My Santos is super bitter and not drinkable

Dear all,

currently I am trying to roast “Santos Natural NY 2/3 Scr. 19 (Arabica)”.

Now, after one week I tried the first espresso and it was not drinkable. Very bad in taste - super bitter.

What do you think what the problem could be when you are looking on the curves of this roast here:
Roast World - Cup, grade, and analyze your coffee roasts in depth [

I have also tried with preheat temp of 180 degree. The coffee taste is the same. Not drinkable :frowning:

Do you have some hints for me why this could be the case ?


1 Like

link doesn’t work.

ooops, sorry. Hopefully that link is working:

lots of fiddling with the power. Try for a steady power decrease and fan increase. first crack looks too low. Santos will crack around 200 consistently. I’d just try again removing the p6,p5 leaving a steady p7 you’ll see a bigger temp drop for the p6 as you introduce f4 at the same time which will have a cumulative effect which you tried to correct by increasing the power again. I reckon your first crack actually occurred around the 8m mark which gave you around a 50 second dt and 3 degrees of stretch. just my humble opinion. Hope this helps

1 Like

thanks for your opinion … i will try. that means you are a fan of constant p7 and increasing Fan speed right ? so stabilizing a constant power and doing increase of fan. yes it could be that the FC was to fast … which % of DV are you prefer for the santos ? more than 20% ? thanks … do you have santos recipe maybe to do a playback on my side ?

no it means i’m a fan of not making too many nested changes as you have no hope of controlling the roast. I roast 1.2kg per batch when i do So my power settings would be different but that’s part of your experimentation to see what you prefer. Enjoy the coffee.

1 Like

I tried to open the roast, but it seems unavailable now.

There was a time i’d say don’t fiddle with the roast, but looking at the Sweet Maria’s roast examples and many of the “good”(we can’t taste them) roasts here, we see a lot of fluctuation. I’ve also have steady roasts taste worse than some of my fiddly roasts.

The rule seems to not ever go backwards. as long as the temps are still climbing.

Many roasters come from a production background and will aim for simplicity and repeatability, but this machine in a home roasting environment should be about fun and exploration. it isn’t such a precision machine that i’d get too into the weeds, but it can still do quite a bit of good results. obviously it is expensive, so it isn’t about saving money either. have fun and explore, take notes, be able to track your roasts. write the name on the container or bag of each roast, so when you drink then, you really know what roasts you have a preference for.

I’m saying this everywhere, but it would be a huge thing if Aillio would allow us users to access some of the global roasting statistics. it would help take some of the guess work and learning curve away from new users, and encourage experimentation and innovation.

If we all know what global and regional Bullet roast trends look like, we’ll have more people trying to innovate outside the lines a little. we’ll see people learn faster (which is better for the planet and less waste) and everyone can grow together. Right now we have everyone roasting pretty similar and conservative, because we have no idea what the standards are, just sort of roasting on fomo for a year, never sure if the feedback we get from one user is really useful or if they are just some one with an opinion. Going to probably over share this concept, as i feel it is really important to the community growth.

Thanks Jimmy for your opinion.
This was the Profile which seems to have a problem due to an application server:

Yes I know it is all about exploring. But the last 10 roast where not drinkable because it was a fried taste. therefore i am struggling because I have not a glue why this occurs. For me - this is disappointing - but I have to find a basic roast profile to get drinkable and not fried coffee taste - after i have solved this i can explore with different tastes …

1 Like

despite there being any number of variabilities to roasting flavors etc, in general there is no real mystery. for the most part basic roasting should taste fairly good and be quite easy to hit.

You roast curve is no longer available.

One consideration that I don’t think I saw in any earlier posts is that occasionally the green coffee is at fault, and there is really not much in terms of roasting technique that you can apply to fix it.

Or the coffee might be fine for others’ tastes but not yours.

Not saying that either of those was what was happening with your Santos, but I’ve had a few go that way.

When you say a “fried taste” can you provide some more description for that? You mentioned too bitter for your April roasts…are you still experiencing that with recent ones? Or grassy? Bready? Burnt?

What flavor profile do you prefer in a good cup?

There are a lot of really good roasters on this forum…if you can provide more descriptors they might be able to give some useful pointers.

Hi John,

I love dark italian espresso.
No, the last batches were all BAKED. not drinkable. A terrible taste and i am currently not sure what the problem ist …

e.g. this was also baked

yesterday I tried this:

hopefully the baked taste has gone and I have a basic profile to work on the next weeks.

My goal is to find a good 350g profile to use as basis to come into the direction to get a real dark espresso → and to solve the BAKED roast problem :slight_smile:

these two were completly BAKED

SZI Brasilien Red Iapar Bio (Arabica) (Manuell → Röstung anhand Vorgabe Espresso Profil “dark” (Roastrebels, Website)) | Roast World

SZI India Parchment (Manuell → Röstung anhand Vorgabe Espresso Profil “dark” (Roastrebels, Website)) | Roast World

if not accessible i will send the SHARE link:

Yeah, idk, i was thinking that since Aillio collects all of our roast data that they could post a few different averaged roast profiles to help people get started. Going thru individuals recipes one at a time is time consuming, and only a few seem to work across the board. I have tried my own stuff, and i dislike what iam getting, but i guess i was trying too hard to be different. I think when i see the regular posters post roast curves, they all seem to be similar. i don’t think there is any trick to roasting, other than not getting too clever.

