Hey all!

After a couple weeks of having my recently acquired GS/3 get all its seals replaced and put back together after being deconstructed, the time has arrived for it to mature into the machine I always knew it could and should be.

This is Phase 1, and it is very much function over form. The basic premise is to remove T.2.104 that runs from the HX mixing valve to the coffee boiler and replace it with a couple tubing adapters, some 4mm PTFE tubing and a real time profiling needle valve.

Phase 2 will be an ergonomic and aesthetic upgrade to phase 1 with a paddle and group gauge. This will take some time to develop, test and install, but will eliminate external plumbing and integrate the profiling valve into the group, similar to what I did on my Rancilio S20. In this initial post, we'll just walk through the execution of phase 1.

Parts list:

• •1/8" BSPP x M5 threaded bushing: Qty 3 - PN RK05/10
  •1/8" BSPP crush washer: Qty 2 - PN 3500-02
  •1/4" BSPP x 1/8" BSPP threaded Bushing: Qty 1 - PN 2151-04-02
  •1/4" BSPP crush washer: QTY 1 - PN 3500-04
  •4mm tube x M5 Quick connect fittings: Qty 4 - PN QR2-F
  •4mm x 2mm PTFE tubing: Qty 16"
  •Miniature needle valve with M5 ports: Qty 1 - PN M-MNV1-K
  •Custom knob and bracket

Steps:
Begin by powering down your cool machine, unplugging it and removing the right side and top covers. In the photo below, the rear cover and rear frame member are also removed, but this shouldn't be necessary to do the mod:

Image: Machine Right Side - Covers Removed


To access the tube fitting on the coffee boiler, first loosen the 14mm hex nut securing the coil to the tea water solenoid valve, remove the coil and set it aside. Then, using a 13mm flare nut wrench, loosen the tube nut that secures the T.2.104 tube to the coffee boiler:

Image: Coffee Boiler Tube Nut Loosened


Next, slightly loosen the 18mm tube nut for the OPV feed pipe on the steam boiler and gently pivot upwards to access the tube nut on the HX mixing valve. Using a 14mm flare nut wrench, loosen the tube nut:

Image: Tube Nut Removed from HX Mixing Valve


Remove the T.2.104 tube from the machine and set aside:

Image: T.2.104 Tube - Removed

Next thread two of the four M5 quick connect fittings into two of the 1/8" x M5 bushings and thread in one of the bushings with a crush washer to seal against the face of the tee fitting on the coffee boiler:

Image: Quick Connect Fitting Installed on Coffee Boiler

Next, stack the other assembled 1/8" x M5 bushing into the 1/4" x 1/8" bushing, using the remaining 1/8" crush washer between them and install this whole assembly into the HX mixing valve with a 1/4" to seal it (the bushings I'm using feature a captivated O-ring seal, so I've omitted the 1/8" crush washer between the bushings in the image below). Once this is tightened down, lower the OPV pipe back down over the HX Mixing valve and tighten the tube nut on the steam boiler:

Image: Quick Connect Fitting on HX Mixing Valve


Next, insert and firmly seat each end of the PTFE tubing into
the quick connect fittings:


Images: Tubing in Place

Finally, assemble the ending fittings onto the needle valve, find a suitable location to place the needle valve, then cut and connect the tubing accordingly. The valve I chose to use has a male M5 inlet fitting, which I threaded into the last 1/8" x M5 bushing. This provided female M5 threads for a quick connect fitting as wil as male 1/8" BSPP threads for mounting the valve assembly. I threaded the final quick connect fitting into the female M5 outlet port of the needle valve and positioned valve atop the steam boiler as shown below:

Image: Profiling Valve Assembly


Here is a shot showing all the modified plumbing components:

Image: GS/3 Profiling Modification - Phase 1


And here it is mocked up with the top covers in place and my 3D printed knob poking through:

Image: Minimalist Profiling Knob


Up close and personal:

Image: Detail - Profiling Valve and Cup Warmer Grate


And finally, a quick video of the my profiling GS/3: Video: Profiled Shot on La Marzocco GS/3 AV

Hmmm. Interesting design. Some worry-wart questions:

1. What's the pressure rating of the quick-connect fittings?
Need to be at least 12 BAR to withstand reheat pressure, preferably higher.
2. What's the temperature rating of the quick connect fittings and PTFE tubing?

