Should I PID my Heat Exchanger?

[Ken Fox]

Should I PID my Heat Exchanger? This is a question I am asked almost every day, each time I venture outdoors

I was often at a loss to know how to answer this very important question. Although I'd PID'd my current espresso machine (with Jim Schulman and Barry Jarrett's help) early in its life, I'd grown accustomed to living with it over time and not given it much additional thought. I have done some Scace Device datalogging on this, my Cimbali Junior D1 Rotary machine, but I haven't had much "before" data to compare with the "after."

Enter my former primary machine, the Cimbali Junior S1 Pourover Vibe machine which was reposing in my basement, awaiting duty at a moments notice should the rotary fail. I've always had a soft spot in my heart for this machine, my first *real* espresso machine, the one I had special ordered for me 10 years ago (and waited 3 months to receive) because I "needed" to have it in brass, from the now defunct Boise office of the now defunct Portland Oregon Cimbali dealer. But I digress.

I'd entertained the possibility of PID'ing this machine for a couple of years but could never justify it. Recently, I decided I didn't need to justify it, so W-T-F, just do it. I begged Barry J. to make me a special length probe, purchased a Fuji PXR3 from an ebay auction, bought a project box and other minor supplies, and had all the other stuff I needed (an SSR and wiring) already sitting around in the parts drawers.

I should add that this entire project, although really fairly simple, took longer at every turn than I'd expected. Each time I devoted 45 minutes to do some simple thing, it ended up taking 3 hours, most often due to my own silly screwups, such as machining the project box upside down from what I needed in order to have access to the wiring connectors used on the Fuji. I've probably got around 25 hours in this job, but if I had it to do over could probably do another one in a fifth the time. But I don't want to do another one, I'd like to encourage you to consider doing something similar on your heat exchanger machine, by showing what kind of results I've been able to get on the first day with my revitalized machine.



Before starting the PID installation, I did some baseline thermometry with the Scace device and my Fluke datalogger. I intended originally to post some of the "before" graphs but in the interest of time and space I'll simply say the obvious; they were quite variable and limited by the hysteresis of the deadband of the pstat that was used. The machine made a lot of tasty shots in its life as a pstat controlled machine, but it was not practical to control shot temperatures other than by differential flushing, a technique familiar to many HEX machine users. Differential flushing does offer some range in shot temperatures but allows little or no certainty about what one is getting nor any assurance that it can be repeated reliably at a later time. Fiddling around with pstat adjustments when one changes coffee blends would be enough to make most people tear their hair out, so I doubt many do this at all often.

I don't claim to have figured out all the ins and outs of how best to use this PID'd machine, certainly not at this point. I autotuned the PID and did a slight adjustment to the "D" parameter, about doubling it; that's the extent of what I have tried so far. What I have done is some preliminary study with sequential shot and "walk up" shot temp curves, trying to judge control over temperature in both repeated shot series and for individual "walk-up" shots. The graphs below will show that I have not yet defined either the upper or the lower range of shot temperatures which will work; I think I can probably get a useful degree on each side of the graphs I have to show now. Not all flush volumes were tested, either, but 50mls used uniformally seemed to work fairly well.


Anyway, herewith the shot data in graph form:

At the low end, shooting for a shot temperature around 199F, and using a boiler temperature of 239.5F:

Next, a boiler temperature of 241F yields a shot temperature around 201F:

"Walk Up" shot testing at this temperature shows pretty good temperature stability across 4 random shots:

A shot temperature of around 202F is produced with a boiler temperature around 242.5F:

which has decent "walk up" shot temperature stability:

A boiler temperature of 245F produces a little bit hotter shot in the range of 203F:

Clearly, the result of pushing up the boiler temperature is non-linear as it relates to shot temperature; further work would be needed to fully define the usable shot temperature range derived from the boiler temperature setting.

Several points to make about these results and then I'll let the data speak for itself. The shot frequency on shot series was that each shot was followed by 2 minute idle period. This is different than I have done before with results from my PID'd rotary machine. The interval was determined empirically when I found that shot temps declined a little sequentially with a shorter interval, in spite of the boiler temp increasing as shots were pulled. This machine was sold as a low volume catering machine; it is, after all, a pourover. The rotary machine appears to have faster thermal transfer between the boiler and the HEX; I know not why, but there were 7 years separating the manufacture of the two machines and presumably Cimbali made some internal design and or materials changes.

