Adventures with BFC Junior DB TCI. A case study with pictures. - Page 3

Need help with equipment usage or want to share your latest discovery?
tonythewonderful (original poster)
Posts: 92
Joined: 8 years ago

#21: Post by tonythewonderful (original poster) »

As kindly pointed by Eric (erics), this machine has 50K ohm NTC (negative temperature coefficient) thermistors as its boilers' sensors and not thermocouples.

I will add edits to my previous posts to indicate/correct this mistake.

Thank you, Eric :)

tonythewonderful (original poster)
Posts: 92
Joined: 8 years ago

#22: Post by tonythewonderful (original poster) »

A big switch. An adventure where the BCF gets a new boiler.


So, the decision was made. I started looking into how I can get a new boiler. I tried to google custom made boilers, and stainless steel boilers, but all I could get at first were those huge boilers which are some part of a power generator factory/plant, that takes a building and a few hundred people to put it up.




Soon I realised that I do not need to look for a boiler, but rather for some metal shop that would get interested in a project. From there several places were found that were doing different stuff out of stainless steel and I wrote a few e-mails asking if they can make a replica of a boiler, using stainless steel and not the dreadful copper.

After a few unanswered e-mails, one promise to consider, and one "we are too busy now" kind of reply, I have found something very promising, something with a strange name "INVIL": http://www.invil.nz. Their gallery (http://www.invil.nz/Gallery) strongly suggested that I am now looking into the right direction.

I wrote them, and got a very quick confirmation that, yes, they can do it. In a few days, I took the boiler (yes, again!) out of my machine and brought it to INVIL for a final examination. Jack from INVIL is a kind of guy you want around when you start something new. You bring ideas, you discuss, you make a decision, and he makes it happen. Ah, yeah, you need to bring some money too, to power the process on. The estimation was about 10% of the price of the new coffee machine, affordable, but not cheap.

So, since the boiler had to be fabricated all anew, I thought why not make it a bit larger? The machine inside had some extra space. The only thing that was difficult to decide was whether I wanted a boiler made out of a thicker pipe (7.62 mm) in which case it will be heavier, or slightly thinner (5.49 mm) in which case its internal volume will be bigger.

Calculations showed that Schedule 40 pipe (that's 5.49 mm wall thickness) will give me almost 100 mL extra internal volume (~860 mL in total) than a thicker Schedule 80. So, I closed the deal on the bigger volume.

The next thing to decide were the openings on top and the bottom for the thermo sensors and heating element. The first idea was to machine them, however INVIL did not have necessary equipment, and a nearby machine shop was consulted. They quoted a price which almost doubled what Jack estimated for the whole boiler, which was way over what I was prepared to spent.

So, instead Jack suggested a plan B - he could weld in appropriate fittings - 1/4 on the top for PID sensor, 1 and 1/4 for the heating element, and a piece of pipe for the hi-limit thermostat sensor (because the original had a thread size a bit less than 1/4, so the factory workers would not mix them up).
The welding of the fittings worked very well for the money. The only problem was that material shrinks after welding, and the fittings lost their perfect round form. I mean, visually they were absolutely fine - I could not tell any difference at all. However, when I tried to screw in the heating element and the sensor, some extra effort was needed to turn them.



So, here they are, the two boilers for comparison:


By the way, the true volume of the original boiler was only 680 mL (with heating element), and not 0.8 L, as claimed in the machine's specifications.

I do not know, may be they calculated the volume without the element, and added the volume inside the group head and pipes?


At home I have boiled the new boiler and then screwed in the heating element.


After it was installed, and I fired up the machine, the boiler was leaking around the gasket.

Examination showed that the teflon tape that I used, interfered with the gasket, pushing it aside. Jack suggested not to use the teflon tape at all, and just make sure to have the gasket squashed evenly between the heating element and the boiler.
It worked like a charm.




A fur coat for a boiler.

Of course the old insulation would not fit the new boiler, and I started thinking how I can make a new one.




My first idea ended up in a complete disaster. I used some packaging foamy sheets, bubble wrap and baking aluminum foil. Although it looked cool - a kind of a rocket, it was a terrible idea.





It started give off a terrible plastic smell after the machine heated up completely.

First I decided to wait. May be it will disappear on its own, I thought. In a few days it became evident that if I wait any longer my wife will make me disappear. So, the boiler once again was taken out of the machine... Oh dear!!! By that time I was thinking that machine would be a death of me, and was good only as my tomb stone.

Here is what I found:


The inner part of the "insulation" had melted, and stuck to the boiler. It was nasty, sticky, and above all smelly. (think of a smell a burning plastic bag would give out, and you will not be far away).

That all had to go. In the process of scrubbing, the letters INVIL detached, and only their shadow remained. (Sorry about that, Jack).


