Rotary pump conversion leads to 'Rituale suicide'?

[Dr Jim]

Ok - so I'm the poor schmuck who decided it would be a Good Idea to convert their Isomac Rituale from 48w Ulka vibratory pump to an externally mounted rotary pump in a quest for more Platonically Perfect espresso...

Which conversion appeared to have worked well until yesterday, when a new and most disturbing symptom has surfaced. Now, from time to time, the boiler-fill logic appears to enter some sort of psychotic fugue state such that when the external pump is running to pressurize a shot, water is also being sent to the boiler as well as the HX loop.

This excess water rapidly overfills the boiler, the boiler pressure rises to 2.5 Bar, the safety valve on top pops open and water gushes out of it until the pump is shut off.

As near as I can trace the circuits involved, it looks like there is a level sensor on top of the tank near one end of the HX loop, it feeds a green wire which goes into the Gicluer control box mounted underneath the boiler. The microswitch which is activated by the grouphead lever also feeds this control box, so I am assuming that this box enables the logic function which says "Send pressurized water to the HX loop, not the boiler, during extraction" - but I've not yet discover the solenoid valve which performs this routing.

I believe that the boiler-fill sensor and circuits are working correctly, since there are boiler-fill events lasting about 15 seconds occurring roughly every two hours which is approximately what the machine did with the Ulka pump in circuit, and the boiler does not appear to overfill.

My working hypothesis is that the rotary pump is too big power draw for the Gicleur's internal relay to handle, so not enough amps are left for the Gicleur to effectively control both the pump and operate the solenoid valve simultaneously - since the solenoid valve is normally open to the boiler, and must be closed by the Gicluer controller - failure to do so means that the incoming pressurized water has open paths both through the HX loop and the grouphead, and through the boiler itself.

So, throwing myself upon the assembled wisdom of the group, would it make sense to use an inline relay to control the external pump?

This would be easy enough to do, just give the pump its very own 110VAC plug and then switch one leg of the AC circuit through a 110V 20amp SSR to control the pump and reduce the load on the controller.

Or, have i missed something stupidly obvious in all of my fussing around?

Cheers

Jim (grouchy without his morning triple)

[AndyS]

would it make sense to use an inline relay to control the external pump?

This would be easy enough to do, just give the pump its very own 110VAC plug and then switch one leg of the AC circuit through a 110V 20amp SSR to control the pump and reduce the load on the controller.

Yes, but I would use an inexpensive mechanical relay (an "ice cube relay"); there's no reason to use an SSR.

And your brain box's brand name is "Gicar."

[HB]

And your brain box's brand name is "Gicar."

Actually the Isomacs use the Giemme controller.

[Dr Jim]

I really shouldn't have posted from last night's notes, in which the capitol letter 'G' stood for 'G'iemme, and not Gicleur ...

With that hurdle mastered, I did some more Web research and found a site which had an actual pinout and translation of the Giemme controller:

http://homepage.mac.com/etorres/junior_espresso/Personal5.html

and a shot of the Giemme controller itself:



Which has proven to be pretty useful, connection #3, labeled 'EV Caricamento' which translates to 'Refill valve' has a black wire with a red plug that traces back to a solenoid valve on the input section labeled 'CEME 5510VV2,05.R04' which I would never have found since it was installed backwards and upside-down underneath the boiler, but this valve looks to be the culprit, since it is appears to be open to the boiler, which is back-asswards, since it should be closed EXCEPT when filling the boiler - at which point 110v is applied to its control lead and it should open.

Further exploration leaves me even more puzzled and confused, since it looks like 110VAC taken from the power cord's brown lead is fed via a brown wire to the microswitch behind the grouphead lever, this switch then connects via a white wire directly to the main pump. Another white wire also connects from the pump back to pin 4 of the Giemma box, which is labeled 'pompa.'