This slower recovery in the HEX has no bearing on ability to froth; with the PID the machine froths as well and recovers as quickly as the rotary machine, just the temps in the HEX lag a little. Every machine design is going to respond differently to electronic temperature control, and only by experimenting will you discover what works best for your particular piece of equipment.

If you are considering modifying your HEX machine by putting in a electronic temperature control with a PID, I hope this post will encourage you to do so. Every machine is going to respond to PID'ing differently, but I think it will be a rare HEX machine that does not have better inter and intrashot temperature consistency after the modification, plus you will be able to actually control the temperature of your straight shots with not a lot less accuracy than some of the mega expensive double boilers that are either now being sold or will soon be released on the market. When the occasion calls for frothing, just push up the boiler temp and froth away; temperature stability of shots going into milk drinks is not all that important, at least in my opinion.


ken
(this is crossposted on alt.coffee)

[Ken Fox]

Much is made of shot temperature measurements made in repetitive shot series; this is what the WBC Water Brewing Temperature Test consists of. But any well designed espresso machine will reach some sort of temperature equilibrium after repeated shots, and if the interval between shots is within the duty cycle of the machine, the curves will look good. How can we use this in our daily low volume home based espresso making? I dunno, except if you have a bunch of espresso loving friends over for a party.

Most of the time, low volume setting users (home users this means YOU) have to deal with a much more difficult situation; pulling shots on a machine that has been sitting at idle for minutes or hours since the last shot was pulled. Depending on the machine, that first shot after idle is apt to be too cold or too hot compared to what one would get with more regular shotmaking. But the first shot is the only shot a home user is for sure going to drink:-)

I have a bias about espresso shot temperatures; it is that sub degree differences in shots are beyond the tasting ability of most tasters, even for most people with high standards for espresso. I think the spectrum of espresso brewing temperatures can be broken down into maybe 3 large "chunks;" for high temperature blends, low temperature blends, and those in between. In this context you can substitute "SO" or Single Origin for "blends" if this is what you drink. I decided to try to come up with a way to get these three distinct temperature "bands" on a random walk up basis with my newly PID'd 10 year old Cimbali Junior S Pourover.

After some prior experimentation I'd more or less determined the boiler temperature range needed for consideration. And of course, this machine has its own shape of shot temperature curves which are very much humped and definitely not flat lined. The machine is what it is, it produces very tasty shots, and it remains to blind tasting to determine, for sure, what is the ideal curve shape for espresso shots. As to the flush, the lower the boiler temperature the easier it is to flush an accurate repeatable amount, and the less you will have to flush. I chose my flush volume measuring device with care, and ended up with a plastic 1 cup measuring cup from Walmart, that cost approximately $1.65 USD. Such measurements are not entirely accurate, plus this is a semi auto machine; I'd put the really obtained flush volumes as about 45-60mls. Notwithstanding this, I was able to get the results below.

With much fanfare, I present to you the results of my testing; 3 distinct curve sets based on 3 boiler temperatures that I think will produce good shots with basically all the good blends and single origins out there, whether they like high temps, low temps, or temps in between.

Have a lookie:

For low temperature blends, using in this case a PID'd boiler temperature of 236F:

For Intermediate temperature blends, using a PID'd boiler temperature of 240F:

For Heat loving blends and Single Origins, using a PID'd boiler temperature of 243.5F:

The machine that is producing this is now 10 years old and was built long before electronic temperature control of boilers was even considered desirable. This is a retrofit. If I can do this on my 10 year old HEX machine, you can probably do it also.

ken

[Marv]

Hi Ken,

I just joined the forums.

My pressure stat is shot after less then one year. I want to PID my Cimbali Junior D1.

Where can I get a thermocouple that I can screw into the boiler? I can get the fuji pid and ssr from espresso parts NW.

thank you for any help.

Marvin

[Ken Fox]

Hi Ken,

I just joined the forums.

My pressure stat is shot after less then one year. I want to PID my Cimbali Junior D1.

Where can I get a thermocouple that I can screw into the boiler? I can get the fuji pid and ssr from espresso parts NW.

thank you for any help.