For some reason my post was truncated but I didn't notice until now.

Hmmm. Interesting design. Some worry-wart questions:

1. What's the pressure rating of the quick-connect fittings?

Needs to be at least 12 BAR to withstand the reheat pressure set by the expansion valve. I'd go with 14+ BAR for a safe margin

2. What's the temperature rating of the quick connect fittings and PTFE tubing?

Tough to spec this. The water \in the HX at idle must be pretty close to boiling, if not somewhat higher (the tube is sitting in ~265F water, but it's under 6-9 BAR of pressure.) Once the brew cycle starts the HX water temp will drop, but the more you close the needle valve the slower the flow through the HX, which could cause water in the tube to boil. It can definitely flash boil when it hits the boiler water. Makes a nasty noise. That happens when I reduce the line pressure to 1-2 BAR and turn the motor off for a Slayer-like pre-infusion flow rate.

Note that Assaf has effectively the same configuration, except he uses copper tubing and a precision needle valve. He has a gear pump for pressure profiling, so the needle valve is for Slayer-like flow profiling.

Note that while the expansion valve still protects the boiler, it no longer protects the pump. The configuration should work with the stock rotary pump because it has a bypass valve, though I wonder if the stock bypass can deal with the very large backflow when the needle valve is shut way down. LM redisigned the MP group valve -- the so-called "conical" valve -- to bypass the excess flow, and I've wondered why they bothered if the pump bypass valve can deal with the flow.

I have bigger concerns about the configuration when used with a pump that doesn't have a bypass valve, like my gear pump. As I said, the expansion valve no longer protects the pump. That said, I did some tests that indicate the gear pump pressure doesn't exceed spec when the flow ahead of it is reduced to rates practical for flow profiling. But it's a different story if the needle valve is shut completely. I'd want a needle valve with an adjustable safety stop to prevent that. Assaf and I have had numerous debates about whether my gear pump could survive the needle valve being closed completely. I have the model with a current-limited motor, but while this does protect the pump to some extent I believe it'll allow the max pressure to exceed the pump specs.

Jake, can you post your source(s) for the parts and flare nut wrench? And a photo of the wrench would be cool, too.

pcrussell50 wrote: Bravo Jake, for returning one of the machines we used to dream of in our "espresso youth", back to the leading edge.

I don't know about leading edge just yet, but thanks all the same!

mgrayson wrote:Jake,

You continue to amaze!

And I wish I'd heard of flare nut wrenches before. I've often wanted something with more contact on soft metal.

Matt

Thanks Matt!

The flare nut wrenches are a godsend.

Peppersass wrote:Jake, can you post your source(s) for the parts and flare nut wrench? And a photo of the wrench would be cool, too.

I picked up the standard fare Craftsman 5 piece metric Flare Nut Wrench set back when I was a tire buster at Sears automotive in the early aughts.

Image: Flare Nut Wrench Set - Available at Lowes

1. What's the pressure rating of the quick-connect fittings?

250psi

2. What's the temperature rating of the quick connect fittings and PTFE tubing?

The PTFE tubing is fine at really high temps and pressures, but I'm asking more of the fittings than these particular ones are really rated for. Aventics (formerly Bosch/Rexroth) manufactures a heat resistant QR2-F fitting that would be preferable in this application. All that said, I'm pretty comfortable with my selection, even with the lower temperature rating. The important bits are made from appropriate materials and the application is static, so fatigue failure is not a primary concern. It's a great observation and question on your part and I'd there weren't an impending phase 2, I would have sourced more costly and robust quick connect fittings, or gone with compression fittings, for sure.

