PID on a HX...this is my thinking, is it flawed?

[jasonmolinari]

I've been reading on up on Ken's discussion of his Cimbali PID, and i've had a few ideas of my own, which i've sort of, kind of confirmed with another person.

What i'm wondering is would the following work for a HX machine, more specifically, an Isomac Tea.

Honestly, i don't like doing the water dance, i do it, but i don't like it. I'd rather hit my my target brew temp accurately, and more importantly repeatedly. I don't make many frothed drinks, i'd say 90% straight espresso. One might argue that i got the wrong machine, but that is neither here nor there. So this is what i was thinking.

Given that the water dance is necessary because water in the HX tubing is sitting at above boiling point, since the boiler is under pressure, could one replace, or augment the pressurestat with a PID, which takes readings from the boiler water (just like Ken's), but instead of setting the PID to 240 or whatever, you would set it to espresso brew temp. The water now idle in the HX would equilibrate with this boiler water (which is accurate, and repeatable given enough stabilization time between boiler water and HX water, which i think should be less than 4-6 minutes), and would be at espresso brew temp. At least for the 1st 4-6 ounces (or however much is in the HX tubing inside the boiler). This is more than enough for a shot. No dance necessary.

Now adjusting the PID would adjust the boiler temp and therefore espresso brew temp. Installation of a bypass switch would allow the machine to go back to using the stock pressurestat, and allow the boiler to return to steam mode.

The downside to this mod, is the rate at which espresso can be pulled, since after each shot the machine will have to be left idle to equilibrate HX water and boiler temp. Given i'm the only drinker in the house, this isn't that important to me.

Essentially this would be creating an expensive single boiler machine with an E61 group (zaffiro i guess), controlled by PID.

I'm guessing that given current PIDs go for about $40 from Auber, this mod could be done for about $130 including a sheathed TC from Omega, and swagelok fittings.

Ideas? Flaws? Issues?
jason

[jesawdy]

Jason-

I think your idea is somewhat flawed.... and it would make your machine a bit of a bastard as you admit.

First a caveat, I do not have an E-61 heat exchanger machine... I have a PID-controlled Silvia, and a Cimbali HX (no PID), so take what I am saying with a grain of salt. The biggest unknown to me is the volume of the heat exchanger on your Isomac, that may affect what I am about to suggest.

A heat exchanger has a shell and a tube.... on your Tea, the shell is the steam side, and the tube is the brew water side. In order to do any heat "exchanging", a temperature gradient must exist. In an E61 the gradient is created by the hot steam side (superheated water and steam), and the cooler tube side, namely cold water entering the tube from the water source. You also have heat loss to the grouphead via the thermosyphon (which also relies on a temperature gradient) and radiant heat loss from the grouphead.

If you had a perfect heat exchanger (and you could approach one with a long enough coiled tube for the cold side, and/or increased dwell time of the tube side in the shell, and no heat loss from the grouphead) you could set your PID to the desired brew temp and rock and roll. The problem is that you don't have a perfect heat exchanger, you have one that was designed to have a superheated water on the shell side, and a transfer efficiency such that once you run some cold water through the tube side, you approach the desired brew temperature as the water exits the tube within a reasonable period of time.

Yes, when idle, you get superheated water in your heat exchanger tube, and yes your grouphead gets overheated too. But, in continuous operation (at a reasonable shot building time), you get a system that can take 70 degrees F water, and heat it to 200 degrees F, again and again, accounting for losses in the system and the transfer coefficient of the heat exchanger design.

I should think that the tube-side heat exchanger volume is too small to do what you suggest. If you have the boiler set via PID at brew temp, the water at idle in the heat exchanger tube will also be at brew temp, and your grouphead won't be overheated (but perhaps underheated). But as soon as you hit the pump switch, you will introduce cold water to the system and decrease the brew temp. The same thing happens on a PID-controlled Silvia, the difference is that you have a larger volume of water that your are adding the new cold water too. Even so, most people run Silvia at about 220-230F (temp reading of the brass boiler, not the water) to account for grouphead losses and water mixing in the boiler as the shot is pulled.