Unless I'm having a Stupid Attack, I would assume that the microswitch is passing the AC current necessary to run the pump motor when it is engaged by lifting the grouphead lever - and not the Giemme controller. I believe that the white AC lead from the Giemme controller is to activate the pump when the boiler level sensor on pin 8 (labeled 'LIV') is ungrounded by the boiler level sensor - this circuit does appear to work correctly as the boiler autofills normally.

Indeed - some disassembly later it was proved that the small metal slug had become trapped in the up or 'open' position and a little positive re-enforcement with a small hammer persuaded it to close - it currently works, but I will order another one for stock tonight since this one now only faintly resembles the part as CEME originally produced it ....

Mutter, mutter, mutter

Cheers

Dr Jim

[AndyS]

Actually the Isomacs use the Giemme controller.

Thanks for the correction, but Gicar sounds more like "gicleur" than Giemme does.

[Ken Fox]

Ok - so I'm the poor schmuck who decided it would be a Good Idea to convert their Isomac Rituale from 48w Ulka vibratory pump to an externally mounted rotary pump in a quest for more Platonically Perfect espresso...

Which conversion appeared to have worked well until yesterday, when a new and most disturbing symptom has surfaced. Now, from time to time, the boiler-fill logic appears to enter some sort of psychotic fugue state such that when the external pump is running to pressurize a shot, water is also being sent to the boiler as well as the HX loop.

This excess water rapidly overfills the boiler, the boiler pressure rises to 2.5 Bar, the safety valve on top pops open and water gushes out of it until the pump is shut off.

As near as I can trace the circuits involved, it looks like there is a level sensor on top of the tank near one end of the HX loop, it feeds a green wire which goes into the Gicluer control box mounted underneath the boiler. The microswitch which is activated by the grouphead lever also feeds this control box, so I am assuming that this box enables the logic function which says "Send pressurized water to the HX loop, not the boiler, during extraction" - but I've not yet discover the solenoid valve which performs this routing.

I believe that the boiler-fill sensor and circuits are working correctly, since there are boiler-fill events lasting about 15 seconds occurring roughly every two hours which is approximately what the machine did with the Ulka pump in circuit, and the boiler does not appear to overfill.

My working hypothesis is that the rotary pump is too big power draw for the Gicleur's internal relay to handle, so not enough amps are left for the Gicleur to effectively control both the pump and operate the solenoid valve simultaneously - since the solenoid valve is normally open to the boiler, and must be closed by the Gicluer controller - failure to do so means that the incoming pressurized water has open paths both through the HX loop and the grouphead, and through the boiler itself.

So, throwing myself upon the assembled wisdom of the group, would it make sense to use an inline relay to control the external pump?

This would be easy enough to do, just give the pump its very own 110VAC plug and then switch one leg of the AC circuit through a 110V 20amp SSR to control the pump and reduce the load on the controller.

Or, have i missed something stupidly obvious in all of my fussing around?

Cheers

Jim (grouchy without his morning triple)

I replied to your original post in a lengthy way, hit the "submit" button, and my response was lost in the internet ether. First I thought maybe it was crossposted to another forum but I could never find it. Unfortunately, I did not have time to retype it.

What I said several days ago was that knowlegeable people I've spoken to with real experience doing espresso machine repair work have told me that there is a lot more to vibe to rotary conversions than simply swapping out the vibe pump for a rotary pump and motor. I have been told that in general it is not worth the effort. Taste tests that Jim Schulman and I did between my two Cimbali Juniors, one a pourover vibe circa 1995 the other a 2.5 year old rotary junior, did not support the idea that there was any sort of taste benefit of going from a vibe to a rotary and that the only benefit was apt to be reduced noise levels when pulling a shot, which is apt to be a minute or two or three or possibly 4, minutes out of your day.

