Hello to everybody:

Please excuse me for double posting, as this was also posted on CG. Unfortunately I have not had any responses on CG.

My name is Ron and would appreciate as much help and info as you can spare. I have an ASTRA Gourmet, plumbed in version, and am thinking of installing a PID controller.

In addition to PIDing the machine I 'm also going to extend the Thermosyphon tubing length between the output of the Heat Exchanger and the input to the group. I've already received the new tubing parts from ASTRA as this will be phase 1 of the project. The objective here, before installing the PID, is to adjust the PSTAT such that the brewing temperature of the water is as close to the desired temperature (200°) as is possible. The hope is that the boiler pressure does not drop below 1.0 to 1.1 bars, which is what I need for ideal steaming. In the event the boiler pressure ends up to low, I can always lengthen the tubing as needed and readjust the PSTAT.

I'm going to use a home built Scace type of device to measure the brew water temperature. With this device the calibration of the system should be assured. After PID installation, the various PID adjustment parameters will be set to allow as close to 200° brewing temperature as possible. The PID will be wired in series with the existing PSTAT. At this point I will need to readjust the PSTAT to approximately 10% higher than the set point of the PID. The PSTAT will then be used to control the steaming, as well as act as an over limit safety device, while the PID will be used to adjust brewing temperature.

I plan on using an OMEGA sheathed T type thermocouple. Now comes the question. Where is the best place to install the thermocouple? My first choice would be to install it as close to the group head as is possible. I can simply add a T-fitting and install the thermocouple (via and Omegalock compression fitting) on the branch side. The T-fitting would be located in line just before it enters the group. My thinking here is that the water temperature at this junction will be as close as is possible to what I'm looking for. I understand that drilling a hole into the group, as did J. Scace in his ASTRA, is an option. However, I really don't want to drill and deface the machine. Again the question begs, are there any other locations which you feel would be better?

Any advice or comments would be greatly appreciated.


Thanks

Ron

ronace wrote:My thinking here is that the water temperature at this junction will be as close as is possible to what I'm looking for. I understand that drilling a hole into the group, as did J. Scace in his ASTRA, is an option. However, I really don't want to drill and deface the machine. Again the question begs, are there any other locations which you feel would be better?

For reference, below are a couple photos of Greg's Astra modification:


External front view


Probe exit above dispersion screen

The probe placement above would track closely to the actual brew temperature during the extraction, but only pick up the temperature of the themosyphon stream indirectly. For further reference, the thermosyphon's effect on temperature is readily detected by Eric's E61 thermocouple adapter:


E61 adapter with 1/16" sheathed thermocouple

The Astra grouphead doesn't have the E61 Allen screw above for easy probe placement; maybe tap one from the backside of the grouphead for monitoring brew temperature?

The thermocouples above are used to monitor brew temperature, not control the steam boiler. For that purpose, a probe inside the boiler makes more sense. Ken detailed his installation in Should I PID my Heat Exchanger? His Cimbali had an unused boiler port that simplified the installation; I don't know how convenient it would be for your Astra. A surface mounted thermocouple for the boiler's PID controller would probably work too, but it would not be as responsive. No doubt Ken has tried both options and knows whether the in-boiler thermocouple is worth the extra effort.


See thermocouple in upper right section of boiler

Here is an alternative (emphasis on "an") to doing any drilling and/or tapping:



You would be measuring temperatures (close) of the thermosyphon loop into and out of the grouphead. You could also wire the thermocouples together if you wanted and get an average signal - lots of power industry applications do this. This would give you an indication of the success of your mods.

Put the PID on the back burner - it only adds convenience and quietness - but if that is what you are after, then move it to the middle burner. Most PID's, even with relay output would have a rough time with the Astra's heater current.

As regards your "homemade" Scace Device, details/pics would be interesting - it is certainly not a trivial exercise.