Now, yes you CAN use a PID-controlled heat exchanger boiler to make it a bit simpler for a machine to come from idle to brew temperature, or to hit a brew temp with less flushing. And yes you can set it very low (and have anemic steam), and get there quickly. But you will always have the boiler temp higher than the desired brew temp to account for heat losses to the grouphead and mixing with cold water.

The PID will also let you change things without going into the machine to fiddle with the pressurestat. If you are desiring some cappas that day, you might bump it up... if you are using a blend that likes a really low temp, you might bump it down. All of this will take some experimentation on your end. For this goal, I don't think the idea of a PID-controlled heat exchanger is a flawed design.... but it's gonna take some experimentation (and perhaps a lot of work fiddling with PID settings and set temperatures) to get there.

[jimoncaffeine]

Jason,

"Now, yes you CAN use a PID-controlled heat exchanger boiler to make it a bit simpler for a machine to come from idle to brew temperature, or to hit a brew temp with less flushing. And yes you can set it very low (and have anemic steam), and get there quickly. But you will always have the boiler temp higher than the desired brew temp to account for heat losses to the grouphead and mixing with cold water.

The PID will also let you change things without going into the machine to fiddle with the pressurestat. If you are desiring some cappas that day, you might bump it up... if you are using a blend that likes a really low temp, you might bump it down. All of this will take some experimentation on your end. For this goal, I don;t hing a the idea of a PID-controlled heat exchanger is a flawed design.... but it's gonna take some experimentation (and perhaps a lot of work fiddling with PID settings and set temperatures) to get there.

I think he is right on the money and I have a pid controlled h/x machine... I use the pid for the sake of the ease of temperature setting, tighter temp control (less overshoot than the Sirai pressurestat that is on the machine is capable of), its quieter, but most of all because it was a fun, cheap and geeky project to do!

I paid about $15 for a used pid(ebay), about $15 for a new ssr(ebay again) and about $12 for a thermocouple. It took an evening of fiddling around with some spare wire, looking for likely looking cast off parts from prior projects and I had it working on my machine. It was pretty easy to do and a lot of fun. Yep, I still need to surf but I don't use quite as much water. If I want to go up or down on boiler temp its just a matter of pushing a button.

Regards,

Jim

[jesawdy]

I paid about $15 for a used pid(ebay), about $15 for a new ssr(ebay again) and about $12 for a thermocouple. It took an evening of fiddling around with some spare wire, looking for likely looking cast off parts from prior projects and I had it working on my machine. It was pretty easy to do and a lot of fun. Yep, I still need to surf but I don't use quite as much water. If I want to go up or down on boiler temp its just a matter of pushing a button.

Jim-

I read your Faema rebuild articles on homeroasters.org with great interest... nice job!

I assume the PID is on your San Marco? That is also a gorgeous machine! Did you document your PID trials as well?... I'd love to hear more. I have a Cimbali M32 that I might consider installing a PID on at some point. Other than Ken Fox's threads and those on an NS Oscar, I haven't read enough to be totally convinced that it is a worthwhile project.

Thanks for sharing.

[jasonmolinari]

Thanks for the replies.
I do see your points Jeff, but given the amount of BOILING water that is required to be flushed, which amounts to at LEAST 4-5 ounces (i think, i'll check tomorrow), i can only assume that amount of coil HX tubing is within the superheated area, and as such would be the amount of water i have to work with if i set boiler temp to brew temp.

As i mentioned throughput is not my issue. If i can only pull 1 espresso per 5 minutes, i'm ok with that.

When you say

But as soon as you hit the pump switch, you will introduce cold water to the system and decrease the brew temp. The same thing happens on a PID-controlled Silvia, the difference is that you have a larger volume of water that your are adding the new cold water too. Even so, most people run Silvia at about 220-230F (temp reading of the brass boiler, not the water) to account for grouphead losses and water mixing in the boiler as the shot is pulled

i don't see this as the same as a silvia. When i hit the brew switch, the water which, in theory, is at brew temp would move through the HX tube to the coffee, to be replaced by cool water. This small amount of new cool water in the HX would have little impact on the boiler water temp, and any effect would be too late to have an impact on the water which is currently on the way to the puck to make espresso.

