RoR crash Troubleshooting
Reply
Junior Member 3 posts alexandre.avoyer Junior Member 3 posts
alexandre.avoyer
Hi all, 

Would like to know if there is any way to stabilise the RoR during the yellowing phase and stop the crash and re rise. Is anyone else having this problem, is it a question of energy to grain ratio? 

Thanks your your thoughts.
Reply 0 0
Sverre Simonsen Co-Founder & CTO 249 posts
Sverre Simonsen
Hi, 

Could you please post a picture of your log? 

Its perfectly normal that the bean temperature reading crashes during first crack,  that's the results of very sensitive sensors and a small bean mass.  Personally, I use it as a tool to predict first crack  in washed coffees👍 
Reply 0 0
alexandre.avoyer Junior Member 3 posts
alexandre.avoyer
Hi,

Its not a problem during the crack, it's more during the first 2 min of the roast, absolutely crazy numbers.
Reply 0 0
Sverre Simonsen Co-Founder & CTO 249 posts
Sverre Simonsen
Could you post a photo?
Reply 0 0
Kris Junior Member 1 post
Kris
I guess Alexandre is refering to the drop and rise between 2 and 4 minutes. To me also not clear where this is coming from. How is ror calculated? 
Reply 0 0
Adam S. Carpenter Member 52 posts
Adam S. Carpenter
There are 2 ideas rolling around here. 
1. Should we follow the advice of Scott Rao and never see our ROR change directions.
2. Should we understand what a sudden burst of moisture does to the bean temperature probe? 

Just like Sverre said in his comical note, "I judge first crack by my crash..." (I am paraphrasing) you can also judge that point where your coffee smells like bread or fresh cut hay/straw by the sudden burst of moisture during the endothermic phase after dramatic turning point 2-4 minutes.

ROEST is giving us such a beautiful readout of our roasting screen - it actually looks bad... due to it's accuracy. If the averages and ROR was calculated slowly, we would get these nice smooth lines rather than jagged slopes and unconventional (non-Scott-Rao-abiding) roast curves.

I digress. Let's identify the starting point. Is it:

1. My coffee tastes funny and I want to know if it's because of my roast curve?
or
2. My roast curve looks funny and I want to know if that's a problem?

The answer to Q.2 is taste the coffee and find out if changing your roast curve makes it taste better. Otherwise, if your coffee tastes amazing, just ignore a funny looking roast curve.

Am I missing something my friends? Thanks!
Adam
Reply 0 0
Eric Scace Eric Scace Junior Member 10 posts
Eric Scace Eric Scace
I have been investigating these questions about RoR management as well.

As I understand it, the RoR calculation algorithm is the lagging ∂T over a 30-second interval. (This is the same interval as used by Rao.)

One can fill in the valleys or trim off the peaks of a RoR curve by changing the air temperature profile during the 30-45 second interval before the peak/valley.

My preliminary — and very rough — rule of thumb is to shift the air temperature curve up or down by about ½ the change in ∂t one wishes to achieve, and then run another sample. There appear to be momentum and timing issues that may play a role in how much actual change will occur. But the idea is to balance added heat, reduced heat, and the heat generated/consumed in exo-/endo-thermic process.

I usually roast about 5 experimental tweaks in this way, marking each batch, and then cup/brew/extract the next day (and usually again the day after that) to form opinions as to the significance (if any) of these changes.

There is a certain amount of "gilding the lily" — but it's also educational.

A related, but best addressed as a separate thread, is what happens when one attempts to create a RoR-based profile (rather than air temp). The 30-second measurement interval for RoR and time it takes to impact bean temperature creates some "interesting" radical oscillations. More investigations are needed in this area.

— Eric
Reply 0 0
alexandre.avoyer Junior Member 3 posts
alexandre.avoyer
Greetings all,

First of all I would like to thank everyone for their replies and participation, this makes me very confident in the Roest community and our ability to work together to better understand this amazing machine.

Just to be clear, the question I asked was more to understand if this peak and drop in RoR is normal and if not what measures must be taken to course correct. Needless to say I have no illusions that the Roest is not a fluid bed roaster nor a traditional gas roaster hence new norms and mindset.