I recommended that you reconsider the project. I still recommend that you reconsider the project.

ken

[Dr Jim]

Ken -

While I understand what you are saying - and agree that such a conversion on a more sophisticated machine like a Cimbali or other commercial device with automation, electronic controls, and careful attention to engineering details would most likely end up being a dreadful hack - on a machine as fundamentally under-engineered and as basically crude as the Isomac it's really not all that far out of character.

Given that the pressure management on the basic machine is pretty simple-minded, and that there are already versions using essentially the same boiler, HX loop, and grouphead with rotary pumps fitted - it was appealing to create a test bed where I can fairly easily go back and forth between the external rotary and internal vibe pump by merely changing the water intake and output lines and swapping the power leads.

Yes, this is hot-rodding, and has no more elegance than shoehorning a supercharged 390 CID Chrysler V8 into a 1953 Studebaker Coupe - which abortion was truly terrifying to drive on the highway, but was clocked at 218MPH at Wendover, Utah one year. There are no commercial or other possibilities with this project, and it is far too crude to go shouting about from the housetops, but it does serve to amuse, educate, and edify me - which is all that I really care about.

Given that most of the parts for this conversion came from other people's junk bins, the total cost including the pump, motor, hoses, fittings, and the SSR to switch the rotary pump has yet to exceed $100 it's been an interesting experiment.

What have I learned?

While the microswitch is quite capable of switching the rotary pump during extraction, the Giemme controller is a 10amp device which isn't sufficient to run both the rotary pump and the CEME valve during boiler-fill, so a relay is needed to isolate the pump's load and keep the valve from chattering and jamming open.

Purely based on observing the Isomacs brew pressure gauge, it does look as if the rotary pump comes up to working pressure significantly faster than the vibe pump, it appears to maintain a more stable pressure during the extraction, there is little or no apparent flutter in the gauge - I often saw a rapid .5 - .75 Bar flutter with the vibe pump - and it can maintain full flow/pressure for the entire extraction.

While there are some definite positives to the conversion, the effect is fairly subtle - there is no massive shifting of the flavor palate, no sudden enhancement of one taste over another, no obvious WOW! effect - but rather it's as if the matrix within which the taste and flavor is embedded has clarified and become somehow less grungy and obscuring - good shots are now simply more vivid and appealing.

OTOH - the machine is now utterly unforgiving of a even minor technique errors in grind, distribution, tamp, and temperature control - pooh up any aspect and it will gleefully spit, channel, and spew vile nastiness all over you - it is shockingly easy to produce a 'sink shot' with but a single instant's inattention. So in this sense I guess it remains true to its hot-rod roots with twitchy, vile handling and idiosyncratic quirks, but very good at what it does do.

Cheers


Jim

[AndyS]

Purely based on observing the Isomacs brew pressure gauge, it does look as if the rotary pump comes up to working pressure significantly faster than the vibe pump, it appears to maintain a more stable pressure during the extraction, there is little or no apparent flutter in the gauge - I often saw a rapid .5 - .75 Bar flutter with the vibe pump - and it can maintain full flow/pressure for the entire extraction.

<snip>

OTOH - the machine is now utterly unforgiving of a even minor technique errors in grind, distribution, tamp, and temperature control - pooh up any aspect and it will gleefully spit, channel, and spew vile nastiness all over you - it is shockingly easy to produce a 'sink shot' with but a single instant's inattention. So in this sense I guess it remains true to its hot-rod roots with twitchy, vile handling and idiosyncratic quirks, but very good at what it does do.

The more stable pressure with your rotary pump is a positive that is improving the quality of your extractions. Coming up to working pressure faster is a negative that is making your distribution errors more critical.

There's no reason why the rotary has to remain a "twitchy, vile handling hot rod." Put an adjustable needle valve ("gicleur") such as mcmaster.com #4555K13 or #7810K11 directly after the pump so that you can regulate how fast the pressure comes up. Adjust the needle valve so that the volume of water that issues from your group with NO portafilter installed measures 70-75ml in the first 10 sec after turning on the pump. Perform this calibration with care. It will make your hot rod docile without ruining its good qualities.

