Hey all,

The time has come for me to begin the documentation of a series of modifications I'll be doing over the next little while to make my home machine more forgiving and versatile. I've covered the "why?" in my long and rambling path to preinfusion and pressure profiling. So let's get to it.

The equipment:
Mostly original 1993 Rancilio S20 MIDI CD 2 group mini commercial machine, which can be seen here, next to the Mazzer SJ with Duranium burrs from back when it still had a hopper. The S20 is 220V and fully plumbed in with 2 HX groups, an internal procon pump and a 5L boiler with a single steam wand on the left and hot water tap on the right. The single pressure gauge shows boiler pressure only. The CD model is stripped down from the volumetric DE version. The brew switches power their respective solenoids and energize a relay that fires the pump. The fluid path is unadulterated by gicar flow meters and only contains a valve body with non-return valves and an expansion valve along with a branch that feeds the boiler fill solenoid. Each of the non-return valves feed the HX injectors and the thermosyphon tubes feed the twin groups. It is definitely a dragon in terms of HX behavior. It also has notoriously little headspace and is extremely unforgiving of any inconsistencies in technique.

My longer-term plans are to execute the modifications outlined here and here. I intentionally titled these as options and not phases or stages because, while they serve different purposes, they compliment each other and the various bits and pieces off each can be carried out separately. As of my writing this very sentence, the only modifications to the S20 so far are machined jet breakers to accept countersunk shower screen bolts. I know, it's some pretty progressive stuff...

In a few hours I'll snap some photos and walk through the process of my first minor modification, which should be complete in time for my morning cappa... Much like my first topic, I hope that this one provides some reading enjoyment to a few folks with a special kind of curiosity and sense of humor, triggers some fascinating and lively discussions on the pros and cons and ins and outs of what I'm embarking on and brings some folks together to chat about really nerdy stuff that most people just can't be bothered to care about.

Cheers!

- Jake

Chapter 1: Line pressure preinfusion

This mod will be far and away the most simple to implement and will hopefully be the sole electrical modification to the machine.

The plan is simple:
-Interupt the signal from the brew switches to the pump relay
-Place switch inline and mount in convenient location
-Open switch contacts during preinfusion to prevent pump from running
-Close switch contacts at end of preinfusion to start pump.

Supplies:


Locate signal wire on pump relay:


Connect wires to relay and signal wire:


Both wires follow the original loom to the switch location:


Wires routed and connected to switch:


Switch "mounted" in discreet location:


Results:
Prior water debit before any modifications was 770cc/min, or 385cc in 30 seconds. After adding the 0.5mm gicleur to the left group, I got that number down to around 250cc in 30 seconds. Running the right group with no back pressure while measuring the water debit out of the left group yielded 235cc in 30 seconds.

So, what does taking the pump out if the equation do for us? Initial water debit while pre-infusing was 155cc in 30 seconds. Cheating by opening the right group solenoid yields 145cc in 30 seconds. I still suspect that opening the right group has a more profound effect on the flow through the left group when there is a puck to preinfuse versus simple water debit calculations. I'll know more once I have a puck pressure gauge...

More to come in the next chapter...

Chapter 2: The design begins

With preinfusion working like a champ, I made some progress on the mechanical design of using the needle valve for pressure profiling this weekend.



Below you can see the nuts and bolts of my pressure profiling design path. The bit sticking out past the gicleur cap is the .25" adapter knob on the needle valve, which will allow me to either attach a shaft and have a dial control on the front of the machine, or to mount a lever directly on the knob and have more of a traditional paddle that extends out through the top cover of the machine. I'm leaning towards the latter solution for aesthetic reasons if nothing else.


Closer view showing the specifics of what I need to manufacture to get this all put together is shown below.

The big tan part is the miniature needle valve with its control shaft sticking out to the left, passing through a modified gicleur cap. The silver bit on the right is a bushing I will fabricate from an 8mm stainless steel set screw. The set screw will be faced off and drilled out and tapped with #10-32 threads from the back to preserve the internal hex for assembly.

Assembly will involve threading the bushing onto the needle valve with thread locker to ensure it doesn't become stuck in the group during disassembly. The valve will then be threaded into the group. Upon inspecting the diameter of the shoulder on the valve after the threads, I think I will have a calibrated bushing between valve and the back face of the group to control the axial location of the valve. A bigger bushing will move the control shaft further out of the group, a smaller one recessing it further, allowing fine tuning of the interface between the needle valve and the gicleur cap.

The cap, highlighted in blue, will be machined to pass over the control shaft of the needle valve and will have a recess that will compress the o ring, shown in black, over the bonnet of the needle valve. Obviously, it is also critical that the group cap seal against the group as it does now. As such, the aforementioned bushing for controlling the inserted "height" of the valve is pivotal in assuring that the valve is positioned such that the o ring is sufficiently compressed while the cap is still seated firmly against the group.

