I recall one of the Mill City YouTube videos talking about measuring density and what that means for roasting but I don’t remember the details … watched it a year ago … but it is still on their channel I think. I think bean density may have to do with moisture in the bean. (I am not an engineer… just a measly former techie 
)
EDUT: I have been thinking about measuring density of my beans but it’s additional work… 
Mill City has one where they talk about the importance of density and how you can pump more heat into a high density bean earlier in the roast: high density beans transfer heat more efficiently.
I did two more roasts last night with the first sticking to the same profile as the last one above, but going back down to 350g and changing to La Morena Cabulco from 2022:
Yellowing started earlier than I like (I shoot for 5 minutes) and first crack was a bit late. I just let the recipe play out with no changes along the way. This is roast R1.15:
The beans still showed scorching or facing, but not as bad as the Costa Rican shown above in my earlier posts. Note the beans here are not consistently charred. (My DSLR battery died, so these are from my iphone)
For the second roast last night, I changed the recipe to use D9 the entire roast:
I also let the roaster cool down to Off and then waited around 15 minutes before starting the preheat to try to recreate the conditions of the roast above. Here is the profile for this second roast (R1.16). Note that I hit yellowing a bit early, but first crack was near the 9 minute target:
The beans still show scorching/facing, but not nearly as bad. These were the worst I could find:
The entire batch looked good, but close up I can see black marks, they just are not as pronounced as the earlier roasts. you can see some of that here:
How do the roasts taste? The first roast in this post has some burnt aroma and a bit of burned flavor. The second does not have that. At less than 12 hours after roasting it is hard to tell how they will taste once properly aged, but so far so good. I also don’t have agtron scores for them yet.
For this older roast (also in an earlier post as R1.10) Roast World - Cup, grade, and analyze your coffee roasts in depth, once I took out the worst of the charred beans it was great. I was happy enough with this by a few days later that I sold a 340 g bag to a customer. The difference between this and the roast I shown below is that this has a preheat temp of 250c.
This next roast (R1.11)has a hint of burnt/roastiness. This was going to hit first crack too late (due to the lower preheat temp of 220c) so I tried to bump power up, which probably scorched the beans: Roast World - Cup, grade, and analyze your coffee roasts in depth
I discuss both of these roasts (R1.10 and R1.11) in earlier posts above.
Do you mark YP based on color or temp range (e.g. 330-335F) or a bit of both? Just wondering. When I roast Ethiopian beans they tend yellow a bit quicker.
I also wonder if the lower D at the beginning is speeding that drying phase up…speculating.
I mark based on temp: 160c. I watch the color, but the shift is so gradual and some beans yellow a bit faster than others even in the same batch. Marking 160c gives me a benchmark across roasts and beans and is usually about when it starts to look yellow. If the color seems way off I will mark outside of 160.
As long as you’re consistent. I tend to mark around 333-335F (167-168C) which will explain why your YP is earlier compared to what’s on my roast graphs.
I use a small stainless steel bowl inside the cooling tray bowl during the roast to catch the chaff falling out the front. I remove it close to the time when I’m getting ready to drop the beans. then I don’t have to clean that chaff out of the beans. Can get quite a bit in there with a natural process coffee.
An update. In short, I have had some success and some improvement, but I’m still not where I’m comfortable using the R1 for production roasting except for one coffee (and I need to scale that up to 1kg if possible). @pnrenton had posted a recipe on the FB group that he had found to work for a number of different beans. This used a 760g batch size and looked like a reasonable profile to try, so I roasted three different beans, back to back. I blind cupped them the next day and since then have been tasting them as V60 brews using Gagne’s method of 22g coffee, 212 F water to 374 g of water. I grind all V60 brews on a 1zpresso k-plus and note that grind setting.
The profile I copied is here: Roast World - Cup, grade, and analyze your coffee roasts in depth
Visually it looks like this:
The first coffee I roasted was Guatemala La Morena Cabulco (from GO). Altitude: 1600-1800 MASL. This coffee as a V60 is fantastic: no bitterness, very round sweetness and chocolate notes. It tastes similar to the roasts I do on my Artisan 3e fluid bed air roaster (where I never had trouble with scorching). The profile:
The second roast is the Costa Rican Tarrazu Honey that I had trouble with up above. Altitude: 1200-1750 MASL (grinding this it feels like most are lower elevation). On cupping this it had a mild grassy, strawlike flavor, indicative of being underdeveloped. I saw a lot of burnt chaff during the roast, but the beans looked fine. However, I’ve now tasted this at two different grind settings and it is clearly burnt. This is the only one I’ve measure color on. Agtron: 56/79, which is a typical delta for this roast level, though I like to have a delta that is lower.
On cupping the third (Mexican Chiapas) I noticed a mild burnt flavor. Altitude: 1750 MASL. The beans looked fine, but as a V60 it is definitely burnt. This one also shows a crash and flick.
So I really need to get close to a 1kg batch size on these, but I hate to waste so much coffee.
For this latest profile I wonder if starting with F3 would help. I would rather have as much convective heat as possible, rather than conductive heat.
Other possibilities:
Soak at a lower power, such as P7 until turning point.
Lower preheat temp.
Slowing the roast on the two burnt coffees to a 6 minute yellowing time and 10 minute first crack is probably best. How to get there on the R1 is, in part, why I bought it. On the PID controlled 3e I just draw a curve in Artisan and the roaster follows.
Have you made sure your roaster is level?
I saw his post on FB and copied his recipe and tried it on some older ethiopian beans that I had on the shelf. My main goal was to try to learn more about recipes. My resulting profile wasn’t as smooth as his.
I am going to try scaling down to 500g and proportionally scale down the power settings so that I don’e go through so many beans.
Yes, one of the first things I checked. The table is level in all directions. Roaster is also level. The honey processed beans produce a lot of chaff and that chaff is clearly getting burnt and mixing with the beans (at least at the front window where I can see). Slow motion video also shows that beans at the bottom of the drum are getting held in place for a split second by beans falling on top of them, even at D9. I think the drum is just too hot and the fan too low. Also, recent wisdom is that convective heat is much better than conductive, but the Bullet heats the drum, not the air coming in, so one hypothesis is that there is a limit to the amount of convective heat that it can develop. Obviously it is still possible to get great coffee from it, but conventional approaches may not work. I’m also more and more convinced that a closed fluid bed roaster with PID burner control is probably the way to go for profile roasting. It is primarily convective and the PID burner control with Artisan allows you to directly draw a profile and have the roaster follow it, instead of trying to guess heat, drum, fan settings to get a profile.
Yes, we seldom discuss/mention bean moisture content here. It has such a huge impact in how the bean roasts and the time it takes to reach each milestone. While a lot are working in low humidity areas where it is common for beans to lose moisture, for us in high humidity regions, we do struggle with higher than ideal moisture content. The fact that we use basic moisture meters does not help.
I knew I recognized your name on the PNG slack channel!
I think you should try the high fan speed profile with this costa rican! 
ive been messing around with f9 or f10 for the entire roast on a few beans, and i really liked it on an eth. anaerobic natural that used to get oily after just 1:15 dev… it really lightened up the roast and helped even things out.
Ive found that costa rican honeys and naturals can be very two toned like what you are talking about, and i think high fan might mitigate some of the scorchiness. I have a costa rica lajas natural that is very chaffy and two-toned. I will do a test roast of it when i roast the PNG.
I kind of think the extremely high convective energy of your artisan is why it roasts more evenly…