My guess is that the end temperature is too high for your tastes. Try ending at a lower temp.

Yeah, I hope to find a basic profile very soon which works with most of the beans. It does not matter if it is a real italian espresso or more a light one. I want to solve the issue to get an espresso which is not drinable (it tastes like hell currently),. So getting a basic profile for little samples (max 350g) is my goal and then use this profile to move forward. As of today I was not able to find something like that and therefore i fired 10kg into hell … i dont think that there is somenone on earth who will say it is drinkable :slight_smile:

1 Like

I will send you two pics of my beans (one with photo flash and one without).
Do you see something which is not good (scorching or something else) ?

Any hints appreciated. This coffee is not drinkable … bad bad taste.

Hey Steffen, first off, I’m also relatively new and eager to learn… and troubleshooting bad flavors. The pic above (non-flash) definitely shows signs of ‘tipping’ which would be consistent with the scorched flavors you’re reporting. From what I read that’s an indicator for the scorched flavors. I see about a dozen examples of tipping in the photo. Recognizing nobody here is roasting on a Loring, the Loring website has this to say about tipping:


This defect is caused by the coffee being heated too rapidly. Frequently it has to do with fast roasting profiles with coffee whose density is not able to handle the heat without creating this physical defect. This defect is recognized by the bean having a blackened spot on only one tip of the coffee bean either during or post roast.

Note their comment about density of your green, which may apply here. I also notice you’re a fan of a dark-roast espresso. I’m targeting a ligher City or City+ roast and have also been wondering how I’m getting toasty, over-roasted flavors (think Starbucks Pike Place) when according to my roast curves, my bean temp has never exceeded 415 deg F / 212.8 deg C, which Sweet Maria shows as only slightly hotter than a typical first crack temp (412 deg F / 211.1 deg C).

Hope this is helpful and happy roasting! :slight_smile:

@jimmybulletroaster what kind of statistics would you like to see? I was able to query for Roast World roast data for nearly 40k roasts with roast degrees from 0 to 3 (lightest to medium). I’m personally biased and don’t care about any roast data for roasts taken beyond medium. I haven’t had a chance to review and clean up the data yet, but I’m assuming that there are much fewer than 40k roasts that have full data points…but there should still be a decent chunk of clean data to work with.

What kind of statistics would you like to see? Everything but the roast curve data? Or would you see value in somehow visualize roast curves in aggregate too? I see some examples of visualizations mentioned elsewhere by @fwbennies, but I’m not sure how the roast curve visualization would work when dealing with multiple roasters as opposed to a single roaster’s outputs.

If you have any ideas on how to represent the data, please let me know. I don’t think we have to wait for Aillio to officially support these kinds of stats to start gleaning information from everything that is already publicly available on

From what i’ve seen, we can access these roasts, but it is really hit or miss.

Being able to see trends is a great way to know where the average approach lies.

If a trend emerges because people are consistently roasting a certain way, you have a better understanding of more general standards.

But the more specific the ability to narrow trends would be interesting too. Based on region (weather climat impact) and altitude (water evaporation and barometric pressure impact).

If this data collection is truly mandatory for the purpose of helping us learn, i feel that more comprehensive analysis tools could help. Where else can you get such high level general data? And it really only directly applies to Bullet roasters anyways.

In a perfect world, would like to be able to analyze trends more than just individual roasts. i can already see my own trends, but it would take years to seek out and test individual roasts. Seeing trends can help in speeding up learning and trying more educated approaches.

Trends would be an average time, temp, and curve with certain parameters applied. Parameters could focus these trends as:

  • bean origin
  • bean density
  • end temp
  • roast color (full city etc)
  • elevation of roaster
  • elevation if bean
  • global over all roasting trends averaged
  • regional roasting trends
  • star rating
  • flavor descriptors
  • differentiating between just roast history and those that have entered more data
  • compare trend to specific roast curves
    and so on…

I think when you can see bell curves and roast trends in the data, it works to give perspective and a reality check. If we go by one off curves one at a time, that is slow going.

I’ve seen the ability to find trends like this in other industries and professions, and it is really impactful and important. It is surprising that with ai and ML, as a roasting community we’re so far behind the curve (no pun intended).

If you want to see faster progress in learning, and want to see more critical participation from the community to enter data properly, there has to be some sort of tool or incentive available to make it all be a motivational factor. At this time it is all close you eyes and pick a roast from hundreds of options and hope for the best.

The specifics for some of those trends would be harder to collect like elevation of roaster. only has general location/region data per user, so it would be a rough estimate, if anything.

However, some of the other ones we could start using to look at trends. I validated the original 38,597 roasts that were roast level 3 or less. As far as I can tell, that should be all roasts from level 0 to 3 that exist in the Roast World database. Out of those nearly 40k roasts, only 16,354 include all data points.

The majority of these “invalid” roasts were only missing ambient temperature and humidity, but for this initial assessment, I’m ok with throwing out anything beyond the solid 16k roasts because I think it reasonable to assume that any roast that doesn’t capture ambient temperature and humidity do not provide enough insight for evaluating trends.

I’m probably going to start with looking at some of the simple numbers like start and end temperature and then expand from there. I’ll start a new thread with any interesting findings since this is getting too off-topic for this thread.

You can filter some of the results. but on the “discover” page, you can still only click on one roast at a time.