Marcism
I just tried Artisan 1.5.0 today.
Before I was on 1.3.1
PID has changed between those two versions.
"P on Input" is added as an option to chose instead of "P on Error".
I expected P on Error to be the same as in 1.3.1 but it is not.
In 1.5.0 I too get the Power to max on start. I did not get that in 1.3.1
That causes the bump on the curve that is the main purpose of the addition of "P on Input", as far as I understand.
In 1.3.1 I did not get that, but instead had a rather slow climb to the background curve.
In 1.5.0 the PID is speeded up at start, but I am not sure I like that.

I tried to switch to P on Input, but no matter what I do, I can not tune it.
I've read all I can find, but with no manual tuning guide I find it hopeless.
So back to P on Error.
Will try your approach tomorrow. Thanks!
/Henrik

If it is not doing what you expect and you have read the release notes, please post an issue here:

https://github.com/artisan-roaster-scope/artisan/issues

Proportional on Error is the older PID logic, and P on Input, more exactly "proportional on measurement" is the new addition.
The PI coefficients are different than PonE, that's why you see different behaviour.
In PoM mode the P term is a braking force, not the factor that led the power increase. The power is driven now mainly by the I term.
You can read more about this mode here:
http://brettbeauregard.com/blog/2017/06 ... asurement/

Sinobi wrote:Marcism
I just tried Artisan 1.5.0 today.
Before I was on 1.3.1
PID has changed between those two versions.
"P on Input" is added as an option to chose instead of "P on Error".
I expected P on Error to be the same as in 1.3.1 but it is not.
In 1.5.0 I too get the Power to max on start. I did not get that in 1.3.1
That causes the bump on the curve that is the main purpose of the addition of "P on Input", as far as I understand.
In 1.3.1 I did not get that, but instead had a rather slow climb to the background curve.
In 1.5.0 the PID is speeded up at start, but I am not sure I like that.

I tried to switch to P on Input, but no matter what I do, I can not tune it.
I've read all I can find, but with no manual tuning guide I find it hopeless.
So back to P on Error.
Will try your approach tomorrow. Thanks!
/Henrik

I did a test yesterday and set my lookahead to 5 seconds and it did still shot power to 100 but managed to correct itself once the alarms kicked in. Stick to 10s for now.

EDIT: Looks like I accidentally had Start PID on CHARGE ticked back on? Might be the cause there!

Here's my alarm set up:

renatoa wrote:...."proportional on measurement"...
You can read more about this mode here:
http://brettbeauregard.com/blog/2017/06 ... asurement/

Thank you renatoa, I appreciate the input.
I can tune P on Error thanks to a few guides on the net especially wikipedia.
As I wrote I have read what I can find on the net in relation to proportional on measurement (not much out there), including the link you provided.
And I did understand the more or less inverse function of the P.
Yesterday I updated from 1.3.1 to 1.5.0 and turned on P on Input, but 3 hours of trying to manually tune it, made me give up (that rarely happens).
I feel I miss something like a guide as how to approach manually tuning it.
Just doing as with P on Error, but adjusting P in the opposite direction didn't work for me.
When switching back to P on Error I was surprised to see that its behaver has changed a bit since 1.3.1, in the first minute or so from PID ON, just like Marcism described.

So does "P on Error" prevent the initial jump from happening? I am only familiar with 1.5.0 and its behaviour.

I did another test with Lookahead set to 5s but "Start PID on CHARGE" set to off. No problems at all apart from a slight overshoot caused by not having the PID completely tuned. I ended up with Lookahead set to 10s, however.

On 1.3.1: Yes, no jump to 100
On 1.5.0: No, like you described, it jumps to 100
When you read the description of the P on Input, two of the main reasons for implementing this, is because it's better modelling a system with a changing input (like a roaster), and that helps getting rid of the overshooting in the beginning that P on Error typically have.

Very Interesting! I think my workaround is going to have to do for now. I guess this could be considered a bug? Unless we are missing something key here. I do feel like the more I delve into Artisan, the more I realise I lack a LOT of knowledge in these areas