I'm thinking of adding a couple of electronic sensors to my Gaggia Classic Pro. In the short term, I'd just like to use these to collect data, so I'll probably just connect them to a microcontroller. In the longer term, I'd like to incorporate these into a control loop for the vibration pump. Once I've got electrical signals to work with I'll be good to go--that's the easy part of things for me.

At the moment, the Digmesa Nano Brass (https://shop.digmesa.com/wp-content/upl ... GB_V05.pdf) is looking like the best choice for measuring flow directly. If you know of something better, I'd love to hear about it.

For the pressure sensor, I'm not sure. Lots of options out there that aren't necessarily food-safe, but I'm having a harder time finding something appropriate for this application. If you know of anything at all, please let me know!

Finally, I'd like to add both of these after the pump, and I'm not sure of the best way to do this. The Digmesa Nano Brass has G 1/8" female connections, so I can add whatever fitting I want to this, but I'd love to hear from anyone who's tried something similar what works best. I suppose big factors here are ease of installation and maintenance, and compact form factor.

Thanks!

The Digmesa Nano Brass is a good choice -- high resolution and a very small form factor for fitting in tight spaces. I tested the same model in my GS/3 and it met the accuracy specs (+/- 2% from 35 ml min to 700 ml/min.)

I use the flow meter in my machine for three things: 1) to set the preinfusion flow rate with a needle valve, 2) to control the flow rate during the back half of the shot (i.e., keep it from accelerating as the puck loosens), and 3) for testing and troubleshooting (e.g.,"is the free-flow rate what it's supposed to be?")

I don't do volumetric dosing, but I'm sure the performance of the two flow meters is about the same for that -- mostly on target but sometimes a little off due to variations is the age of the coffee, grind, dose, puck preparation, etc. Better to use a scale with Bluetooth capability and cut the shot by weight. That's what my automation setup does.

Both flow meters are fine for setting the preinfusion flow rate, which for the Slayer-like shots I do is in the 100-180 ml/min range. Both live up to the +/-2% spec in that range. Both are also fine for testing and troubleshooting, which tends to take place in the 100-500 ml/min range on my machine.

However, neither flow meter works well below its rated linearity range, which is 35-700 ml/min for the Nano Brass and 32-1400 ml/min for the Gicar. I verified with an oscilloscope that the impellers tend to stall below about 30 ml/min. I suspect that water just slips past the fins when this happens. That make both flow meters problematic for keeping the flow rate steady for the 7g singles I usually pull. I try to keep the flow rate in the 20-25 ml/min range, but often the flow meters are stuck at zero or the flow rate they report is much lower than the actual flow rate (a lot of zeros getting mixed in with the occasional pulse.) I verified this by coding the scale reading routine to calculate and graph the flow rate based on the weight. There's some noise on that signal as momentum and bounce of droplets can make the scale momentarily spike up or down, but the readings are in the right ballpark and can be smoothed by code or by eyeballing the trend on the graph.

As there wasn't a lot of difference between the performance of the flow meters for my applications, and the 20x flow rate of the Nano Brass disabled the volumetric dosing feature on my GS/3, I removed the Nano Brass and went back to the Gicar. I only used the Nano Brass for about a month and plan to list it for sale on HB. If you'd like to buy it PM me.

As for connecting a flow meter, I put off trying the Nano Brass for years because I didn't want to fabricate custom copper tube connections. But I found the answer when I read our member Jake_G's post on Simple Profiling on a La Marzocco GS/3 AV. Jake came up with an inexpensive and simple way to install a needle valve for flow profiling using plastic 3D-printer tubing, quick-connect fittings and a few adapters. Check out Jake's post -- it has photos and a parts list. Initially, I used the connection method for the needle valve Jake recommends and a solenoid bypass valve so I can do Slayer shots. Later I replaced the needle valve with a high-resolution metering valve. Recently, I used Jake's connection method to test the Digmesa Nano Brass.

On the pressure transducer, I used a Huba 501, which I believe is the same transducer as the one in the La Marzocco Strada. I think I paid about $80 for it on eBay, but I can't recall if it was new or used. My pressure transducer is connected directly to the output of the gear pump that replaced the GS/3 stock rotary pump. The flow meter is after the pump, too.

I have Arduino code to sample both the flow meter and pressure transducer. It's pretty straightforward, but I'm happy to share it if you'd like to save some work. The signals are a little noisy, so averaging is required to avoid readings that bounce around.

Wow! Thanks for all the information.

Do I understand correctly that the issue you had with the DIgmesa Nano Brass was just that it wasn't compatible with the signal the GS/3 was expecting--like, for a custom project it wouldn't present an issue?

The Gaggia uses sheathed 5 x 8.9 silicone tubing secured with an Oetiker clamp, as shown here:



I suppose one option here would be an Oetiker clamp on that end, and a G 1/8" quick connect fitting on the flow meter. Anyone have experience with Oetiker clamps and PTFE? I'm having a hard time finding official commentary.

You may find interesting what components are used with the Leva!/ito kit which is installed on the GCP.

https://www.softwareandcircuits.com/div ... ure-v2.php

http://projectcaffe.bplaced.net/features_leva.html

Could be the same parts, didn't check, just connecting some dots on my way past.

