Scott Rao - ROR rate of rise "the flick"

[hankua]

I copied your setup last time, but just put a newer hand me down laptop in service, and the smoothing setting were probably default. Also got a pop up warning regarding less than 3sec sampling.

[JavaBuzz]

Hankua,

Marko (the creator of Artisan) has a blog post that has some good information on temp readings in general and provides some recommendations like not setting the sampling interval too low (he recommends 5 seconds, though a little faster is nice if you can do it). Basically go as low as you can on sampling interval while still maintaining readable live graphs:
http://artisan-roasterscope.blogspot.com/2014/01/sampling-interval-smoothing-and-rate-of.html

Enabling Oversampling is recommended by him I know (Config > Oversampling). However, I believe oversampling is disabled when you set the sampling interval less than 3 seconds, even if it's enabled in the Artisan menu (I believe he mentions this in another blog post, though maybe newer versions don't do this).

I have my sampling interval set to 3 seconds, with oversampling enabled. I found that going less than that with my setup made the temp graphs - especially deltas/RoR - very choppy and at times almost unreadable during the live monitoring (when it really counts). Depending on your meter and temp probe types, you might want a little higher or lower sampling interval.

You can up the "Smooth Curves" and "Smooth Deltas" (under Tools > Extras), and that might help a bit, but Marko also recommends upping these settings as a "last resort" (and as minimally as possible).

NOTE:
From a post I read somewhere, the "Smooth Deltas" setting actually only affects the RoR graphs after the roast has completed (after you have clicked Stop). This appears to be true from my testing as well. However, the "Smooth Curves" does appear to make the live delta/RoR readings a little bit smoother (as well as BT/ET), so best to up that if the deltas/RoRs are hard to read during the live monitoring and you can't smooth the readings at the probe/meter level (if they are choppy).

Side notes:
I have a TC4 meter. I have currently settled on an 85% "filter" setting for the temp probes. This makes the temp readings much less jittery (though does make the temp readings lag slightly more).

The temp probes that came with my Huky do seem a bit over-sensitive. Not sure if all Hukys have come with the same probes.

[SusanJoM]

Enabling Oversampling is recommended by him I know (Config > Oversampling). However, I believe oversampling is disabled when you set the sampling interval less than 3 seconds, even if it's enabled in the Artisan menu (I believe he mentions this in another blog post, though maybe newer versions don't do this).

Oh good, maybe you can explain to me what "oversampling" is?
I've had sampling set to 1 since I first set Artisan up with Moshe's help (SAS at H-B) and it has seemed to work just fine, but I have no idea what "oversampling" is or how it would change things

Susan

[JavaBuzz]

I believe this should be a good simplified explanation of oversampling:

Within a sampling interval, Artisan can receive multiple readings from a meter...

With oversampling enabled, Artisan will record as many of those readings as possible, then at the end of each sampling interval, it will "average" all the readings in that interval (or might technically run some type of equation against them?), and record/graph that as the temp. This can help substantially with smoothing out the graph, reduce "jitter" from errant or sporadic readings, and make the graph more readable in general.

With oversampling off, when the end of the sampling interval is reached, Artisan just takes the newest reading received from the meter. If that reading is way off (which can be more common with high-sensitivity probes) you might see a lot more "jitter" in the ET/BT/etc. readings, which in turn could make your delta/RoR curves even worse.

Hopefully that makes sense!

[JavaBuzz]

Here's some Artisan basics, as I've come to understand them, for fixing "jitter" (my definition of "jitter" would be too much variation in the  temp readings, making the Artisan live graph hard to read/follow):

Too much jitter (making graph unhelpful):

1) If you can, buy temp probes that are not overly-sensitive, but sensitive enough to give you helpful readings (look for ones that minimize grounding issues too)

2) If you don't want to/can't spend money on new temp probes, if possible adjust settings on your meter to smooth out readings from the probes (pretty sure both the TC4 and Phidget can do this, not sure on others)

3) If you can't do the above (and/or need to do fine adjustments), try turning on oversampling, and possibly also slightly increasing the sampling interval

4) If you think you're in the ballpark on the sampling interval, but want to smooth out readings just a bit more, increase the "Smooth Curves" setting slightly.

5) If, after a roast is done being recorded, you want to make your RoR curves look "prettier" (and hide some brief spikes) you can increase the "Smooth Deltas" number. You can actually increase this after a roast is done and watch how it affects the delta curves behind the window.

[SusanJoM]

The explanation of oversampling made sense to me but since using a 1 second interval seems to work  just fine with the Phidget I think I'll stick with it.

The T/Cs that came from Mr. Li and the Phidget and Artisan all seem to communicate flawlessly, and tell me what I need to know about my roasts.  What I do see now, however, is that communicating with people who might have installed different T/Cs or whose Artisan settings are different from mine becomes problematic.  What we are seeing on the graphs becomes more and more unclear.

