Tweaking a PID for La Pavoni

"On the road" set for coffee

Being on a holiday trip is a great time to work with smaller equipment and to take a closer look at the configuration. Spending the day with these machines saves you much time and money which would otherwise be spent on sunscreen, sculptures and cathedrals.

Earlier this year, during a stay in Poland, I noticed that both Pavonis deliver equally delicious espresso, even if one just has its internal pressostat to control the pressure/temperature of the boiler and the other has an elaborate PID system with a Fuji PXG4 controlling the exact temperature of the brew head by regulating the energy of the heating element inside the boiler, and Artisan above it all, monitoring and logging what the PID is doing, how much energy the element is spending and the effect over time on the brew head temperature.

As you can see in the illustration above and below, I noticed this time that the ideal level of brew head temp was not successfully managed. I tried to start the Fuji's 'autotune' but as the Fuji fired up the element to measure its effect, the over pressure valve began to hiss urgently and the manometer dial showed a pressure well into the red 'danger zone' so I stopped that and tried to change the settings manually.

Red line showing overactive PID control

Starting a recording in Artisan. Tasting proved that here, with my Ethiopian beans, the temperature on the outside of the brew group before start of the shot is best at 77ºC (the extraction temp inside the puck is higher of course). 

Close up of the temps during / after an shot of espresso: 12.5g in the basket, 0.4 bar on the gauge, 12 seconds pre-infusion, 20g of espresso in 30 seconds. Temperature probe inside the cooling ring against the brew group varies just over 1ºC. The red line shows the target 77ºC, the orange line represents the heater activity and the black line shows the deviation of the temp from the target.

Larger view over 104 minutes of idling with one espresso shot at minute 57:

For those with a similar setup, I enclose what I think are all the menu settings one needs to do the same.

Colors in the screen are set like this:

With some trial and error I found these PID settings to work best:

Illustration copied from https://en.wikipedia.org/wiki/PID_controller

This is the content of the Device Assignment menu in Artisan:

Extra Devices:

For a while, it was unclear to me how to display the deviation, the difference between target temperature and actual measured temperature. In the current new version of Artisan, this works differently and another formula was needed.

Marko Luther pointed me to the blog about Signals, Symbolic Assignments and the Plotter on the Artisan website and I also found the Help info below in the Extras menu of Artisan:

Trying out the different symbolic variables I arrived at the abs(Y2-Y3) formula which works beautifully again.

Below is my current Serial Ports menu. Mostly this is different with Parity=0 instead of N but if I want to comply with this, I need to dive into the configuration of the Fuji.

Hopefully this info will someday be of help to another user of the Fuji PXG4 in a similar setup.

Thanks to Marko Luther for Artisan!!

6 shots. First one at 15min when the machine wasn't fully warmed up. Cleaning flush at 42mins.

PS 28 september 2016: This PID setting also works well:

PS update 16 Oct 1016:

Switching on the 'autotune' around the operating group temp of 77ºC did not work well recently since the procedure involves having the heating element on full capacity for a certain amount of time and at the given temperature that would result in considerable overpressure for the boiler with a violently hissing overpressure valve.

Resulting PID settings read from Fuji PXG-4 and saved

Following the advice of a friend, I tried the autotune at 45ºC group temp and then went very well. Next, I tested this same PID setting on the internal PID of Artisan and the system worked very well. Then I also tested this at higher temperatures, in steps of 10ºC and finally on the working group temp of 77ºC.

Testing the PID controller in several steps ramping up the group temp

This morning I did a full warmup using the PID and bypassing the pressostat from the start. I opened the steam wand a little when the pressure approached max and the OPV started hissing, closed it again as the PID was lowering the power for the heating element and boiler pressure went down.

After 26 minutes I pulled a delicious espresso and observed the group stabilising neatly.

Warmup and one espresso at 26mins