So, given these caveats:
1) There is more than a few ounces of water in the HX tubing in the boiler water
2) the E61 thermosyphon would still work with boiler set at brew temp, and not cool the HX water too much

The group losses could be made up by bumping the temp a few degrees possibly.

do you still think it wouldn't work?


I appreciate all the long replies!

I actually have many of the parts i need to test this, not a PID, but an on-off controller, i may try installing that just to see if the theory works.

jason

[erics]

Jason -

I too think everything Jeff said is right on the money - a very thorough & well written explanation.

That hefty E-61 group plays a major roll in temperature stability or instability as the case may be and there will always be worthy discussions as to whether or not presenting the puck with a near constant temperature (as measured at the top) produces the better espresso.

The idea of PID'ing a Hx machine, IMO, is catching on fast thanks to the well written and informative postings by Ken F. I can't think of any downside to this adventure other than $. However, you can always crank the pressurestat down and try it yourself. Granted, a pressurestat won't give you the temperature stability of a PID but it will allow you to test out your thinking. The problem with a reduced pressurestat setting is that it also lowers the group temperature and, in your case, may take it below "optimum" brew temp such that the initial flow of water is trying to heat up the group. For someone who makes mixed beverages, the pressurestat should be lowered to the point of having adequate steaming capacity but that can be pretty subjective. In any event, its certainly not as low as you proposed earlier in another post - 236-241 F. If you want those temps in your boiler (not Hx) then you need a setting of a tad less than 0.70 Bar - assuming you are near sea level.

I just guessing here but I believe you could reach a nice compromise with the Tea, at some setting higher than 0.70 Bar, that would eliminate all but a small cursory "cooling flush".

Eric S.

PS - If you filled in your profile , somebody in yourtown, yourstate might very well have a Thermofilter to lend to further calibrate your tastebuds.

[jasonmolinari]

ok, that test failed miserably. Using a on-off temp controller just has way too much hysterisis even to see if the idea would work.
Temp control set at 200, boiler went all the way to boiling!...
need to find a PID to test with

[jesawdy]

Thanks for the replies.
I do see your points Jeff, but given the amount of BOILING water that is required to be flushed, which amounts to at LEAST 4-5 ounces (i think, i'll check tomorrow), i can only assume that amount of coil HX tubing is withing the superheated area, and as such would be the amount of water i have to work with if i set boiler temp to brew temp.

....

So, given these caveats:
1) There is more than a few ounces of water in the HX tubing in the boiler water
2) the E61 thermosyphon would still work with boiler set at brew temp, and not cool the HX water too much

The group losses could be made up by bumping the temp a few degrees possibly.

do you still think it wouldn't work?

Jason-

I don't know. If I had ever had an Isomac heat exchanger apart, I think I could say better whether your assumptions are correct. I'm thinking that, yes you may be getting 4-5 ounces, but how much of that was in the heat exchanger tube, and how much was fresh cold water that just flashed in the boiler/grouphead?

Also, fluid dynamics and pumps are tricky things. I have about forgotten everything I learned in school regarding those two subjects by now, but I think you may have more mixing in the typical espresso machine heat exchanger tube than you think. If it was a simple 1/4" diameter copper coil in a heat exchanger shell, I might agree with you that cold water is just pushing hot water out, but it is a goofy looking test tube shaped reservoir that "dips" into the boiler water.

[jasonmolinari]

Jeff, yep, i think i can see your issues....but that doesn't mean i won't try:) I've also forgotten most of my fluids and heat transfer.

Hrm..another thing i thought about is that there is a run of tube outside the boiler, connecting the boiler HX tube to the group head. The water in there would not be at brew temp...that might be an issue.

[jimoncaffeine]

jesawdy,

Thanks! I had a great time writing the articles and doing the rebuild.

The PID is indeed on the lsm. I have done some documentation and another (single) article will be written.





It will cover:

- PID selection
- SSR selection
- Thermocouple choices (including some ideas for placement.)
- Installation and ideas about how 'permanent' you want it to be (I made mine so I can uninstall, leaving no traces, in about 10 minutes.)

