I finally found the time to continue to work on this project. And it looks like this time it works!
Still learning how to handle the PID and have some questions: the OUT / AT light almost blinks continuously, I think that means that the PID is switching the heating elements in the grouphead on/off. Is it normal that the on/off interval is one second apart and never stays off? The light goes out when I set the PID to a cooler temperature (which sounds logical).

Not a question, but a remark: I also use an external type K probe attached on the outside of the grouphead (readings are shown on the green Mastech device in the picture). While the PID holds the target temp very accurately, the type K probe measurings fluctuate more ( 1 - 2 degrees Celsius)

BartG wrote:Still learning how to handle the PID and have some questions: the OUT / AT light almost blinks continuously, I think that means that the PID is switching the heating elements in the grouphead on/off. Is it normal that the on/off interval is one second apart and never stays off?
Not a question, but a remark: I also use an external type K probe attached on the outside of the grouphead (readings are shown on the green Mastech device in the picture). While the PID holds the target temp very accurately, the type K probe measurings fluctuate more ( 1 - 2 degrees Celsius)

I didn't follow closely the latest posts here, so pls ignore any redundancy here:

the PID should have a cycle parameter, set at lower values, say 2s for SSR, and at higher intervals, 20s for direct relay (not the case); I do not know the PID you are using, but AT light if meaning autotune, should go off;

as I understood theese, the heating elements, are placed above in the group, outside the 'working area' (lower part of the cylinder), hence, the thermostat placed there, or the PID in our case, are not meant primarily, to react to the temperature changes in the group (actually, they can't, given the group thermal mass and those 100W power), but to dissipate a rather constant amount of heat 'flow rate';
by PID-ing the group one cannot control the group behavior during the extraction, but the temperature in the idle status, which combined with the HX steam and water, results in a, hopefully, desired effect!
and for the consistency of the shots it is important that the amount of steam and water, hence the heat, comming from HX is - for similar procedures - the same (providing there is a complete HX recovery between shots, and the same boiler pressure);
in this way, the parameter under control with PID - idle status group temperature - could be used to:
- eliminate completely the flushes
- set a certain thermal regime for the extraction; as discussed earlier, seems convenient to consider a cooler shot for PID=85-87C, a medium one for 89-90C, and a hotter one for 93-94C;

I am using also a K type T/C placed on the lower front wall of the cylinder, and the maximum temperature during the shot is showed at the beginning, usually at the end of preinfusion; then follows a descending slope of cca. 3C;

the temperature measured on the cylinder wall is in idle, rather constant; perhaps you should identify any cause for variance in heat dissipation during idle, as ambient temperature, insertion of a colder PF etc.

for ex., this is the temperature profile, during a warm-up with the PID set for autotune:

and this is a session log, to notice the PF insertion effect, at the beginning:

To start PIDing 1 need to understand where their disturbances are coming from.For any espresso machine the obvious disturbance is the group head temperature and for any heat exchanger design, whether if it is the typical industrial heat exchanger or espresso machine heat exchanger, the temperature of the brew water will be influence by both the flow rate of the brew water and the heat source which is your heating element or boiler water temperature. To make it simple the faster your water going out, the lower your temperature of water because water dont have enough dwell time in the hx to get heated up 2 the desired setpoint. Unless you increase your heat flow rate to the boiler.so pid the heat exchanger, set your process variable at outlet of the heat exchanger by putting a temperature probe there and set your control variable as your heating element. if possible, do a master and slave pid control system where you pid your heat exchanger outlet to control another pid system which is controlling your boiler temperature using the heating element. This will provide you with accuracy in temperature. Better still, if really feasible, do a feedforward and feedback cascade pid control system by putting a flow rate transmitter at the inlet of the heat exchanger and connect it to the master PID controller . This has the accuracy of cascade PID control system and improves the response speed of the whole system by eliminating the disturbance before the feedback can kick in .

For more info , http://controlguru.com/the-feed-forward-controller/

BartG wrote: Is it normal that the on/off interval is one second apart and never stays off? The light goes out when I set the PID to a cooler temperature (which sounds logical).

