Power profiling

Tell us about your experience, ask if you're unsure of something, let us know if you have a problem.
Post Reply
User avatar
kaffelogic
Site Admin
Posts: 498
Joined: Thu 18 Oct, 2018 9:38 am
x 72
Contact:

Power profiling

#856

Post by kaffelogic »

Power profiling is now available in a beta version of the firmware. This is a basic implementation for evaluation by early beta testers.

The firmware version containing the power profiling is 7.4.4.1092. https://drive.google.com/open?id=15jC8a ... M99T-ccOit

The way it works is this: in a zone, disable PID control and enable power profiling by setting the multipliers to zero. That is, set zone multiplier Kp and zone multiplier Kd both to 0. This will turn off PID control and activate power profiling for the zone. In power profiling the zone boost specifies the percent change in power during the zone. So if the boost is zero, the power will remain unchanged during the zone. But if the boost is 5.0 the power will increase steadily (linearly) to add 5% during the zone. If the boost is -5.0 the power will decrease by 5% during the zone. Once the end of the zone is reached normal PID control resumes.

Power profiling can be used to smooth out the power curve (black line) during specific parts of the roast. Run a roast with no zones and identify any zones where you want to straighten the power curve. Read off the power at the start of the zone, and the power at the end of the zone. Convert this into a percent difference. For example, if the power at the start of the zone is 1075 W and at the end of the zone is 1123 W, the difference is a gain of 48 W. As a percent this is 48/1075*100 = 4.465%. Enter 4.465 as the zone boost. If it starts at 1075 W the power will steadily increase from 1075 to 1123 during the zone. If it starts at some other value, which it might due to variation in temperature, humidity, air pressure, etc, then it works off that other value. For example, if the roaster enters the zone with a power of 1065 W, then the change will be +4.465% of 1065 which is +47.6 W and the end value will be 1112.6 W.

You will need to start a power profiling zone at a point where the roast is behaving nicely - following the profile curve and without significant wiggles in the power curve - otherwise there will be too much variation in the starting power.

I will post an example as a further post on this thread.
Damian
Posts: 111
Joined: Sat 07 Dec, 2019 9:30 pm
Location: NSW AU
x 17

Re: Power profiling

#857

Post by Damian »

I'll play with this some more, but here is my first try
I'm not sure this is going to work as I hoped
11.jpeg
11.jpeg (157.46 KiB) Viewed 3424 times
12.jpeg
12.jpeg (78.2 KiB) Viewed 3424 times
Steve
Posts: 173
Joined: Fri 30 Aug, 2019 7:04 pm
Location: NSW central coast
x 47

Re: Power profiling

#859

Post by Steve »

Excellent Chris, i think this will useful for bracketing problem areas ( briefly taking control from PID) and is likely to work better than just negative or positive boosts alone.
User avatar
kaffelogic
Site Admin
Posts: 498
Joined: Thu 18 Oct, 2018 9:38 am
x 72
Contact:

Re: Power profiling [example]

#863

Post by kaffelogic »

Note that this feature can only be found in firmware v7.4.4-beta. Earlier versions will simply turn off PID when you set zone multipliers to zero.

Here's the promised example: (using very ordinary Columbia Excelsio beans)

I started with this log:
with no zones.klog
(62.54 KiB) Downloaded 235 times
Looking at the log you can see the power curve has only small wiggles up to 5:20. From 5:20 to 7:30 the power curve wiggles quite a bit. My objective was to flatten these wiggles. At 5:20 the power is 1.121 kW. At 7:32 it is 1.187 kW. That's a difference of 0.066 kW. You should open the log file in Kaffelogic Studio and verify these numbers for yourself. Expressed as a percentage this is 0.066/1.121 x 100% = 5.888%.

So I made a power profile zone from 5:20 to 7:32 with boost 5.888. I also made some other changes to the profile while I was about it, but I did not touch the part from 5:20 to 7:32 because that was the part I was power profiling using the power values from the original log.

Running that roast produced this log:
with a power profile zone.klog
(76.78 KiB) Downloaded 230 times
You will notice that the straight section of the power curve is from 5:11 to 7:22. This is offset from the set values by about 10 secs. This is a consequence of the way Kaffelogic Studio makes up for the inherent time delays in the control system.

In this roast the system hit 5:20 at a slightly different power level from the original roast, but the way power profiling works it adjusts so that the power increase is still 5.888% and this keeps the roast pretty much on the temperature line as planned.

After the power profile zone ends some wiggles come back. Actually we might be able to make improvements by ending the zone earlier, using a corner, and careful adjustment of minimum rate of rise. However, I decided to have a go at a second power profile zone. The result was it made things worse, but serves as an example of a power profile zone with a negative boost value. It doesn't quite work as intended because the temperature curve ends up completely flat. The tricky part to this profile is that it tries to flatten the curve exactly at first crack - always a tricky thing to do without either overshooting or crashing. Expect to make further adjustments, but I give you this example as a work in progress:
with two power profile zones.klog
(73.39 KiB) Downloaded 216 times
Post Reply