I'm waiting on data collection... Currently I'm using a surface mount thermocouple that I have pinched between the boiler and the large hose clamp that locates the boiler in my lsm. I had initially thought that this was a ridiculously lame idea but I had the tc in my hand, 20 minutes to myself and gave it a try. It is working far better than I could have hoped. In truth, its worked so well that I haven't pursued a direct probe and the elaborate methods I had worked out to get it into the boiler.

The article will reflect the choices I made but will cover the reasoning I used and (as usual) mistakes I made along the way. - I setup the pid to run in parallel with the pressurestat figuring that the surface mount tc would be slow enough reacting that it wouldn't turn on the element if a lot of steam were pulled from the boiler and it had not autofilled yet. In actuality the tc is fast enough reacting that it *does* read the temp change and act on it. Currently the pressurestat is still in place - and functional - but adjusted down to approximately 0.9 bars of pressure. Its a redundancy factor... If the pid dies, I can crank up the tension on the pressurestat and still drink my morning coffee. (priorities! )

Total cost really was about $45 and the pid I went with is a used Barber-Coleman unit. It was a fun way to spend an evening. If you've got to replace a commercial pressurestat it was about half the cost of that. Some theorize that by keeping the unit at a constant temp instead of the roughly 2 degree deadband that the stock pressurestat cycles at (all day long...) it will actually be easier on the machine. I don't know if I agree with that or not. Going from stone cold to 240 degrees and back again each day seems like it could thermally stress the components but I'm not sure about a 2 degree swing...

Comments to adding one or not adding one-
- I have found that I can produce totally acceptable espresso using either the pressurestat or the pid.
- I still do the cooling flush and its become second nature.
- You can easily change temperatures on a whim.
- It's quieter. (No: click - pause - boiling water noise - click cycle going on all day long.)
- If your pressurestat is going out, it could well be cheaper than a new one though probably not as cheap as rebuilding your existing one (assuming its a rebuildable unit.)
- If your anti-depression valve (vacuum breaker, what ever you want to call it) for your boiler stuck closed with a pressurestat machine, pressure would build up prematurely and turn off the element early.
- If the same valve stuck on your pid'ed machine you might trip the safety release valve since it will happily heat until the desired temp is reached, boiler pressure be hanged. (I haven't had this happen yet but it is possible that it could happen.)

In the end, its about enjoying the coffee. It's hard to justify the pid end of things since a tremendous amount of making good coffee is based on the skill of the barista. I am not a tremendously skilled barista, by my standards anyway, but I adapt to my equipment fairly quickly. I can tell by the sound of the cooling flush when its about right and that's good enough for me. It may not be good enough for others and putting a pid on a machine that had a mechanical control will definitely give you tighter control of the temps. With everything setup via the pid and the pressurestat in control, I would see a constant 2 degree fluctuation. If I pulled some water it would jump to a 3 degree overshoot. When I turn down the pressurestat and put the pid in the loop that drops to a 0 degree fluctuation when static and a maximum of a 1 degree overshoot if I pull a lot of water.

Anyway those are some scattered thoughts on it. (The caffeine must be wearing off...)

Regards,

Jim

[espressme]

jesawdy,
Thanks! I had a great time writing the articles and doing the rebuild.
The PID is indeed on the lsm. I have done some documentation and another (single) article will be written.
snip
It will cover:
- PID selection
- SSR selection
- Thermocouple choices (including some ideas for placement.)
- Installation and ideas about how 'permanent' you want it to be (I made mine so I can uninstall, leaving no traces, in about 10 minutes.)
snip
Jim

Hello Jim,
Please post it here! Or, at least a URL to it!
I have a PID and SSR of correct ratings in the mail to me and am looking forward to having at least one machine PID'd. I have the basic idea, but the CG Ham Beach article was lost.
I have a used, dirty, but basically sound 1993 1 group Salvatore commercial HX to overhaul that could really benefit by the PID. Salvatore checked it out!
I look forward to your posting and article!
Sincerely
Richard / espressme