The question was, what on Earth, I was to do now! I started googling thermal insulations, and it turned out, it was quite difficult to buy some sort of a stand alone thermal pad of a sensible size in New Zealand. The majority of offers were from China, promising 500+ tons of material per month. I visited a couple of car parts places, however, there they could offer only some sort of exhaust tape, which was too narrow, and as far from "food grade" as you can think of.


Finally I had to compromise and improvise. I thought I'd waited enough, and to wait for a couple more weeks for some insulation pad from China was too much...

I bought a "BBQ buddy", which was a sort of a re-usable "aluminum foil", and took some insulation material from under my roof, which is a kind of furry, soft stuff made of plastic (and not from glass wool). The final result was a bit less attractive than I wanted, reminiscent of a Santa Claus in a leather clad Nazi uniform, but by that time I did not care any more - I wanted my coffee!




This worked very well - it does not smell at all, it insulates well - only warm to touch. By the way, as you can see I insulated the group head pipes too (to the extent I could, as the space was getting tighter). Why manufacturers would not do that in the first place?


The moment of truth.

So, now came the moment of truth. Will the new boiler affect the taste of water? Will it perform better than the copper one? Did it worth all the trouble?

I fired up the machine, and once everything was heated and the boiler flushed, I gave it a try. It did not affect the taste/smell of water at all. Remembering my fault with the copper boiler, when a complete thorough cleaning helped only for a day, I was cautious not to jump to conclusions. However, my fears were ungrounded - a few days passed, then a few weeks, and the water from this boiler is as good as it was the very first day. It does not have any unpleasant odour/taste. In fact its taste is indistinguishable from my kitchen kettle.

I do not know if this has anything to do with the water I use, or with my sensitivity to copper. Probably someone will not notice any difference between the two (although I would be very much surprised). Speaking for myself, I am absolutely happy with the results! I can think of no excuse why any manufacturer would still use copper boilers. Can you?

How about coffee? It got better too. Although here I might be biased, and besides I have no opportunity to compare both boilers side by side. At least now, when I am absolutely satisfied with the quality of water coming out of the brew boiler, this question "may be it tastes bad because of the water?" is closed. If I do not get a satisfactory result, it's something else. That turned out to be crucial to know that one of the variables is under control.

I now think, I would buy a "flat pack" coffee machine. There is nothing really to it to put the parts together. If this means I will get stainless steel boilers (and preferably pipes too), better boiler's and pipes insulation (and may be something else, like rotary pump) for the same price...



So, that is the end of the biggest story in the adventures: since then I have been mostly concentrated on some minor improvements and techniques, and finding out what else could be done to get my coffee on a proper level.





Coming up next: BFC gets an Eric's thermometer, and I get a new headache.

Futahaguro
Posts: 14
Joined: 8 years ago

#23: Post by Futahaguro »

Well after that I am glad it got you the end result you wanted! To be fair to the manufacturer you need to make sure to compare apples to apples when you are measuring the volume of the boiler. It seems like you are measuring your new boiler volume without the heating element but you are measuring the old boiler with the element in. You said the old one has 680 mL with the element in but the manufacturer said 800 mL. Did you measure the old boiler with the element out?

You said the boiler fittings were warped slightly but I did not see any comment about a fix. Did they end up working fine? The heating element hole looks like it was machined, were they able to do that? It's just weird that some look machine but some look welded.

tonythewonderful (original poster)
Posts: 92
Joined: 8 years ago

#24: Post by tonythewonderful (original poster) »

Futahaguro wrote:o be fair to the manufacturer you need to make sure to compare apples to apples when you are measuring the volume of the boiler. It seems like you are measuring your new boiler volume without the heating element but you are measuring the old boiler with the element in. You said the old one has 680 mL with the element in but the manufacturer said 800 mL. Did you measure the old boiler with the element out?
The volume of both boilers was measured with the heating element in. Since I have only one element, I just measured the old boiler first, and then the new boiler. A boiler was disconnected from the machine, had their fittings closed. Then I filled it with cold water through the opening on the top, and measured the volume of the water.
The original boiler had also the hi-limit thermostat pipe removed. The new boiler has it welded in, so it could not be removed. So, the original boiler had a (very slight) advantage.

I will not be able to re-measure the new boiler without HE, however, the old boiler is still here, and I will do a "fair" measurement: without the HE and without the thermostat pipe in, later today.

So, just to clarify:
840 mL - is the volume of the new boiler with HE in. 860 mL was the calculated volume of the empty cylinder, before the boiler was actually made.
680 mL - is the volume of the old boiler with HE in. So, without the HE, it will be slightly bigger.