Marvin

Edmonton; never been there but I used to go to Jasper every year. What a great place! Now that I live in the Idaho Rockies (now being the last 13 years) I've had less motivation to get up your way, although I do always make my yearly one week pilgrimage to Vancouver. In fact, a GREAT restauranteur and friend in Vancouver (John Bishop of Bishops Restaurant, Vancouver) was the guy who told me to buy my first Cimbali, my (now 10 year old and now PID'd "S" pourover) which I still own. But I digress

My TC was custom made for me by Barry Jarrett, but you could get one from Omega. The boiler port I used was a spare port not being used for anything else, with a 1/8" BSPP fitting. Here is a picture of where I put the thermocouple:

My TC is a "K" type, although a "T" type might be slightly better. About 1" length for an ensheathed probe is about right in this location, and will keep you away from the boiler walls and the HEX itself, both of which could perturb the temperature measurements.

My recollection is that it is a 1/8" BSPP fitting but you should confirm that. BSPP is a British type fitting that is commonly found in espresso machines (do not ask me why). It stands, I believe, for "British Standard Parallel Pipe." This differs from your standard US/Canadian pipe thread. IF you are very handy you could take the original plug out of the boiler, drill a hole in it, put in SS sheathing, and then put the probe in there. But I'm not handy enough to do that so I won't try to tell you how to do it.

As for the rest of the installation, you need a box for the controller that you can (hopefully) mount on the machine itself. The most functional place to mount the controller in a box is on the underside of the machine. I have mine mounted like this:

The small aluminum project box I used, which is only slightly larger than the fuji controller it contains, is mounted on the underside just to the left of the on off switch on the underside. The box I used was obtained from mouser.com but there are other sources. There is also an SSR to mount. I put it on the underside of the machine also, further back, and got a plastic OEM cover for the relay to help protect from electrical risk. My relay is a Crydom and mouser sells the plastic covers for around three US$. The underside of the case is a good heat sink, but you will need to sand off the black paint and to use thermal paste when you mount the SSR.

I bought some short metal screws from Home Depot that I used to screw in the small controller project box and the SSR from beneath. OBVIOUSLY, you need to inspect exactly where your drill holes are going to come through, or you could do some real damage like destroy your brain board or cut through plumbing. This is not a job to do without being very careful and attentive to the risk of doing real damage. I was sweating bullets while I did the install. Even more obviously, the machine needs to be unplugged from the wall when you install the components. We don't want to read about you in the paper afterwards:-)

Remember that water and electricity coexist in espresso machines. Stuff like this has to be mounted considering the risk of exposure to water. I have had one drip tray overflow since I installed the PID on this machine and am very glad that I considered this potential problem before doing the install. The last thing you want to do is to get electrocuted or to risk having all your work get ruined by a water overflow accident. So, consider very carefully where you put stuff and take advantage of the waterproofing that Cimbali has already put into this machine, by mounting stuff under the waterproofed undertray compartment, for example.

I would suggest retaining the old pstat as a safety if this is feasible. It will work fine whatever the deadband as long as it still functions. I have my pstat set to turn off at about 1.6 bar so it never actually has any effect in normal, PID driven, temperature control. It would kick in if the SSR got stuck in the on position, for example.

Good luck.

ken

[Ken Fox]

Hi Ken,

I just joined the forums.

My pressure stat is shot after less then one year. I want to PID my Cimbali Junior D1.

Where can I get a thermocouple that I can screw into the boiler? I can get the fuji pid and ssr from espresso parts NW.

thank you for any help.

Marvin

If you haven't already bought a controller and an SSR, there is an ebay seller who is blowing out the Fuji PXR3 for $90 US delivered and also good Crydom SSRs for $13 (I don't know what the shipping to Canada would cost but since they use the post office I'd bet the shipping cost is only slightly more). I have this exact combination of controller and SSR installed:

http://cgi.ebay.com/ws/eBayISAPI....em&item=7588586166

and

http://cgi.ebay.com/ws/eBayISAPI....em&item=7588585216

If you want to install the controller as I have done, on the front underside of the machine, http://www.mouser.com has a very good selection of metal project boxes, and they also have thermal paste in packets that you can use to put under the plate on the SSR. They will also sell you a plastic cover for the crydom relay for two or three dollars. They shipped my order via the postal service also so I'd imagine it won't cost much to get this stuff up to Canada if you don't mind buying ex USA.

Finally, Barry Jarrett recommends ventilating the box that contains the controller, because he has seen temperature oscillations in PID installations when he did not do this. You would need to drill some small holes on each side of the project box to let air pass through.

ken

[Marv]

Thank you very much!