I assume you have done a search on Google Groups (Alt.Coffee, author = Greg Scace) which would detail his successful adventures with his Astra.

HB wrote:The thermocouples above are used to monitor brew temperature, not control the steam boiler. For that purpose, a probe inside the boiler makes more sense. Ken detailed his installation in Should I PID my Heat Exchanger? His Cimbali had an unused boiler port that simplified the installation; I don't know how convenient it would be for your Astra. A surface mounted thermocouple for the boiler's PID controller would probably work too, but it would not be as responsive. No doubt Ken has tried both options and knows whether the in-boiler thermocouple is worth the extra effort.

Actually, I have no experience with surface mounting of thermocouples, unless you want to count my (for fun) recent on-the-fly brewhead monitoring of my old vibe Junior



Even this probe is internal, at least half a cm. inside the grouphead in a hole that was intended for some sort of mounting screw in another Cimbali model.

I would not personally use an externally mounted TC for temperature control. For one reason, it is simply something with which I have no experience and my track record at "guessing" how something out of my range of experience will work, is honestly "not good." In addition, there is a small element of risk. If for some reason the probe became detached and started to monitor temps of say, the inside of the espresso machine rather than the outside of the boiler, you might end up cranking up the boiler temp to the point of blowing the safety valve, assuming that (hopefully) it was functional.

ken

HB wrote:The thermocouples above are used to monitor brew temperature, not control the steam boiler. For that purpose, a probe inside the boiler makes more sense.

OK, but if the thermocouple is inside the boiler, as opposed to the entrance to the group, wouldn't this introduce excessive lag time between the brewing temperature water and the PID?

Ron

ronace wrote:OK, but if the thermocouple is inside the boiler, as opposed to the entrance to the group, wouldn't this introduce excessive lag time between the brewing temperature water and the PID?

The precision of the PID controller's ability to maintain temperature is related to the quality of the input data it receives. The "I" and "D" parameters try to compensate for delay in reaction times by attenuating the "on" time, but ideally the input reading detects the response without significant delay (see Good all-around explanation of PID stuff). I assume a pressure transducer would be even better for measuring the boiler pressure / boiler temperature, but they're very pricey.

I'm no PID expert, but I know these small details make a difference. For example, I juryrigged a PID installation on an Isomac Amica. It tended to overshoot the target temperature in successive shots. Years later Jeff and I evaluated the Quickmill Alexia with PID controller, which has a kit install by Jim Gallt. His PID'd Alexia never overshot the brew temperature. I asked Jim about it and he said the key was the thermocouple placement. It has to be close enough to where the temperature changes to react quickly, but not too close to avoid over-reacting.

ronace wrote:OK, but if the thermocouple is inside the boiler, as opposed to the entrance to the group, wouldn't this introduce excessive lag time between the brewing temperature water and the PID?

Ron

A PID temperature controller, or any controller for that matter, is not going to work well if what it is controlling is far removed from what it is measuring. The "what" it is controlling is of course the boiler element.

There are both practical and safety concerns. Presumably, the group is going to heat up with a significant lag as the boiler heats up. This means that any simple controller is going to continue "telling" the boiler element to heat up the boiler water, when in fact the boiler water is already hot "enough," and what is really needed is more time for the heat to conduct and convect towards the group. The result will be huge overshoot as the boiler heats up and needless oscillation in boiler temperatures before the boiler temperature stabilizes; this assumes that in the process you don't blow the safety valve, which is surely going to be unpleasant. Even after the boiler temperature stabilizes, how such a system (with a remote probe controlling the boiler element) would function is unknown and would vary by machine.

PIDing a boiler is a relatively simple procedure that eliminates or reduces the hysteresis in temperatures that a boiler would otherwise have with a p-stat. There is considerable experience with this approach, both in consumer and professional espresso, and also in industry where PID controllers are most often used. You can take advantage of this experience and know, in advance, that it is going to work out, as you are not asking the system to do anything other than refine the functionality of what is already in there (a pressurestat, presumably).