[HB]

Given that the water dance is necessary because water in the HX tubing is sitting at above boiling point, since the boiler is under pressure, could one replace, or augment the pressurestat with a PID, which takes readings from the boiler water (just like Ken's), but instead of setting the PID to 240 or whatever, you would set it to espresso brew temp. The water now idle in the HX would equilibrate with this boiler water (which is accurate, and repeatable given enough stabilization time between boiler water and HX water, which i think should be less than 4-6 minutes), and would be at espresso brew temp. At least for the 1st 4-6 ounces (or however much is in the HX tubing inside the boiler). This is more than enough for a shot. No dance necessary.

Have you read MOSFET's Six Million Dollar Expobar? He "de-HXed" it by adding a PID and insulated preheat coils.

If it was a simple 1/4" diameter copper coil in a heat exchanger shell, I might agree with you that cold water is just pushing hot water out, but it is a goofy looking test tube shaped reservoir that "dips" into the boiler water.

You're thinking about the Cimbali Junior's HX. The Tea's HX is a straight horizontal copper tube (it's pictured in JimP's article Single Boiler vs. Heat Exchanger?).

Back to the original question, I'm interested in hearing if others can reproduce Ken's results on a prosumer E61. When it comes to thermal memory, the Cimbali Junior is an elephant, which plays well to the PID'd steam boiler strategy. I'm not sure the smaller HXs (110ml) of a prosumer E61 and lesser thermal mass will benefit as much. From the machines I've seen, the other extreme of the "thermal memory" range is the Gaggia Achille. My gut feeling is that HX machines whose thermal memory are at the extremes will benefit most from precise control...

Another approach I've been mulling over is to automate the cooling flush. That is, combining an E61 thermocouple adapter, an inexpensive PID, and a SSR controlling the pump. Non-lever type E61s would be good candidates for this modification (e.g., Valentina and the Expobar Pulser).

[coffee_monkey]

Jason,

I have done exactly what you proposed on a very similar machine (Rituale... should be almost same as Tea w/ some minor layout differences) with a very important addition.

What Jeff said pretty summed up the missing link in your proposed setup. To achieve what you want, you basically want to get as close as a "perfect" HX as possible. What I did was to created a pre-heater loop around the boiler and insulate the boiler+loop. The length of the loop was calculated to give me at least 3 oz of pre-heated liquid which is heated close to boiler temp such that the HX is not doing any further heating during pulling a shot. You basically increased the size of your HX so that it has enough capacity to get you thru a shot (commercial HX are much different and have much larger capacity... some also are heated by steam only and will not work w/ this setup).

I then calculated/tested to get a minimum rebound time needed to get to steady state again, and make sure I do not outpace the machine when I use it.

The result is basically a pretty crappy single boiler machine with no ability to steam. Also, there is the issue w/ the vacuum breaker valve. If you set temperature crosses the threshold temperature (210 on my machine), the boiler will go from close to open (and vice versa). It takes more energy to close the valve so whenever this happens, the rebound time will become non-linear and became a bit random. It also will spit some water when it goes from open to close. Normally this is not an issue as it only happened once during initial ramp-up. But if it does that everytime you pull a shot, you will get water all over the inside of the machine (Which is very bad...). I learned this lesson the hard way (valve got stuck while I was at work and steam came up for hours which warped my cabinet above the machine...) and have since re-routed the exit steam to the drip-tray.

Despite the limitations, the machine is quite consistent currently. My wake-up shots are consistent within 2 degrees and no more cooling flushes are needed. In fact, I need to pull a blank to warm up the head before the wakeup shot. I can also converge on steady state much sooner (about 2 to 3 shots) and intrashot stability also improved.

Oh, make sure you put in voltage regulation or replace the vibe pump w/ a rotary pump if you choose to PID the TEA. The voltage drop as result of the PID will cause the pump pressure to pulse at PID interval (1 sec usually) and the clarity of the shot will really suffer...


Ben

[jasonmolinari]

Thanks to all.