Yes, the mountains up here are very nice. I try and get out there a few times a year - not nearly as much as I would like!

Now I am on the hunt for a 1 inch ensheathed probe. I will let you know how it goes.

Marvin

[Ken Fox]

Thank you very much!

Yes, the mountains up here are very nice. I try and get out there a few times a year - not nearly as much as I would like!

Now I am on the hunt for a 1 inch ensheathed probe. I will let you know how it goes.

Marvin

One inch in length will work, but if you want to be more precise this is what I'd do. Remove the bolt from the side of the boiler (of course this assumes the boiler is drained down below this point and the machine is cold and unplugged), then stick something like a chop stick in to measure the distance until you run into the heat exchanger. What you want is for the probe tip, ensheathed in the SS, to be about halfway in between the boiler wall where you enter and the heat exchanger, so that the tip is not perturbed much by either one. It could well be that 2" is more like what you want, although 1" would certainly work.

ken
p.s. If I was going to order it from Omega, and if the price was about the same, I'd get a "T" type probe since they are more accurate in the range of boiling water temps. Be sure to program the controller so that it knows that the input is from a T type probe, however, as they are usually set up originally for K-type probes.

[blu]

My recollection is that it is a 1/8" BSPP fitting but you should confirm that.

hello from southgermany,

I would like to pid my junior (new modell) and found your great instructions and comments - thanks!
I just ordererd a few parts but the thermocouple (want to use the same position as you mentioned) is still missing. the reason is, that I do not know, what kind of fitting to order. is there anyone who can verify the 1/8" BSPP?

thx a lot,
sorry my english is not quite good ,

blu

ps might be interesting for some cimbali junior owners (new model - water leak solution for exp. valve):

http://www.kaffee-netz.de/board/v...hp?t=9364&start=10
http://popovic.info/download/caffe/junior_exp.avi

[Ken Fox]

I just checked; it IS a 1/8" BSPP fitting (British Standard Parallel Pipe). You should be able to get one of those fairly easily out of the UK. The place in the States where I got mine appears to be out of business now.

ken
p.s. if I had it to do over again I'd consider getting a probe long enough to get it to go right up against the submerged bottom of the heat exchanger. This way it is going to get a very good idea of the temperature of the water that is being used to make your drinks. I don't know if the difference would be measurable but I can't see how it would be any worse. In order to do this properly, though, you should measure the length approximately going in straight through the port where the probe will go, then estimate how much extra you need to bend the probe and get it in this position. You could always put in an extra bend or two if the probe is too long. When you install it you will need to remove the heat exchanger, then install the probe, and then bend it through the hole of the heat exchanger. A bright narrow beam flashlight would be invaluable. You might need some instruments like some long beaked small pliers to bend the probe in order to pull this off!

[erics]

I have used fittings from these guys to adapt various thermocouples and other probes:

http://www.swagelok.com/

http://www.swagelok.com/downloads...s/EN/MS-01-140.pdf - see page 15

The ISO parallel thread is identical to BSPP. Basically what you want is a male adaptor, "bored-through", and sealed with a copper washer to accept the thermocouple of your choice (preferably type T). Follow Ken's advice on length but be sure and take into account the additional length that may be necessary because of any adaptor you screw into the female 1/8" BSPP threads.

Become good friends with your local Swagelok distributor - it makes all the difference.

If you do end up using a Swagelok fitting, be sure and use nylon ferrules vice the metal ferrules that the adaptor will come standard with.

Eric S.

[expressotaglong]

I'll second erics take on Swagelok stuff. Their small diameter tube fittings are good for gripping stainless sheathed thermocouples. Another good source for BSPP fittings and adapters is Parker Hannifin. The Tube Fittings Division catalog is a good place to start looking for European fittings and adapters. This catalog has a wealth of information on almost any threaded pipe fitting you will find in an espresso maker.
http://www.parker.com/tfd/cat/pdffiles/open.pdf

BSPP is very common on European industrial equipment. The threads and how the fitting seals is unlike anything else in common usage here in North America; you will not find them in any hardware store. In my experience European equipment can feature a mishmash of fitting types. I once worked on Swiss die casting machine which, in addition to the usual metric threaded stuff, featured 2 varieties of British pipe threads, two different SAE fittings, and one oddball (non-ISO) DIN fitting. If you are doing some machine plumbing modifications you may want to look at the Port Details section in the aforementioned Parker catalog. Case in point, the tapered pipe fittings you see in your machine are probably BSPT, not NPT. I've seen people cram the two together but I'd avoid doing so with high temperature pressurized water. The good news is that as Parker swallows up manufacturers worldwide they have become a one stop shop for European spec fittings and valves.