Interesting. Looks like the first link is using a Digmesa FHKSC flow sensor installed on low pressure side of pump and a Delphi automotive 300 psi pressure sensor. Second link seems to be using the same flow sensor, and could be a similar pressure sensor as well. They're both fairly non-specific about things like fittings, but seem to be using PTFE tubing and barbed fittings--no clamps shown, but they would clearly be needed without quick connect fittings.

crwper wrote: Do I understand correctly that the issue you had with the DIgmesa Nano Brass was just that it wasn't compatible with the signal the GS/3 was expecting--like, for a custom project it wouldn't present an issue?

That's basically correct. The Gicar produces one pulse for every .55 ml of water flowing through it. The Digmesa Nano Brass impeller produces one pulse every .025 ml, which is about 20x the pulse rate for the same volume of water. The GS/3 CPU counts the pulses whenever the machine is in brew mode.

Earlier I said this throws off the volumetric dosing. But all that does is play back the recorded number of pulses, so the higher pulse rate might not be a problem. But there could be an issue with the number of pulses produced by the Nano Brass overflowing whatever variable the GS/3 is using for the recording. I didn't test that because I don't care about volumetric dosing (though if I ever sell the machine the buyer might care and I'd have to reinstall the Gicar.)

The bigger issue I faced was that the GS/3 also counts the pulses to determine if the machine is "running away", i.e., the user forgot to turn off the brew button or maybe there's a leak or solenoid failure. So, after a certain number of pulses, the GS/3 shuts of the brew cycle (it also does that if a 60-second timer has expired, but that's another issue I was able to overcome with a firmware change.) Consequently, the brew cycle only runs a few seconds before the GS/3 shuts it off, obviously a huge problem.

Previous to installing the Digmesa Nano Brass, I connected my Arduino optoisolator interface to the Gicar in parallel with the GS/3. That way, the GS/3 could do volumetric dosing and my Arduino could show the flow rate. But when I installed the Nano Brass, I had to disconnect the flow monitoring circuit of the GS/3 and connect the Nano Brass only to the Arduino. That worked, so yes, the Nano Brass is fine for a custom project that doesn't connect it to the machine's CPU.

However, there was one downside for the GS/3: when brew is on and no pulses are counted, it displays the error message "ALARM FLOW METER". Normally, it only displays this message when the grind is too fine and the machine is choking or something is wrong with the cold water input. The machine still worked for my purposes, but I wasn't able to see the shot timer, temperature or boiler heating statuses during a shot. My Android/Arduino setup provides it's own shot timer, but I found the loss of the GS/3 display during a shot annoying. Hence, with the Digmesa Nano Brass flow meter offering no real advantage over the Gicar, and causing some issues with the GS/3, I went back to the Gicar.

Hard to say what the best approach is for connecting the pressure transducer and flow meter to your machine without knowing more about the water flow of your machine and the exact type of fitting (barbed or smooth) the sheathed tube is connected to. I'd be tempted to forego PTFE and quick connects, and put the pressure transducer and flow meter inline with the sheathed tubing. You'd need adapters with a hose barb or smooth tube on one end and an appropriate threaded connector on the other end (e.g., male G 1/8" for the Nano Brass.) You'd have to cut the sheathed tubing in a couple of places and get four or five clamps. You'd need a tee connector for the pressure transducer. A short piece of the sheathed tubing would be needed to connect the pressure transducer to the tee, but that's OK because I think you'd need to shorten the sheathed tube a bit anyway. If you go this route, best to get a replacement tube to keep on hand if you decide to remove the mod and return the machine to stock condition.

[EDIT: I started out with an inexpensive pressure transducer like the one shown in the second link. I can't remember whether it had significant accuracy issues. It might have been reasonable for experimental work. I replaced it with the Huba because the price was reasonable and I wanted to make sure the readings were accurate. Besides, if it's good enough for the Strada, it's good enough for me!]

If you are already using an Arduino, you could just produce an signal for the GS/3 to simulate the old one.
You use the Digmesa in your machine and count with the Arduino to show your flow. Every 22nd count (0.55/0,025) you also trigger a pin, which is connected to the sensor connector from the GS/3. Then you could have both.

crwper wrote:Interesting. Looks like the first link is using a Digmesa FHKSC flow sensor installed on low pressure side of pump and a Delphi automotive 300 psi pressure sensor. Second link seems to be using the same flow sensor, and could be a similar pressure sensor as well. They're both fairly non-specific about things like fittings, but seem to be using PTFE tubing and barbed fittings--no clamps shown, but they would clearly be needed without quick connect fittings.

They're links to the same thing on different sites

Please take a look at this project's wiki.

Pflunz wrote:If you are already using an Arduino, you could just produce an signal for the GS/3 to simulate the old one.
You use the Digmesa in your machine and count with the Arduino to show your flow. Every 22nd count (0.55/0,025) you also trigger a pin, which is connected to the sensor connector from the GS/3. Then you could have both.

Great idea. I never thought of that. However, I don't have any wires left in the 8-wire Ethernet cable that connects the Arduino to my GS/3 interface board. I could add a second cable, but that would be kinda ugly. Not worth the effort. Besides, I found that the Digmesa stalls at a higher flow rate than the Gicar in the region where I need to control the flow rate during the back end of the shot.