Ain't it always the way?:  the more you learn the less you know 

Edited:

So I just went again to the link JavaBuzz posted at the beginning of this thread and read it for the umpteenth time.  My eyes glaze over at various places, but this paragraph stands out:

The sampling module of Artisan takes readings according to the user specified sampling interval without applying any smoothing. If oversampling (introduced in Artisan v0.7.4) is active, two readings are taken per interval and the average is computed and taken as reading, replacing the two raw readings, which are dropped. The drawing module applies live smoothing on the values recorded by the sampling mode according to the amount specified by the user during recording and static smoothing after roasting and for displaying the background graphs. Therefore, a time lag might be observed during roasting that is eliminated afterwards. The smoothed curves are only computed for display purposes in Artisan. Artisan always stores and loads the raw values gathered by the sampling module

Seems to me that what it says is that if you are wanting to see your temps in real roasting time that the lowest possible interval is your best bet;  for the purposes of reading a graph after the fact, perhaps increasing the interval and adding oversampling would make more sense.   

Thanks JavaBuzz for reminding me that Marko really is the best teacher of all.

[JavaBuzz]

This thread has definitely got off-topic now (and I'm partly to blame), but it's an interesting discussion nonetheless. Maybe we should move some of this to a separate forum post?

If you are able to...
-Set a 1s sampling interval
-Set "smooth curves" and "smooth deltas" to 0 (at least temporarily to see the true output from the probes and meter)

...and the graphs still look somewhat tolerable (especially delta/RoR), I'd say go for it.

If they look aweful and totally unhelpful, you probably want to adjust the sampling interval at least 1s higher (and turn on oversampling if over 3s). Rinse and repeat if still look bad.

If you want, you can readjust smooth curves slightly up to make the graphs even a bit better.

Susan (and all), the Phidget has a USB-native interface, and that is probably at least partly why it can better handle the higher sampling interval of 1s. Also, it might internally do better sampling/averaging itself of probe readings.

The TC4 (and many other temp meters) use a USB-to-serial converter chip to convert the signal sent to the computer. I think that this is a bottleneck with the TC4 meter. If I lower the sampling interval to 1s, I can watch the delta curves get dropped to 0F every few readings because something in my system/setup (probably the TC4) can't keep up with the speed (though the BT/ET curves look fine, I think because "Smooth Spikes" is enabled). If I raise the sampling interval to 2s, I stop seeing the drops, but I decided to go to 3s so Artisan would use oversampling and give a slightly better looking reading. I am still tweaking my own settings slightly here and there, but that is where I'm at currently.

[JavaBuzz]

If people are interested, I think I might try to put a document together on "basics of adjusting Artisan for your roasting setup". Then I could post it here (and maybe HB) as well.

I'm sure there would be differing opinion on how best to do this, but I think it would be helpful at least as a guideline, and could always update it with tips and pointers from people.

The Artisan menus can definitely be confusing, and I don't even know for sure exactly what some of the settings do after hours of researching the subject.

If people are interested, let me know.

[SusanJoM]

I have a couple of questions about the Extras Page.
What  does "Delta Span" mean?
What does "Window" mean -- t here next to Smooth Spikes and Smooth2
and
What does Path Effects mean?

As you can see I have stuff in those boxes, but I'm not sure what what I have in there does or how changing it will impact my graphs and I'm always a bit scared to change things without a better idea of what I am messing with.

[JavaBuzz]

Those are all settings I'm not sure of myself. I just leave them at defaults too.

Smooth2 is another type of smoothing (similar to Smooth Spikes). From what I recall, I think Marko had a blog post where he said most people will probably want to stick with the defaukt of only Smooth Spikes checked. I think it was a lot more CPU intensive and he hadn't tested it as much (might have been written by someone else)?

My best guess on Window is its a sensitivity setting for Smooth Spikes and/or Smooth2. Just guessing though.

I think Delta Span might be how far back in time delta readings look when averaging the current RoR reporting, but I could very well be wrong here too. When you adjust it on a saved profile, it moves the delta curves left or right.

Pretty sure Path Effects is a basic visual setting, but no clue what it does.

If anyone else has thoughts, I would be interested to hear them.

[hankua]

If people are interested, I think I might try to put a document together on "basics of adjusting Artisan for your roasting setup". Then I could post it here (and maybe HB) as well.

I'm sure there would be differing opinion on how best to do this, but I think it would be helpful at least as a guideline, and could always update it with tips and pointers from people.

The Artisan menus can definitely be confusing, and I don't even know for sure exactly what some of the settings do after hours of researching the subject.

If people are interested, let me know.

That would be awesome! 

[SusanJoM]

My first flickless roast....Whoopee

[phischmi]

Looks great!
Do you wanna tell something more about it?
Look's like you kept the fan at 50%. Or did you just forget to record those adjustments?

[SusanJoM]

Don't forget:  I am using the perforated drum and slow motor.
My settings won't do you any good.

But....the only thing that isn't registered is that the gas actually went OFF when BT reached 405.

I left the grate open and the fan at 50% for the entire roast.

[Gregr]

That is one pretty profile. Have you tasted it yet?

My first flickless roast....Whoopee