Here's my method for avoiding the "flick" on a gas drum:

1. Enter first crack with a reasonable amount of momentum.
2. Cut your gas by 30-50% around halfway through first.
3. Starting around 1:15 into first, keep a very close eye on your ROR curve. If it starts to rise, consider dropping gas a bit lower (possibly to zero if you are executing a city roast in a solid drum).

With this approach, I get no flick on around 50% of my roasts, and I can reduce it to a mere "flicker" (a small increase followed by a quick return to a steady or declining ROR) in most of the remainder. FWIW I do not find that a flicker harms a cup nearly as much as a rise in ROR that is then sustained at the higher level; frankly, I'm not even sure I could detect the effect that a flicker has in a blind cupping.

For me .. in my Santoker, the flick is the consequence of the heat I apply to the drum at the start of first crack to avoid the profound RoR drop that frequently happens if I don't anticipate, and not backing off and applying a bit more air flow soon enough while first crack continues ... usually starting to back off off at about 45 seconds post first crack start and decreasing heat throughout. If I do that well I will not have a flick and avoid the drop of FC. On my machine I must also cut heat (and or increase airflow a bit) prior to first crack to avoid the slow RoR that happens before FC. Throw the fact that every bean seems to want to release and absorb heat differently and it seems daunting. I love the analogy of roasting being like the captain of an aircraft carrier where you have to anticipate your move way in advance and realize you have momentum (heat in the drum) ... really a challenge but so rewarding when you do it well. I also find it does make a better cup. I'm sure everyone's roaster is different and experiences things differently but learning your machine and trying to master it is the fun in roasting. Getting closer all the time ... learning from mistakes ... and trying different things. Enjoying the journey ... Steve

I'm in the same boat. Cut heat approaching first crack then increase just after the start of first crack, then decrease again about 45 seconds into first crack and keep decreasing through to drop. I generally am increasing air as I approach and go through first crack as well and that seems to make it a bit easier to control. Always seems to help, but don't get it just right every time for sure.

Jon

Has everyone here (that wants to get rid of the flick) actually done blind A/B comparisons with "flicked" roasts and "non flicked" roasts of the same coffee? This was the first thing i did when Scott's book came out. I really wanted the declining ror to instantly make all my coffee better. But thats not what i found. So I tried again. Who wouldn't want a simple thing like this to make all there coffee better? Again, the results were the flicked roast wasn't worse.

Ive repeated this same thing multiple times on this site. But I know from experience people are prone to believe the "experts", and not what they have experienced themselves. Im confident in my tasting capability's, and so far in my experience non flicked coffee isn't magically better than coffee with some degree of flick. So before you take the declining rate of rise as gospel, maybe do a few roasts where you match the weight loss and times (for the most part) and see if you prefer the non flicked roast?

Thank you Dustin. Perhaps the emperor is indeed clothesless.

Scott stopped by a while back to discuss some of this with us; you can find the exchange here:

Development time as a ratio of roast time by Scott Rao

In that thread I posted about one blind comparison I did, where a SS-FF, flat-line (but not flicked) profile beat out a Rao-rules profile in a blind cupping. Soon afterwards I did another similar experiment using that coffee, with similar results (at least at city roast levels):

Roast and Learn Together - August 2014

But since then I've mostly become a convert, at least when using a traditional drum roaster. In the March-April R&L thread, Marshall encouraged us to couple Rao's "commandments" with a restriction to never increase heat input during a roast. (I had typically bumped heat up as a way to draw out drying and then have a fast ramp, for all my prior experiments.) When I tried doing Rao-style profiles in this manner, they consistently beat out all my other efforts, including the SS-FF, flat-line approaches that I had liked best in prior trials:

Roast and Learn Together - March/April 2015

So now I tend to treat his recommendations as much more valuable. But none of this has included an attempt to isolate just the "flick." I am always game for some blind testing of roasting theories, so I'll give that a shot at the next available opportunity, and post about the results.

Here are two nearly identical roasts I did 10 days apart. In the first graph, even though I was decreasing gas, and increasing air settings during first crack, you can see a classic "flick". In the cup, this roast was very noticeably baked, which I believe (other opinions or thoughts welcome) was caused by this "flick". The second graph was my next attempt with this bean and, using the roast log from the first roast, I cut back on the gas a little more at the start and also during 1stC. This pretty much eliminated the "flick" and made a dramatic improvement in the cup with no apparent baked flavor. While there's definitely room for improvement in this graph, like the flat to rising ROR in the 6-7 min. range, and almost flat ROR toward the finish, my over all impression of this cup was that of a nice, classic Colombian, while the first roast was pretty bad.

470g load
Charge temp: 375F
Turn: 1:15 @ 233F
EOD: 4:07 @ 311F
1stCS: 398F @ 7:44
1stCE: 431F @ 10:02
Drop: 440F @ 10:33


470g Load
Charge temp: 375F
Turn: 1:15 @ 237F
EOD: 4:00 @ 310F
1stCS: 397F @ 7:47
1STCE: 427F @ 9:54
Drop: 437F @ 10:45

You've uploaded the exact same graph twice.

I was trying to see the difference in the two and couldn't figure it out. Ha

Hmm, just before the bean name I'm seeing #122 and the second graph says #129. Sorry, I will try to fix.