Peppersass wrote:Dalla Corte uses a Digmesa Nano. It's a conventional flow meter with 2% accuracy and high resolution (lower volume per revolution.) Last I looked it was available for about $80.

Yep, looks like samples are available for CHF 78.00 (https://shop.digmesa.com/en/). That's a pretty good price, really, for what looks like a solid product. There's no way the ultrasonic sensors I've found can match this for price/performance. It will be interesting to see what kind of accuracy I can get out of a home-built setup.

crwper wrote: Ideally, my feeling is that the pump ought to be controlled with feedback from a sensor to produce the desired pressure, and the bypass should be there only for safety reasons.

That's essentially how my gear pump works. It doesn't have a bypass (that's an available option, but it must only be used for safety, never to set operating pressure -- otherwise the warranty will be voided.) I have a pressure transducer and flow meter after the pump, both of which are connected to an Arduino that passes the data to an Android phone or tablet app. I manually set the speed of the pump to whatever pressure and flow I want, but that could be automated for profile recording and playback (Assaf did that with his gear pump and Arduino-based GS/3 mod.)

I do have some automation in my setup do to things like a Slayer shot. I hit a button in the Android app, which opens a solenoid bypass valve to redirect the flow through a needle valve that I've preset to the desired flow rate for preinfusion (I started out using Slayer's 50-100 ml/min spec, but later changed to Decent's 180-240 ml/sec.) A preset timer determines how long to preinfuse (I usually set the timer to go off when pressure has reached max -- usually 9 BAR -- for a couple of seconds.) When the timer goes off, the solenoid bypass valve closes and the flow increases to the normal 9 BAR flow rate, which with no resistance would be around 350 ml/min, but with puck resistance tends to slowly rise from zero at the beginning of max pressure to maybe 50 ml/min for a single shot. The pressure declines a bit, too, because the gear pump has no bypass. But it's still too fast by the end of the shot, so I manually ramp down the gear pump speed to drop the flow rate to a more or less constant 25-30 ml/min through the back half of the shot.

I should point out that the low flow rate numbers that my setup reports after max pressure and until the end of the shot are probably not at all accurate, given the poor resolution of the Gicar flow meter in my GS/3. However, they're fine for relative measurement, which is all I need to keep the flow rate mostly constant through the back half of the shot. One of these days I might get a Digmesa Nano to replace the Gicar flow meter.

Since my cheap flow meter is only rated up to 8 bar, I put it before the pump. Its a $10 YF-S401

I also added a "T" into the water line, between the flow meter and the pump. I connected the output of the OPV into this T, so the OPV should be feeding water back into the line, after the flow meter. Otherwise anything coming out of the OPV would throw off my flow readings.

I'm sure this isn't perfect, but its better than nothing.

I admit my flow numbers are kind of strange. I integrate flow into shot volume, and my volume numbers are higher than they should be at the beginning of the shot, compared to weight and what I know is in the cup.
I've weighed the pucks a couple of times after the shots, and pucks are about double the weight, so I also know how much water is absorbed into the puck. I just assume this is pretty consistent now.

By the end of the shot, accounting for the water absorbed into the puck, the volume is about what it should be, compared to weight. So it slowly goes from being too high at the beginning, to being what it should be by the end.

When I was putting this together, I "calibrated" the flow meter by putting water through it. Then weighing the water and comparing to the calculated volume. I adjusted by about 10% based on these measurements.

I really don't use flow for anything, except for logic to decide when a shot starts, and when it ends. I am eyeballing the flow rate which seems to work ok.

I display and record flow, and adjust the pump manually.

Randy

I would sooner measure pressure (and temperature) and calculate flow. The pump will produce a known flow when pushing against a given resistance (thus producing pressure). This is why pump curves showing the flow rate at a given pressure can exist. My guess is that a pressure transmitter will cost less and be far more reliable than an ultra-low-flow sensor.

If you're going for average flow the simplest (and most accurate) measure is by mass and time. The old "bucket and a stopwatch" technique... in this case an espresso cup and scale with timer, but close enough.

mdurepos wrote:I would sooner measure pressure (and temperature) and calculate flow. The pump will produce a known flow when pushing against a given resistance (thus producing pressure). This is why pump curves showing the flow rate at a given pressure can exist. My guess is that a pressure transmitter will cost less and be far more reliable than an ultra-low-flow sensor.