In my experience this is normal. I've used two Auber PID's one two different machines and the light on both stayed on during warm-up and then started flashing in about one second intervals. During auto-tune the flashing is faster.

theoleejunming wrote:... the temperature of the brew water will be influence by both the flow rate of the brew water and the heat source which is your heating element or boiler water temperature. To make it simple the faster your water going out, the lower your temperature of water...so pid the heat exchanger, set your process variable at outlet of the heat exchanger by putting a temperature probe there ... do a feedforward and feedback cascade pid control system ...

it is about a rather low cost spring-lever with some unusual concepts (hybrid) like the cylinder fill pump and the group heater;
the pumped water fills the cylinder chamber and makes for a slow, zero bar preinfusion; the espresso extraction is in a second phase, under the pressure of the springs; and with an 100W group heating placed high above into the head, the PID has practically no influence during the 25-40secs brew!

sorry for off-topic, and the lenghty post, but the complex cascade PID-ing seems more appropiate for boilers of a DB pump machine, but still I do not consider as being efficient to control in the real time of the extraction (25-35 secs) an HX machine with E61 group!

and yes, the flow rate of the brew water does influence the temperature of the water above the puck, in an E61 pump machine: overall, there could be a declining profile (cold water injection into the HX), but the initial temperature hump is due to an increased flow rate at the beginning, when water is filling all the chambers in the group; the smaller the flow rate, the smaller the heat aport with the water; with a 0.5-0.6mm jicleur into the mushroom, I had no HX hump at all (Bezzera Magica)...

and with a lever machine, the instant temperature of the brew water entering the cylinder chamber is not relevant, as it is the resulted amount of heat brought by flash boil water entering the cylinder chamber (no flush), combined with the rest of the pumped water, and with the thermal status of the heavy brass group in idle;
BTW the water entered into the chamber, there is a total of 56cc available with the piston up, plus the headspace above the puck, for a total volume of about 70cc, roughly half of the HX chamber volume (Strega has a container shaped HX, not a tube)

it is this thermal status of the grouphead in idle which could be controlled by PID-ing the group;
the temperature of the brew water could be controlled by a pump dimmer; for the average temperature of the brew water into the chamber it does count if the filling of the camber at 0 bar was done in 5 sec (full flow), or in 12 seconds (colder, as heat dissipation is for a longer period, and the PID has no effect on such a short interval with 100W heaters)
however, the thermal status of the GH in idle, before starting the shot, could be also controlled by stopping the pump or, better, by inserting a cold PF;

from my experience, PID-ing the group controls the idle status of the group and this, combined with a consistent - from shot to shot - amount of heat brought from HX, could control the overall temperature profile of the shot;
also, by PID-ing the group, one can work completely without flushes;
in a head-to-head multiple shot session without flushes, there should be an HX recovery (cca. 1:30 min) and head recovery (cca. 2:30 min) allowed intervals between shots;

I also mentioned the dimmer on the pump, and - for me proved to be a first must mod - a pressure gauge on the output flow of the pump: 50$ in my case, the gauge and the capillary, mounted outside and using the existing cross connector (union 7304001TR, page 6) after the OPV (same for all E61 Bezzera machines)

ravco wrote:In my experience this is normal. I've used two Auber PID's one two different machines and the light on both stayed on during warm-up and then started flashing in about one second intervals. During auto-tune the flashing is faster.

Thanks, good to know.

radudanutco wrote: ...
from my experience, PID-ing the group controls the idle status of the group and this, combined with a consistent - from shot to shot - amount of heat brought from HX, could control the overall temperature profile of the shot;
also, by PID-ing the group, one can work completely without flushes;
...

Yes, using the Strega without flushing is one of the goals here. I've stopped flushing since I placed a thermocouple on the outside of the grouphead. With experience a certain temperature (dependent on where the temperature is measured of course) became my reference for certain beans. For lightly roasted beans I tried to get the grouphead 3 to 4 °C higher. For darker roasted beans I let the temp go 3 - 4 °C lower than the reference point. This works well, but it would be nice to be automated. So I installed the PID to stop temperature surfing (switching the machine off/on, or removing portafilter and reinserting a colder one to cool the grouphead down).
The PID itself holds the set temperature very well, but since it measures the temp not in the grouphead itself, but at its neck which is mounted to the case, there is a difference in temperature. If I want the grouphead to warm up or cool down fast I have to do an overshoot on the PID and then correct it to stabilize the temp of the grouphead.