[gscace]

Ok - so I'm the poor schmuck who decided it would be a Good Idea to convert their Isomac Rituale from 48w Ulka vibratory pump to an externally mounted rotary pump in a quest for more Platonically Perfect espresso...

Which conversion appeared to have worked well until yesterday, when a new and most disturbing symptom has surfaced. Now, from time to time, the boiler-fill logic appears to enter some sort of psychotic fugue state such that when the external pump is running to pressurize a shot, water is also being sent to the boiler as well as the HX loop.

This excess water rapidly overfills the boiler, the boiler pressure rises to 2.5 Bar, the safety valve on top pops open and water gushes out of it until the pump is shut off.

As near as I can trace the circuits involved, it looks like there is a level sensor on top of the tank near one end of the HX loop, it feeds a green wire which goes into the Gicluer control box mounted underneath the boiler. The microswitch which is activated by the grouphead lever also feeds this control box, so I am assuming that this box enables the logic function which says "Send pressurized water to the HX loop, not the boiler, during extraction" - but I've not yet discover the solenoid valve which performs this routing.

I believe that the boiler-fill sensor and circuits are working correctly, since there are boiler-fill events lasting about 15 seconds occurring roughly every two hours which is approximately what the machine did with the Ulka pump in circuit, and the boiler does not appear to overfill.

My working hypothesis is that the rotary pump is too big power draw for the Gicleur's internal relay to handle, so not enough amps are left for the Gicleur to effectively control both the pump and operate the solenoid valve simultaneously - since the solenoid valve is normally open to the boiler, and must be closed by the Gicluer controller - failure to do so means that the incoming pressurized water has open paths both through the HX loop and the grouphead, and through the boiler itself.

So, throwing myself upon the assembled wisdom of the group, would it make sense to use an inline relay to control the external pump?

This would be easy enough to do, just give the pump its very own 110VAC plug and then switch one leg of the AC circuit through a 110V 20amp SSR to control the pump and reduce the load on the controller.

Or, have i missed something stupidly obvious in all of my fussing around?

Cheers

Jim (grouchy without his morning triple)

Somewhere in your plumbing there has to be a solenoid valve that diverts water flow from the pump into the boiler, rather than to the group. Disassemble and clean the solenoid valve that prevents your pump from filling the boiler when the pump is activated. I'm guessing that your real problem is that you got dirt in the solenoid valve and that it's not closing as a result. You ought to be switching your rotary pump with a relay as well.

-Greg

[another_jim]

I once contemplated using a relay to knock out the boiler refill solenoid while the pump was activated for brewing (have the levetta activate a relay, and run the pump via the NO, and the solenoid via the NC; the Giemme pump switch is in parallel with this switch, so will not get jammed)

I never did it, because I always finish the shot before the boiler refill is activated from steaming. If you refill switch is activating at any time except after steaming, you have a leak somewhere and it's best to seal it rather than having the refill activating at random intervals.

[mike]

There's no reason why the rotary has to remain a "twitchy, vile handling hot rod." Put an adjustable needle valve ("gicleur") such as mcmaster.com #4555K13 or #7810K11 directly after the pump so that you can regulate how fast the pressure comes up. Adjust the needle valve so that the volume of water that issues from your group with NO portafilter installed measures 70-75ml in the first 10 sec after turning on the pump. Perform this calibration with care. It will make your hot rod docile without ruining its good qualities.

Andy helped me with this aspect of my Silvia conversion to rotary (along with a bunch of other mods like the grouphead heater). Adding in the needle valve to fix the debit had tremendous impact on my shots, including significantly reducing pits and fractures. I sort of wish I still had the vibe Silvia I sold to measure the water debit and pressure buildup using the vibe. It would be impossible to determine which of the many mods had the most significant effect, but adding the gicleur had a very noticible effect.