Tough to spec this. The water \in the HX at idle must be pretty close to boiling, if not somewhat higher (the tube is sitting in ~265F water, but it's under 6-9 BAR of pressure.) Once the brew cycle starts the HX water temp will drop, but the more you close the needle valve the slower the flow through the HX, which could cause water in the tube to boil. It can definitely flash boil when it hits the boiler water. Makes a nasty noise. That happens when I reduce the line pressure to 1-2 BAR and turn the motor off for a Slayer-like pre-infusion flow rate.

By definition, if the coffee boiler pressure is at or above the steam boiler pressure, there can be no boiling. Since my needle valve installed in the line after the HX mixing valve, the water on the downstream side of the valve will be cooler than the water in the HX. Flash boiling has so far not been observed. I'll do some thermography to verify, but my assumption based off of researching the Slayer preheat systems (194°F) and the published data from the LM FB80 HX and mixing valve combo (174°F, and presumably the same as GB5 and perhaps GS/3) is that the water in the T.2.104 tube was designed to be cooler than the coffee boiler set point, which wholly eliminates the concern of flash boiling so long as the outlet of the HX is throttled and not the inlet. Remember that all machines experience zero flow at idle, and the GS/3 is lauded for the ability to walk up and pull shots with no flush. By intentional design, this mod lacks the gumption to influence this characteristic of the machine.

Note that while the expansion valve still protects the boiler, it no longer protects the pump. The configuration should work with the stock rotary pump because it has a bypass valve, though I wonder if the stock bypass can deal with the very large backflow when the needle valve is shut way down. LM redisigned the MP group valve -- the so-called "conical" valve -- to bypass the excess flow, and I've wondered why they bothered if the pump bypass valve can deal with the flow.

While running the pump with the needle valve fully closed is not my intent until I update the firmware such as to allow for long enough shots to employ blooming shots and integrate a pump kill switch, there is no reason that doing so now would be any worse than performing a backflush, choking the machine with too fine a grind or even employing the factory programmed preinfusion, which closes the 3 way valve for the pause and soak duration while the pump continues to run. The likely reason LM employed the conical valve was to improve the reliability of the wear-prone original MP design, which had O-rings scrubbing across the pressure and exhaust ports. They could have designed the components with no overlap between the pressure and exhaust ports, but the metering would have been poor and there would have been a third position that was "off", kind of like the middle position on an E61. Adding overlap was a clever way of softening the transition and allowing for some adjustability while eliminating the electric solenoid valve. Clever? Yes. Precise? Maybe kinda sorta. Wasteful of reservoir water? Absolutely!

Furthermore, many a Bianca, and now Pro700, Synchronika and countless other E61 machines with the new profiling devices available now have been profiling using the same method employed here as well as myself with my S20 for the last year with nary a complaint. It just works...

With respect to a gear pump, I agree that it is unwise to use this method without a bypass or a programmed pressure limit in concert with a magnetic clutch or current limit. Even so, if you can backflush safely, I wouldn't worry too much about choking the pump, but it is true that there is an air cushion when back flushing that softens the load on the pump that would not exist between the pump and the needle valve. I know Slayer warns not to close their valve for this reason...

pcrussell50 wrote:Now, Jake's pump is almost surely a Fluid-O-Tech, but I can't imagine it's bypass system being very different or less capable than Procon's though I don't know that as fact.

Yep. FOT Rotoflow in the GS/3 and it has zero issues running in full bypass. Remember that these pumps are comically oversized for espresso shots as they need to be able to fill the boilers in a reasonable amount of time. As such, anywhere from 70-90% of their flow is bypassed any time you're pulling a shot. Its unwise to run any pump dead-headed for any significant duty cycle, but for our purposes I can't see any real reliability or longevity concerns given the defacto method of operation.

Thanks for the comments everyone, keep em coming!

Cheers!

- Jake