All told, I'll need to bore a hole through the gicleur cap along with a sealing surface for the o ring and drill and tap an 8mm set screw. Otherwise, everything else is standard, off-the-shelf parts to implement this.

I'll post an update when parts arrive and I get things fitted. In the meantime, I'll work on modelling the control panel of the machine so I can detail the control mechanism.

Cheers!

- Jake

Alright,
I got a bit more work done on the design side of things. Notable mentions include utilizing a couple small chunks of SS swagelok tubing to set the depth of the needle valve and also as a backup ring to position the o ring into the sealing bore of the gicleur cap. I also detailed out the mounting locations and methods for the thermocouple and pressure sensor, both behind the shower screen using 10-32 threaded holes with a light spot face to ensure a good seal.



Above and below you can see the updated design using the tubing to set the depth of the needle valve in the cavity and position the O ring. Careful measurement should ensure that I get things close enough to seal everything up within the margin of error, but it should be relatively easy to either shave a bit of length off the tube sections or cut a longer one if fine tuning is needed to get the valve and o ring positions optimized. The inlet port on the needle valve can face any direction, so it shouldn't be too hard to get this dialed in.


Below you can see the fittings for temperature and pressure measurement at the puck.


The socket head cap screw on the right is sealed against the spot face on the group with a standard crush washer. The cap screw is drilled through and lightly countersunk to provide a basin for the thermocouple bead to rest against. The cap is also drilled through so the thermocouple wires can pass through the central bore and then exit through the side of the cap so an Allen wrench can be used to snug up the bolt without damaging the wires. I plan on sealing the thermocouple bead and wires to the cap screw with either silicon or epoxy before installing in the machine. Any thoughts on which approach would be best would be greatly appreciated. This location allows the bead of the thermocouple to be in the direct path of the water as it enters the jet breaker cavity behind the shower screen. It should yield a nice hybrid of an Eric's group thermometer and a Scace device readings, since it will be influenced greatly by group temperature, so the temperature at the start of a shot won't be artificially low like a Scace profile, but will also be in direct contact with the brew water, so it shouldn't read much (if at all) hotter than the true brew temperature. Time will tell...

On the left we have a standard 10-32 adapter to 1/8" compression fitting to the puck pressure gauge. This fitting comes standard with an o ring face seal which will seal on the spot face on the group surface. This fitting is larger than I would have thought, but I think will still work nicely to get a true puck pressure signal out to a gauge.

I have my thermocouple and readout in hand along with my thermocouple mounting bolt. Once I get the adapter for the pressure fitting in hand, I'll pull my group to mount them and get my final measurements for fabricating the other components. I'm currently trying to locate a supply of low-lead brass compression fittings where I can order individual units. Right now it looks like I'll be ordering a package of 10 . I know I said I wanted my mods to be reversable, but I can always plug a couple holes...

Cheers!

- Jake

Hi Jake,
Some time ago there was a discussion concerning the installation of a thermosensor on the German coffee forum. The patient was a Silvia which has (more or less) the same group head as the commercial Rancilios if I understand correctly.
https://www.kaffee-netz.de/threads/inst ... pe.111093/
Faustino also found the ideal spot to be smack dab in the middle and he drilled out the existing screw hole and the screw holding the screen, documented well with pictures.

Thanks for the link Pio,

While my German is a bit rusty, I was able to gather that the OP procured a 150mm long 0.5mm grounded thermocouple and used an M5 compression fitting to mount it in the top center of the group, with the tip of the sensor poking through a drilled center screw. He drilled through the factory threaded mounting hole with a 2mm bit and then opened it up to 4.5mm from the tip and finally extended the original threads clean through to the top of the group.

Very clever approach, and I had originally planned on this tactic but alas, the S20 group is sufficiently different from that of miss Silvia that it seemed unlikely to be successful in my application. The offset Silvia group is a solid 10mm thick with a flat top from the edge all the way over to the offset boiler mounting boss. Mine is much more contoured. So much so that the center bolt hole has a cast dome in the center of the group to provide thickness for the threads. The M5 fluid threads require a spot face to seal against and in order to gain the needed sealing surface the center dome would need to be ground flat, leaving insufficient threads for even the fluid connector, let alone the shower screen bolt...

Also, I am using a modified TruTemp thermometer (since I got one for free) and my options for it are to either use the factory 1/8" shaft with a compression fitting or to simply remove the bead from the tube (it slides right out) and install the exposed TC junction in the fluid path. Since the group center is effectively off-limits, I figured the next best option would be to mount it in the jet breaker cavity. I suppose an argument could be made to clock the sensor 90 degrees one way or the other to give the stream a chance to do some mixing but I think given the fact that I won't be logging any data other than videos of shots watching the display that the difference would be negligible.