The mass of the boilers was measured without HE, and the new boiler is almost 3 times as massive as the old one: 3, 078 g vs 1, 086 g


Futahaguro wrote:the boiler fittings were warped slightly but I did not see any comment about a fix.
Yes, but this was only noticeable when I screwed in the thermosensor and the HE. It was not anything major - they work just fine. The only difference was: on the original boiler the sensor and HE could be easily screwed in just by hand (of course this is without teflon tape), on the new one I had to use wrench to do so. So, it did not need any specific fix. Once everything was assembled - it has been working great :)
Futahaguro wrote:The heating element hole looks like it was machined, were they able to do that? It's just weird that some look machine but some look welded.
I think this is something that only Jack form INVIL can answer. Maybe what happened, the 1 1/4 fitting was welded in, and then the outside part was machined, so the HE gasket will not leak. At least he did not charge me anything extra, even if it something were machined.

tonythewonderful (original poster)
Posts: 92
Joined: 8 years ago

#25: Post by tonythewonderful (original poster) »

OK, I remeasured the old boiler.

Without HE, it holds 710 mL water (30 mL for the HE, I think is quite reasonable). Not even close to 0.8 L.
The only way it can be 0.8, if the whole volume of the boiler is measured, that is, if the volume of the metal the boiler is made of, is added to the internal volume:
77mm (diameter), 181mm (hight) gives us ˜840 mL. But that's ridiculous. By the same token car manufacturers could measure the external volume of their engines - we would drive 200L cars...

Futahaguro
Posts: 14
Joined: 8 years ago

#26: Post by Futahaguro replying to tonythewonderful »

Bizarre. Maybe there is a good reason for it, or maybe it is just marketing:)

Headala
Posts: 917
Joined: 10 years ago

#27: Post by Headala »

Another possibility is that the manufacturer originally was using a different part and the specs/marketing never got updated. This is something I dealt with frequently at a factory I consulted for. Buyers find a cheaper/better/more efficient part, replace it in the production, but it doesn't always trickle through to documentation or marketing.

tonythewonderful (original poster)
Posts: 92
Joined: 8 years ago

#28: Post by tonythewonderful (original poster) »

Headala wrote:Another possibility is that the manufacturer originally was using a different part and the specs/marketing never got updated.
Yes, it certainly might be the case. After looking at different coffee machines BCF is making, it looks like that brew boiler from Junior Extra TCI and Perfetta (single boiler, not HX) look very similar (the part numbers are different, though), the only difference is the volume 0.7 (P) vs 0.8 (JE TCI).


It is just rotated differently on the diagram. Who knows, maybe they ran out of 0.8L boilers, and used 0.7 instead? It is just when you have a huge boiler, like 2L, 100 mL probably won't make any difference, but when it is something small, then every mL counts :)


tonythewonderful (original poster)
Posts: 92
Joined: 8 years ago

#29: Post by tonythewonderful (original poster) »

BFC gets an Eric's thermometer, and I get a new headache.
Just because it is digital, does not mean it is right. (Dan Ken)


I've heard this from Dan in one of the "Introduction to newbies" videos. It is not a secret that the temperature PID displays is not necessarily the true representation of what pours out of the group head.

The fist time I realised there is more to it was when I tried to adjust the brew temperature from the default 85C (which I thought was too low) to a more appropriate 92C. However, at 92C (according to PID) the grouphead would spit the boiling water in all directions. "That's strange" - thought I, we are not living high up in the mountains... Then I read about the off-set, and PIDs, and it became obvious that the figure on PID was only slightly better than a guess. It gave perfect, but false assurance that I am in control.

Also I have read that if you switch PID to display temperature in F instead of C, the temperature (at least on display) would be stable. And that was exactly what I observed. While in C, the temperature would jump up and down 1 or 2 degrees while machine was idling. When switched to F, it would show the same temperature all the time.

Of course, I was unhappy about this matter. After all, with PID I expected precise control, and no guessing. For a while I lived with it, being busy with boiler replacement - there was no time to worry about PID.

With the new boiler I noticed one peculiar thing - the first shot would get much better than the second. The first one would be richer in taste, have more body. And because I was usually the one who gets the second cup... well... I had to get to the bottom of it.

Eric's grouphead thermometer was the first and only choice. By the way, I heartily recommend it to everyone who has a compatible machine with an exposed e61 grouphead and wants to experiment.