That being said, it is a machine that gives us the possibility to analyse and control the way we roast with real time readings and profile editing, if it was only the cup that mattered none of us would have spent this amount of money on such a great tool, so lets keep the theory objective and to the point.

I have sat in and read books from Scott Rao and I believe he has great content and experience, however the fact that they derive from his experience with mainly gas drum roasters of usually high batch capacity makes his theories for my part non applicable for the Roest setup. So lets not name names and keep it objective.

What I would like to know is, what kind of RoR calculation formula are we working with, how and what are the thermometers reading and to what end, if there are 3 different modes to create profiles it means these three are applicable (I am only seeing consistent results with air temp) so should we consider the Roest an air roaster with a drum? If the argument for the RoR peak and crash being due to the drying phase and hence water loss it means that if I roasted empty this would not happen (I just tried, it did happen). 

Cut a long story short, is this RoR peak and drop (45 degrees/min at peak down to almost 10 and stabilising at 5) a normal occurrence with the Roest and part of its DNA, if so why and to what end. If not....what am I doing wrong.

Thanks all.
Reply 0 0
Thomas Senior Member 129 posts
Thomas
I can't see a so called RoR crash,what I see is that this bean needs more energy in the area before and while RoR drops rapidly than the given ET profile provides.
In this case I would work on a power profile and translate the resulting ET into a ET profile.
Look at these two roasts with the same bean, the difference in power is marginal.
Do I taste it in the cup?
120g,Altieri natural Gesha.

Bildschirmfoto 2020-11-22 um 13.50.51.png 

Bildschirmfoto 2020-11-22 um 13.51.05.png
Reply 0 0
Jon Arne Jørgensen Jon_ROEST ROEST Software Developer 160 posts
Jon Arne Jørgensen Jon_ROEST

Hi all,

I'm in the process of trying to make the ROR algorithm more responsive.

I'm using a currently unreleased firmware that actually logs the temperature exactly once per second.


The ROR curve is calculated in these steps:

1. Calculate the rise in bean temperature from last second.

2. Put this value into a moving median filter, that should remove any sudden spikes.

3. Put the filtered value into a moving average with 8 values (8 seconds).

4. Multiply the average from the window by 60 to transform the value to Change per Minute.

 

I've also written a Python script that can take the downloaded CSV from a log,

and show a plot of how the ROR curve will look with this algorithm.

If you have a log you would like to see with the new algorithm, just send me the CSV, and I can generate the ROR curve.

If you are comfortable with Python, I can send you this script, but I would need to clean it up a bit first.

 

Here are some screenshots for comparison.

The first is the Old algorithm.

Screenshot_20201222_161736.png  And this is the new one. Screenshot_20201222_161650.png

I ran out of coffee, so the roast with the current algorithm is only 90g that's why it's a bit shorter.

Please reply here with any suggestions 🙂
Reply 0 0
Thomas Senior Member 129 posts
Thomas
Maybe something between the two versions, or like in Artisan to have the choice to activate a filter to smooth it.
Reply 1 0
Paohui Law Junior Member 27 posts
Paohui Law
In what units are these ROR values displayed in? Celsius/minute or Fahrenheit/minute?
Reply 0 0
Jon Arne Jørgensen Jon_ROEST ROEST Software Developer 160 posts
Jon Arne Jørgensen Jon_ROEST

@Paohui Law,

They are displayed in Celsius/Per Minute. Or Fahrenheit/Per Minute depending on the settings in the web client.

But I just discovered that there is a bug in the conversion to Fahrenheit for the ROR curve, so I'll fix it during today.
Reply 1 0
Paohui Law Junior Member 27 posts
Paohui Law
@Jon_ROEST can you help me with how differently the old and new ROR algorithm works? I've seen a drastic change in responsiveness between the old and latest firmware and I assumed that we've switched algorithm? 
Reply 0 0
Thomas Senior Member 129 posts
Thomas
Paohui Law wrote:
@Jon_ROEST can you help me with how differently the old and new ROR algorithm works? I've seen a drastic change in responsiveness between the old and latest firmware and I assumed that we've switched algorithm? 
 
Have not roasted since the update, what has changed?
Reply 0 0
Reply