That's an interesting thought. I would be concerned that the pump characteristics wouldn't be exactly the same as the spec, or that they would change over time.

On the other hand, one could characterize the pump using a pressure sensor and a portafilter with a needle valve. Set the valve, measure the flow rate using a scale, measure pressure using the sensor, and repeat for several different flow rates. This could be repeated as part of regular maintenance if there is concern that the pump characteristics will change with time.

mdurepos wrote:If you're going for average flow the simplest (and most accurate) measure is by mass and time. The old "bucket and a stopwatch" technique... in this case an espresso cup and scale with timer, but close enough.

I've been measuring average flow using the technique you mention, but at this point I'm more interested in instantaneous flow rate, with an eye to building a custom controller for my Gaggia Classic Pro.

My thinking has been that we want to maintain a certain flow, not pressure, since pressure itself doesn't seem to have much effect on extraction. But, as you say, since the pump itself has an inherent "resistance", in theory constant pressure is the same as constant flow (until flow stops). I'll have to think a little more on this.

randytsuch wrote:I also added a "T" into the water line, between the flow meter and the pump. I connected the output of the OPV into this T, so the OPV should be feeding water back into the line, after the flow meter. Otherwise anything coming out of the OPV would throw off my flow readings.

I think you're right that this should fix the issue with the OPV. In this case, it seems to me that the only way water can enter the system is through your flow meter. So, if the flow meter were perfect, I think you'd get the right numbers.

randytsuch wrote:Since my cheap flow meter is only rated up to 8 bar, I put it before the pump. Its a $10 YF-S401

This seems like it's probably the culprit. The YF-S401 is specified for about 0.3 to 6 L/min (https://bc-robotics.com/shop/liquid-flow-meter-yf-s401/). Average flow for a typical shot is about 1 mL/s, or about 0.06 L/min, well below the specs for that flow meter. At the very least, it probably doesn't have a very linear response in that range.

Peppersass wrote:That's essentially how my gear pump works. It doesn't have a bypass (that's an available option, but it must only be used for safety, never to set operating pressure -- otherwise the warranty will be voided.) I have a pressure transducer and flow meter after the pump, both of which are connected to an Arduino that passes the data to an Android phone or tablet app. I manually set the speed of the pump to whatever pressure and flow I want, but that could be automated for profile recording and playback (Assaf did that with his gear pump and Arduino-based GS/3 mod.)

I'd love to hear more about your setup. You've got a much better starting point, but my ultimate goal is essentially to gut my GCP and replace all the analog logic with a custom controller. I realize some people might think this is serious overkill for a machine which has some serious mechanical limitations, but for me there's a thrill in getting the best performance I can from that machine--and I think I'll learn a lot about what's important with espresso along the way.

If you're knowledgeable enough to do that, building a driver for a vibe pump that lets you control a DC PWM signal with variable pulse width and frequency should let you do what you need for very little money. Listen carefully to a Decent pulling shots and you'll hear what I mean.

Ira

ira wrote:If you're knowledgeable enough to do that, building a driver for a vibe pump that lets you control a DC PWM signal with variable pulse width and frequency should let you do what you need for very little money. Listen carefully to a Decent pulling shots and you'll hear what I mean.

Oh, that's a lovely idea. I'd been planning just to use a triac "dimmer" controlled by a microcontroller, but retaining the peak voltage and varying the frequency would be much better. I've seen people mention PWM control for vibratory pumps, but never quite put it together. Thank you!

Specified flow rates for meters are misleading, imho

My meter generates 5298 pulses per liter or 1 pulse per 0.189ml

That's better for low flow rates than any of the digmesa meters I've seen. I was looking at the 932-9505-B, but it only does 2382 pulses/l

Not sure how they come up with the min and max flow rates, so I started looking at pulses per liter.

So I'm not sure why my meter is off, unless its not, and the puck is holding more water during prefusion. Wonder if that's a thing.

I also can look at change in weight, so I know how the shot progress. And I currently have no interest in using flow or pressure to control the shot. I wanted to before, but now I'm fine with watching, and controlling real time, so I can make adjustments on the fly.

Randy