After easy and uneventful installation two things became evident: while idling, the grouphead temperature would slowly, but constantly cycle around 197F, while the PID happily showed 225F (I have zero off-set) all the time. I did not pay much attention to that, until I saw what was happening during shots.
During the fist shot (after overnight idling) the temperature would start with ˜196, and quickly climb to 205F - 207F, which is fine, as the temperature at the puck would be 3-5F colder (I have no way of measuring that, though - this is from the thermometer description). The shot would come out very good.
During the second shot (about 5-7 minutes later) the temperature, after a very brief raise to 202 - 203F, would quickly drop down to 195F, resulting into a much less satisfactory shot.
"That's 10F of difference" - observed I

I decided to investigate further. First of all I wanted to know the ranges the temperature would change while the machine was idling. Since I do not have a digital logger, and sitting next to the machine for many minutes just watching the thermometer was too much of a drag, I set up a camera on a tripod to take pictures of the grouphead thermometer at a specified interval. By a few tries, I determined this interval as 12 seconds, so I would get 5 photos/measurements a minute for 15 minutes (at which point the thermometer would time out and shut itself off). Yes, agree, not a digital logger with thermocouples, but workable nonetheless.

Here is what I got:



After that, more thinking (and googling) ensued. Obviously the default PID settings were meant for the standard boiler. Reading about how PID worked helped with general understanding, however, I wanted to know how I am to find settings that would be best for my new boiler.

I found a manual from a completely different machine, and copied the PID settings from there. Here is what I got. Note, that the set temperature (and displayed temperature) was the same in both cases:



That was already much better, much tighter control. But of course I did not want to settle just for better. No, sir! I wanted the best I could get. I needed some sure way of finding the PID numbers. And of course I did not want to randomly assign those numbers and test what happened - I would still be doing that.

So, I asked a question - is there a practical way/rule of thumb to figure out PID settings for a controller that does not have a self-tuning feautre (as far as I know my GICAR does not have it). And surely, there is. I actually did not need to look any further than Wikipedia.

It turned out to be very simple. Although you will need a way to measure the temperature and record it, if you want to go this way.

Here is what you do (Ziegler-Nichols method):

1. Set "I" and "D" to zero, and start with some small "P" (I started with "1.0")
2. Gradually increase "P", and test after each adjustment, until the system starts to oscillate, that is the temperature would regularly go up and down at the same interval. You will need the first "P" value at which this stable oscillations start. Also note the duration of the oscillation interval in seconds.
3. Then you just calculate "I" and "D" according to simple formulas. No need to buy any analytical software, or solve differential equations.

P=0.60 * P(osc)
I=2*P/Int(osc)
D = P*Int(osc)/8

where,
P(osc) = the value of P at which first stable oscillations occur
Int(osc) = Interval of oscillations in seconds

So, I just started with "P" = 1.0, and increased it by 0.5 each time, and got the following. Note, that the set temperature was the same every time: 225F.




Starting from "P"=2.0, there are stable oscillations. I could probably go for more fine search by testing 1.6 - 2.5, in 0.1 steps. However, the method of acquiring these data with the camera was not the most entertaining one, and I decided to settle with what I had for now.

So, P(osc) = 2.0, Int(osc) = 192 sec
PID numbers: P = 1.2, I = ~ 0.01, D = ~ 29. These gave me the tightest control for my boiler, while idling. Now during shots the temperature always goes up!




Obviously, "more research is required". E.g. I'd love to see the relationship between the boiler's temperature and grouphead temperature, temperature at different depths of the boiler, how temperature recovers after a shot, etc., etc. This will have to wait till I get a proper digital logger.

By the way: in double boilers there are two different boilers (who would doubt?), however, (at least in my machine), there is only one set of PID figures. That is, one of the boilers will use non-optimal PID settings. Of course, who cares about precise control for the steam boiler?

Seakiwi
Posts: 1
Joined: 7 years ago

#30: Post by Seakiwi »

I am simply stunned by your journey Tony! I thought I was OCD about coffee but...............

I have had a BFC Junior Extra TCI with rotary pump for a couple of years now and must admit to never having had the inclination to disassemble/reassemble it once let alone as many times as you have!

I too, was a little disappointed in the temp stability in the brew boiler when I bought it but when I queried BFC themselves with the problem I was experiencing (4 degree C overshoot of set temp) they came back quite quickly with PID settings that have given me a less than 2 degree variation consistently over the last 2 years.

The settings they gave me were P=2 (default was 7) i=0.01 (default was 0.10) and D was left at the default which is 15.

I have never had a problem with metallic or copper tasting water, although the machine was commissioned in the tech room of the importer as we played with settings etc and quite a lot of water went through it before we even tasted any shots.

I pull mainly double ristretto shots for my own comsumption and prefer very dark roasts so I guess there could be some masking, however I think I will try and resist the temptation to change to a stainless boiler!

It would be interesting to taste the same coffee/grind/dose through each of our machines side by side to see if there is any noticeable difference now.

I would like a reliable group head temp reading as well to know just what the drop from boiler to group head actually is. I believe one of the BFC importers to NZ may be working at doing something like this.

Oh, and I am in Auckland as well.