So what I'm tasked with is determining how to best seal up a bore that has 2 thin gauge wires running through it in such a way that is food-safe, has no unpleasant aroma or flavors, and won't leak under pressure. Truth be told, I'm leaning towards JB weld or something similar, but if anyone has any recommendations of a preferred soft-setting silicone sealant that they've had good success with, I'd be up for that, as well. Especially since one could likely pull the TC back out of that sealant and try other things and clearly the JB weld approach would be "one and done"...

Cheers!

- Jake

Minor update,

I got my thermocouple mounting bolt drilled out so I can mock up my installation plan. Once my compression fittings and crush washers arrive, I'll pull the group and cut this bolt to length so the bead of the thermocouple is right where I want it. I'm still debating whether to mount it directly in the stream from the group or to move it out of plane so the readings are tempered a bit. I suspect most of the heat transfer from the water into the group occurs in the internal passages of the group, but obviously there is a significant amount of heat lost in warming the puck that I may have a better chance capturing if the bead is shielded from the stream exiting the group port...

Here's a few update photos:


A shot of the wire routing to allow for installation and removal without damaging the wires...


And the whole assembly...


I'll definitely be going the silicone route (over JB weld) the more I think about it. Even better would be to see if I can get a teflon tube routed through the wire path and seal that up. I am certain the TC wires will not survive the vibration-rich environment being routed through the bolt like that if they are in contact with any sharp edges at all. I think a slug of silicone, injected into the bore will keep them safe as long as I keep the wires suspended in the center of the slug of silicone while it sets up... The frustrating thing is that i can't do any of that until I pull the group and get an accurate bolt length target. The good news is that i can seal that hole with a standard bolt while I figure out the best path forward to complete this little task.

Cheers!

- Jake

Parts ordered...

The parts list is short, and mostly in line with my initial design with a few minor tweaks. First, I'm as cheap as they come, so while I was pleased the needle valve I've selected was only $8 from the manufacturer, I was appalled that the drop shipping charge was $14.99... I've been hunting for a similar valve for the last few months, as well as searching for a local distributor to no avail. Right up until Saturday, when I found my valve, in metric skin (had initially been looking at 10-32 fluid fittings) for a whopping $2.40 on Amazon as an add-on item with free shipping!

Since I made the switch to m5 threads on the valve, I figured I'd standardize and picked up a 5 pack of m5 x 4mm straight push-lock fittings for the group along with a banjo 90° push-lock fitting for the back of the gauge. I figured I don't have a good reason to go with 1/8 copper tube with compression fittings when 4mm x 2mm Teflon tube will suffice. All totalled, everything came to just shy of $20 shipped, including the 11 bar gauge. All that is left is to source a couple feet of Teflon tube locally and I'll be in business.

The change away from 10-32 also means a better solution for my shower screen temperature probe. While I quite liked the idea of my drilled-through bolt, sealing the probe wires in a small length of Teflon tubing will make a much more serviceable solution. I'll pilot-drill the group holes (for temp and pressure feeds) through at 1/16" before spot-facing, flat drilling and bottom tapping the m5 threads. This will allow me to drop the temperature probe directly into the push-lock fitting, with the thermistor bead poking through the pilot hole, in direct contact with the water behind the shower screen while still making it easy to remove when the need arises. Going with the tube fittings will also offer much more protection for the sensor wires. I was a little nervous about the sharp 90° turn the wires were making in the bottom of the hex socket on the drilled bolt.

While I wait for the parts to show up, I'll get a stainless 8mm bolt cut down and drilled for m5 internal threads to mount the valve in place of the gicleur. When the parts arrive, I'll test fit the valve behind the gicleur cap and get a feel for axial placement of the valve. This will be critical as I need to locate the o ring that seals the OD of the valve body to the inner bore of the gicleur cap as precisely as I can. Too far into the group and the o ring won't even make contact with the cap. Too far out of the group and I could extrude the o ring or worse, crack the cap when I tighten it up. Critical measurements will be the depth of the back wall and the thickness of the cast gicleur cap. I'm hoping that the whole hex head on the gicleur cap is solid, as I have it modeled in previous posts... If so, I'll breathe a sigh of relief because I'll have plenty of room to chamfer into a sealing bore with a step down to capture the o ring and let the control shaft poke through. If not, I may have to make a new cap or press a sealing insert into mine to get things to line up the way I need them to. Only time will tell. Time and a digital caliper. I'll need both.

Cheers!

- Jake

Hello, old thread.

The Parts:
The orange slug was the "just in case" so could plug...

The Hole:


Cleaning the Bore:


Out with the Old:
This is what Rancilio's 0.5mm gicleur looks like, by the way...

In with the New:



First goes the Valve:

Then goes the Cap:


No leaks!

I still dont have an actuator, so I set the debit at 70g in 30s at 3 bar (pump off) which yields 140g in 30s at 9 bar (pump on).

No shots yet. Too late and im out of beans. Still need a group pressure gauge...

Cheers!

- Jake

Jake_G wrote:No leaks!

Woohoo!