Whether you get your fittings from Swagelock, Parker, or whomever, be prepared to pay an inordinate amount compared to your local hardware store. These are high quality industrial fittings, the same ones your machine was built with, and their cost is part of the reason why good espresso machines cost as much as they do.

[jasonmolinari]

I'm trying to follow what is going on here, thinking about putting a PID on my Isomac Tea.

Ken, clarify a few items for me.

1) If you have the boiler temp at 236 or 241, wouldn't the water in the heat exchanger be overheated, just like it would with the pressure stat? If so, i assume flushing is still necessary.

2) Would it work to set the PID temp the the desired brew temp, and purposely let the heat exchanger water (the 3 or 4 or 5 ounces) equilibrate to this temperature? Giving you optimal set point temp for the first 3-5 ounces, and then requiring the machine to sit idle to let the next "charge" of water in the HX equilibrate?

Doing this obviously would eliminate the steaming potential, making the HX machine one big ole single boiler machine, but PID controlled. Optionally, a 2nd set point on the PID could be set above this, to allow steaming.

Is what i'm asking clear?

jason

[eastpresso]

Hi Ken,

would you mind to share where you installed the probe on your pourover machine? There are no unused ports on the old Junior so the most likely location (without any modification) would be the drain plug?

Thanks in advance!

[Ken Fox]

I removed the pstat and in its place, through the port that it used, is where the TC resides in my pourover machine:



You do "lose" a bit of redundancy by doing this, and theoretically you could get the rare occurrence of the SSR breaking and freezing in the "on" position, in which case you risk having your boiler safety valve blow.

Other than drilling an extra hole in the boiler, or (possibly) figuring out how to "T" in the probe through the port used by the pstat, I can think of no other easy access point for a probe in this machine.

ken

[eastpresso]

Thanks a lot for the picture and the explanation.

The pstat port is fairly high up - looking at the picture (and the tubing) it seems indeed boiler pressure is measured by hot water pressure and not steam? The difference in boiler level (due to manual fill) is close to 3cm. It is difficult to tell whether the lowest position of the pstat port corresponds to the min level on the sight glass, although I would assume that this would be the case? Did you bend the probe in addition? What length did you use?

Any thoughts on using the drain plug (other than that is a lousy position to get to and that it's used for a different purpose too)?

[Ken Fox]

You are right that the Pstat port is high and that a TC probe put there (unless bent) will be in the steam column rather than the hot water unless you have overfilled your boiler. This is of no concern as the steam temperature is in equilibrium with the water temperature; it can be no other way. I doubt there are any response time issues of consequence. My probe goes in straight. The probe in my rotary machine is also in the steam column, by the way.

I don't remember the exact length of my probe; what I would do in your situation would be to stick a long straight wooden stick (such as those used to skewer food for barbecuing on the grill) to measure the amount of room before you run into something, with that "something" undoubtedly being the heat exchanger, then make it a little less long than that. I believe my probe in that machine is about 3" long but that is from memory a long time ago and I'm way less than certain on that.

As to the idea of using the boiler drain port for a TC Probe, I can't think of very many worse locations. A PID is not necessary for the operation of your espresso machine, but the ability to drain the boiler MIGHT be, should you need to change the element or get into the boiler for any other reason. You will probably be unable to put anything other than an extremely short probe in there for mechanical reasons, which may give you readings little better than a surface probe taped to the boiler. In addition the probe will probably bang into the element on the way in, and would interfere with removing the element should it fail or its removal be necessary for some other reason. I would not under any circumstances consider the boiler plug as a location for the probe if it were my machine.


ken

[cafeIKE]

A 0.125" probe has a time constant of about 10 seconds in air, 0.33 seconds in water.
A 0.0625" probe has a time constant of about 4 seconds in air, 0.23 seconds in water.

As HX boiler PID tuning functions over a very wide range, and the boiler thermal inertia is measured in minutes, location in the steam area will work fine. Ken is spot on asserting that existing functionality should be maintained.