I'm not saying that a more complicated type of controller, or a controller with more than one thermocouple couldn't solve this problem (and in the process end up with better and finer temperature control), however in this case we are talking about a much more complex system than what you will get with a single TC and an off the shelf one channel PID.

As an aside, I have thought about the possibility of getting longer ensheathed TC probes for my Cimbali Juniors, in order to get them as close as possible to the actual heat exchangers contained in the boiler. If I ever test that, the idea would be that the system would become more responsive since it would (by simple positioning) be effected somewhat more by what is going on in the HX while still being sensitive to boiler temperature. Exactly how this would pan out in real life I don't know, but it suggests one way of biasing the system a bit towards the actual brew water temperature while still (hopefully) retaining a safety margin. I am however finding that my current system allows quite a bit of accuracy and reproducibility of shot temperatures, so any additional improvement would be small and probably not worth the effort involved.

ken

Ken Fox wrote:There are both practical and safety concerns. Presumably, the group is going to heat up with a significant lag as the boiler heats up. This means that any simple controller is going to continue "telling" the boiler element to heat up the boiler water, when in fact the boiler water is already hot "enough," and what is really needed is more time for the heat to conduct and convect towards the group. The result will be huge overshoot as the boiler heats up and needless oscillation in boiler temperatures before the boiler temperature stabilizes; this assumes that in the process you don't blow the safety valve, which is surely going to be unpleasant. Even after the boiler temperature stabilizes, how such a system (with a remote probe controlling the boiler element) would function is unknown and would vary by machine.

Exactly right. This was also my concern when Ron contacted me a few weeks back.

But I think this overheat problem can be mitigated by keeping the old pstat in place, in series with the PID system's SSR. Setting the pstat slightly "above" the temp/pressure where the system ultimately stabilizes, it should make things work just like a non-PID'd boiler until it gets very close to steady state.

From there, the PID can take over for fine tuning the idling temperature. And if the pstat is set an adequate increment above steady state pressure, it shouldn't have much to do most of the time.

I don't know for sure if this will work or not, but I suspect it will.

Jim

I will be watching and learning. As I get more into my Ms Carmen (named by Yulie) I will be doing mods. I had thought the PID one not needed as it has gages already? Why are you PIDing?
treshell

jggall01 wrote:Exactly right. This was also my concern when Ron contacted me a few weeks back.

But I think this overheat problem can be mitigated by keeping the old pstat in place, in series with the PID system's SSR. Setting the pstat slightly "above" the temp/pressure where the system ultimately stabilizes, it should make things work just like a non-PID'd boiler until it gets very close to steady state.

From there, the PID can take over for fine tuning the idling temperature. And if the pstat is set an adequate increment above steady state pressure, it shouldn't have much to do most of the time.

I don't know for sure if this will work or not, but I suspect it will.

Jim

I think this might work out if the individual either wants to tinker with the pstat each time he changes the PID setting, or if he wants a constant and unchanging shot temperature. One of the advantages of a PID, however, is that one can change boiler set points rather easily by simply pushing buttons on the PID controller, as opposed to having to take of the case of the espresso machine and adjust the pstat with a screwdriver.

Using the pstat as a safety is all well and good, but if the pstat temp is too close to the PID temp then you will in essence have something 1/2 way between pstat and PID control as the temp oscillates above and below, or if the pstat temp is too far away than it won't really damp the oscillations.

How well this would work in practice is going to be heavily machine-design-dependent. With my Cimbalis, which don't have a thermosyphon, I think the odds are better that it would work than in an e61 design. But the only way to really know this would be to test it. In order to be worth the effort, one should first find out how well one will do without reinventing the wheel. If a basic boiler PID setup combined with a regular flushing regimen gets you very close to what you want, then the diminishing returns of improving on it may not be worth bothering with.

ken