Dan, i will check the post on CoffeeGeek when i get a chance. I've also mulled the idea of automating the flush, but can't think of a good way to do it with a levetta machine

Ben, thanks for the details..it sounds like a lot more than i was looking to do, so i might pass on this mod, but it is interesting that someone else had the same idea.

Hrmm..maybe i'll just have to keep dancing until someone comes out with a sub $1000 rotary dual boiler PID machine:)

jason

[hbuchtel]

What Jeff said pretty summed up the missing link in your proposed setup. To achieve what you want, you basically want to get as close as a "perfect" HX as possible. What I did was to created a pre-heater loop around the boiler and insulate the boiler+loop. The length of the loop was calculated to give me at least 3 oz of pre-heated liquid which is heated close to boiler temp such that the HX is not doing any further heating during pulling a shot. You basically increased the size of your HX so that it has enough capacity to get you thru a shot (commercial HX are much different and have much larger capacity... some also are heated by steam only and will not work w/ this setup).
Ben

Ben, after you added the extra length of HX tubing did you notice any effect on your extractions? Maybe a longer time between turning the pump on and first drops of espresso, or a slower flow, something like that?

Thanks,

Henry

[jasonmolinari]

I've read the coffeegeek post mentioned by Dan, and it is similar to what you've done Ben.

I was inside the machine last night, and i don't see much space, or really ANY space to put coils of copper tubing around the boiler, let alone 6 of them as the guy on coffeegeek did!

Not only that, how did you bend the copper tubing cleanly to go around the boiler making good contact with it? And what insulation did you use? I've seen this mentioned a few times, but the only stuff i've seen at home depot is rated to about 150deg. F, which is well below where we'll be.

thanks
jason - still thinking about doing it:)

[coffee_monkey]

Henry,

There was no difference in flow with the pre-heater loop. Upon adding the rotary pump, my built in pre-infusion drop down to about 1sec or less as pressure come up to full almost immediately. I added a needle valve between the machine and the pump which allowed me to adjust the water debit down to 70~90ml/10sec. This make the pre-infusion goes back to the original 6~7. There is a good discussion of this on coffeed right now...

Jason,

You have to clear some spaces around the boiler but there are enough space to get at least 7 loops in. What you do is to start your first loop and put a hose clamp on it. Then do your second loop (or 3rd if you can pull it tight enough) and put another hose clamp on it. Pull on the loop REAL hard/tight before putting the clamp on it. The contact is not perfect but it's as good as I can get it. Another way the I seen the silvia folks did was to prebend the loop into a coil (that is smaller than the diameter of the boiler) and slip it on like a spring. But that won't work very well since there are many ports around the Isomac boiler. This is actually the hardest part of the entire project... changing plumbing is a real major pain in the ass and expect some leaks on your first try... I put a 3 way between the pre-heater loop and the original route. This allow me to bypass the loop and put the machine back to its original config when I need to steam again.

I used the ceramic paper sheets that some ppl used (do a search on CG or HB... I forgot where). Very good insulation but pain in the ass to use as the ceramic fibers are quite sharp and brittle....


Ben

[jasonmolinari]

Interesting Ben, i read up the stuff on your blog, pictures would help for sure, specially in seeing how you wrapped the coils, i can see that being a major pain..

Why install a 3 way? If you just add length to the HX, which is wrapped around the boiler, you could still turn the boiler temp back up to make steam in the boiler, and keep using the extended HX, couldn't you? Obviously your flush would be much much longer...but a 3 way isn't necessary to run in steam mode....

[coffee_monkey]

If you revert back to the normal HX mode, the extended HX will make your flush ridiculously long. Also since there are more cold water to heat now, the recovery time will be excessively long as well. Basically you just make your life that much harder...

Plus, if you go all thru all that much trouble changing the plumbing, a couple more valves/fittings is not going to hurt.

[jasonmolinari]

Agree, adding a couple of valves isn't the problem, but accessing them easily is. Since i'd only go back to HX mode to make steam drinks, i don't care about the flush length.

You mean if you wanted to revert to HX mode to actually use it as it was before, maybe to sell the machine or whatever? Then yes i agree, putting a valve in it is fine, since you wouldn't be switching it every day anyhow.