Moderator Note: Discussion split from If La Marzocco makes a GS3 V2, what changes would you like to see?

smite wrote:I am intrigued by your mod option for the GS3MP not to derail the thread but could you explain the benefits this would provide besides less water waste and the best way to achieve?

John,

MP mods would be identical to my Simple Profiling on a La Marzocco GS/3 AV. The MP (Strada mod or Conical valve) is a good pressure profiler, but has rather poor resolution, as you point out. The needle valve makes it a flow profiler in addition to the inherent pressure profiler that the paddle gives you. I do not yet have a mechanical solution in mind to use the paddle to control flow, but it is always in the back of my mind. In terms of benefits, flow profiling through the needle valve has much better resolution than pressure profiling by dumping to the drip box.

michael wrote:Gear pump

Indeed. Perhaps a GS/3 EP added to the lineup?

Jacob wrote:Timed steaming - I scream, you scream, we all scream for eSteam

Automated (not programmed) flow profiling

Holy smokes! Its Jacob!
I read most of your early GS/3 posts when I found out I was getting mine, but I can't believe I didn't latch on to your early profiling posts when reading up before embarking on the ramble. I read all the TMFR stuff, but I didn't have a GS/3 back then, so I must not have really digested it all that well. Nice work! Did you ever post your code anywhere for automated flow profiling?

JonF wrote:I would move the controls so they were not under the hot-water wand . . .

**Sigh**
I'm with you, but where? Swap sides with the gauges?

michael wrote:So when's the new GS3 coming; doesn't feel like it's a soon thing

I honestly have no idea. Programmable PI is coming to LMLM, but I wonder if LM considers the PI that we have on the GS/3 AV to be "good enough". I don't. Even pump on, pump off would be a major improvement to what we have now.

Thanks for keeping the ideas coming.

Cheers!

- Jake

Jake_G wrote:

• Needle valve profiling. Preferably with a paddle activating a potentiometer and remote actuation of the needle valve to allow flow based, pressure based or valve position based profiles.

Are there electronically actuated needle valves available off-the-shelf which doesn't cost a small fortune and/or isn't incredibly large?

Anyone tried purchasing the valve designed by Dalla Corte for the Mina as a spare part? If so was it available and at what price...

Found this proportional solenoid valve used for the steam wands on the Strada EP. Guessing it has nowhere near enough resolution/precision for it to be of any use in direct flow profiling, but perhaps it could provide some form of control if used for the bypass to a Slayer style needle valve...

pcrussell50 wrote:Other than the fun of watching automation, what exactly is the benefit of a servo driven needle valve versus manual?

Making the valve action physically independent from the location of user input. I can only speak for myself, but if I were to get into realtime unlimited flow profiling, I would not want to be reaching around a hot machine to fiddle with a valve. I would want a fully integrated user experience with say a single paddle actuator on the front of the machine controlling all aspects of the extraction.

Whether it is actually worthwhile, well that is a wholly separate question. I suspect for my own use I would be perfectly happy limited to the Slayer-style with control narrowed down to a single restricted intermediate step. Still require it to be integrated into a single neat actuator though!

Jake_G wrote:Nice work! Did you ever post your code anywhere for automated flow profiling?

Thanks
No, the code ended up being a real mess!
BTW I think the interesting thing about my setup is the automated approach, not the implantation.

My setup have been working great for more than seven years now

Jacob wrote:Thanks
No, the code ended up being a real mess!
BTW I think the interesting thing about my setup is the automated approach, not the implantation.

My setup have been working great for more than seven years now

I followed your posts when I was building the Chimera - what is interesting to me is how you built the loop around flow. Mainly two issues:

1. Stability of the loop (no oscillations etc.)?
2. How to handle hitting compliance limits (e.g. flow is too low due to fine grind. Do you up the motor volts? do you back off? do you switch to PP?).

Now that I revisited the TMFR thread - is your solution closed loop (i.e. receiving feedback from the flowmeter to ensure a certain ml/min) or is it controlling the pump (FG204?) voltage without a closed loop? If it is open loop you can disregard both questions above....