And now I see that I'm temp surfing on the PID to get the grouphead temp where I want it ... not ideal.

BartG wrote:... The PID itself holds the set temperature very well, but since it measures the temp not in the grouphead itself, but at its neck which is mounted to the case, there is a difference in temperature. If I want the grouphead to warm up or cool down fast I have to do an overshoot on the PID and then correct it to stabilize the temp of the grouphead.

And now I see that I'm temp surfing on the PID to get the grouphead temp where I want it ... not ideal.

this is why I consider the thermostat, a good price/performance compromise; even a 12-14C deadband near the heating elements is reflected as a 1-1,4C oscillation on the external wall in the active area (Pt. 1 by erics, where I am logging the group temp)
however, as stated here, the PID is a convenient way to adjust the shot temperature and to eliminate the flushes;
perhaps an on-off controller would do, as the overshooting (larger deadband) does help in decreasing the response time;
as said here, PID-ing the group is a convenient solution for setting the group temperature in idle, that is, after the HX and group recovery interval;
BTW, these intervals between shots are quite tolerant, providing there is a minimum of, say 2-3 minutes as needed for recovery to shot; for ex:


I took a longer pause after first two shots with no logged evidence of influencing the next three shots;the 'HX' sensor is actually on the output pipe near the HX;
the PID=90.5C
sorry for C degrees; and of course, they are only a rough estimate as the calibration was done long time before;

I've been using the PID for almost ten days now and start to like it It does give more accuracy, although the temp surfing is not completely gone. This is caused of course because the RTD sensor is mounted in the far back of the grouphead.
Now I'm wondering: since I already have a type K thermocouple attached externally to the side of the grouphead and because this thermocouple gives me my reference temperature, why not connect this one to the PID instead of the RTD sensor?
The only thing that's holding me back is safety: what if the thermocouple comes loose and just reads room temperature... the PID would just keep on heating the grouphead.
Does anybody know of a way to prevent this kind of issue? Is there a way to program the PID for this?

what I was trying to say is that, IMO, the PID does not have much to do with keeping the temperature steady during the extraction process!
the temperature profile during the extraction is a result of the group temperature in the previously idle status, and the mixture of flash-boil water and the rest of the water pumped through HX; and of course, the duration/flow rate;

the PID can only control the group temperature via those low power heating elements (100W), during the idle status, with a recovery time of, say, about 2-3 minutes, not 5-10 secs response time needed in extraction...

I've worked for several weeks using a K thermocouple mounted in pt. 2, then pt. 1 (as by erics), but the net results were more confusing and definitely for a less consistency!
there is quite a thermal inertia in the group during the brewing, and the PID will react immediately to the temperature changes but with a delayed and less predictible effect than in the case the temperature is kept constant in that neck area and dissipated in a constant heat flow;
for that reason I consider the benefit of PID-ing the group for (1) controlling the extraction temperature (preemtively in the idle thermal status) and, derived from that, (2) working without flushes at all;

radudanutco wrote: ...

I've worked for several weeks using a K thermocouple mounted in pt. 2, then pt. 1 (as by erics), but the net results were more confusing and definitely for a less consistency!
there is quite a thermal inertia in the group during the brewing, and the PID will react immediately to the temperature changes but with a delayed and less predictible effect than in the case the temperature is kept constant in that neck area and dissipated in a constant heat flow;

Hmm, OK, I did not expect that!

radudanutco wrote: for that reason I consider the benefit of PID-ing the group for (1) controlling the extraction temperature (preemtively in the idle thermal status) and, derived from that, (2) working without flushes at all;

Yes, that how I'm also using it

Did you alter the offset of the pid so its temperature reading is the same as your thermocouples (like placed on pt. 1 or pt. 2 from the link to Erics picture)?