC2209 dangling there too, ready for maybe replacing that geared motor with a stepper, but am not there yet.
It’s mostly working, so I don’t want to mess with it too much!
1 Like
I’m glad you tried the load cells but sorry it did not work out as well as you thought. From my recent experiences with different roasts that I was given, seems like the medium and dark is mostly dictated by the color, temperature and whether you get to 2nd crack. Looks like you have the load cells mounted well. Does the particle board absorb a lot of the vibration. Seems like putting cork of some sort of foam might help absorb more of the vibration but someone on this forum might give you a better idea for that.
I tried putting a probe in the side of the roaster about 1.5” down from the top but did not give me as good of readings as the probe from the top just above the height of the wobble disc swing. I covered the abandoned side hole with a SS M5 screw, lock washer and nut. It would be a more accurate temperature where the beans are but not sure I would attempt to drill a hole in that screen either.
Since your heat gun seems permanent in there with the SSR, do you dump your finished roast into something else for cooling or do you just turn off the heat to the heat gun and let its fan cool them?
My new TFT board came in today and I had the modified case already designed. After a couple of test prints, looks like I can fit it into my design ok. I just have to do an Arduino test script to make sure it works, then get ChatGPT to rewrite my code. This TFT board also has an SD card slot and I might have enough pins to use it if I get rid of the LED which with the display is not really necessary. I could save roast temperature data to that.
Looks like I need to use a logic level shifter for the TFT display on the UNO R4 to change the logic from 5v to 3.3v on the display. The TXS0108E seems like not much money for it and should fit in my case easily. I found this writeup on the wiring diagram for it.
That should work. It pays to be careful with those bidirectional level shifters, they can be a bit finnicky about the load on either side.
As long as you’re driving from a low enough impedance output and driving to a high enough impedance input then they’re great. They really don’t like capacitance on their outputs due to the way they ‘detect’ the direction they’re operating in so I wouldn’t use one to drive long lengths of wire, for instance.
In situations where the direction I’m level shifting is fixed I’d usually prefer to use unidirectional shifters or a bidirectional one with a separate control line just to avoid any potential issues, but these ones work pretty well for prototypes and simple logic-to-logic scenarios.
Yeah, that was me. I was pretty sure they’d be a challenge to get working in that situation but had hoped I’d be wrong. From tracking input and output weights on all my roasts with the Kaffelogic, I haven’t seen much to convince me that weight loss itself would be a useful feedback value during the roast, at least. I would never interrupt a roast because it had gotten to a target weight, for instance, given that most of the actual reactions we’re looking for are temperature/time driven.
What weight loss are you actually seeing? I’m at around 20% with my decaf espresso blend roasting to a medium-dark (220C drop temp).