(FWIW - closing the loop proved rather difficult. Even attempting to be consistent - every day the "impedance" of the coffee bed changes... So I ended up with a manual paddle....)

Mayhem wrote:Are there electronically actuated needle valves available off-the-shelf which doesn't cost a small fortune and/or isn't incredibly large?

I am not aware of any. There are proportional valves that will work. But they run a few hundred bucks. Moog makes the E024 sub miniature servo valve for Formula 1. I'm sure it's reasonably priced

Anyone tried purchasing the valve designed by Dalla Corte for the Mina as a spare part? If so was it available and at what price...

I quite like this idea. WLL now carries the Mina, perhaps they have spare parts?

Found this proportional solenoid valve used for the steam wands on the Strada EP. Guessing it has nowhere near enough resolution/precision for it to be of any use in direct flow profiling, but perhaps it could provide some form of control if used for the bypass to a Slayer style needle valve...

You are likely correct on the lack of resolution. Interesting concept on using it in series with an adjustable -but fixed- orifice like the Slayer needle valve. Not sure if it would work or not. I'd have to mull it over. I suspect that you would have a very narrow window between off and wide open, with the needle valve providing your max flow at wide open, so I doubt it would work. It might though. I'd have to see C_v signal response curves to know for sure.

Cheers!

- Jake

Jake_G wrote:You are likely correct on the lack of resolution. Interesting concept on using it in series with an adjustable -but fixed- orifice like the Slayer needle valve. Not sure if it would work or not. I'd have to mull it over. I suspect that you would have a very narrow window between off and wide open, with the needle valve providing your max flow at wide open, so I doubt it would work. It might though. I'd have to see C_v signal response curves to know for sure.

Looking around some more I've found what looks to be promising candidates for flow profiling. Bürkert has a standard series described with "excellent range" and "very good response", as well as a simplified series with "high dynamics" and "good range" but at a lower price.

Both series feature valves with an 0.8 mm orifice and Kvs value for water of 0.018 m3/h, which I am assuming is just a different term for the Cv value as specified for manual needle/metering valves I've looked at? Quick math says 0.018 m3/h is equivalent to 5 ml/s which is very close to maximum flow through Slayer's valve or so I am told. There is a smaller version with a max pressure rating of 12 bar, and larger heavier duty versions (with same size orifice and Kvs) rated for 16 bar. Response curve looks good too, I think...

Could these be useable for full flow profiling? Someone with better engineering credentials than me, please look at the datasheets!

Type 2871 - standard, smaller, max 12 bar
Type 2873 - standard, larger, max 16 bar
Type 2861 - simplified, smaller, max 12 bar
Type 2863 - simplified, larger, max 16 bar
Bürkert proportional solenoid valves

They make their product tables difficult to decipher.

Their response chart is non-dimensional so they can use a single response curve for all of their valves. Their chart plots K_v/K_vs, and K_v is a coefficient used in an equation to calculate Q. Both K values and Q have units of m³/h:


Taking your example of an ideal flow one step further, I would choose a valve that allows a large debit, such as 12 or even 15ml/s at 10 bar pressure drop. 15ml/s in m³/h is .054, which is Q. So, you take the chart and equation above, and first solve for K_v.
K_v = .054 * (960/(1000 * 10))^^1/2 = .0167

Ok

Since the chart goes to 1 @ 100% open, we just need to pick a valve that has a K_vs as close to .0167 as possible.
The 0.8mm orifice gives us a K_vs of .018, as you've found in the table and will get us slightly more than 15ml/s @ 10 bar, but it will be really close to 15ml/s @ 9 bar. Looking at the response curve, if using a 0-10V output, this valve would take about 3V to get you 5ml/s. Using the 15 to 85% rule, it would reliably operate between 1.5 and 8.5 volts and give you water debits between .75 to 14.25 ml/s without too much concern. The blue and black lines represent the hysteresis, so one should take that into account with the control scheme. Generally a response that incorporates slight overshoot will keep you on the blue line, but individual actuators may respond differently based on actual operating conditions.

Overall, I think this would be a good candidate for full flow profiling. But it is an order of magnitude more expensive that just putting a hobby servo on a needle valve...

Cheers!

- Jake