Gicar flowmeter accuracy

Need help with equipment usage or want to share your latest discovery?
User avatar
AssafL

#1: Post by AssafL »

Flowmeter specs are pretty remarkable. Upon examining the Gicar catalog one would expect it to be possible to calculate the volume of water that flows to the puck using the flowmeter. Here are the relevant pages from the catalog:




If indeed true, and accurate enough, it might be possible to add a shot weight scale, an Arduino Uno and replace the rather mundane "volume" buttons with a "brew ratio" buttons. (Wouldn't it be nice to have a Ristretto, Normale and Lungo buttons instead of the "one or two small cups whatever the dose happens to be" buttons? Especially in the mornings when one doesn't want to necessarily mess around with accurate scales but still wants a decent cup.)

So naturally, to test, I hooked up the flowmeter in the GS/3 and a Sartorius Scale with the YADAP-RS RS232 output option to an Arduino to test this assumption. I used a needle valve to adjust the flowrate and used a scale to measure the water deposited from the group.

The tests were cut short as flowmeter performance turned out to be a rather big disappointment.

Here are the figures:
Test Flow rate (ml/min) Weight measured Flowmeter Pulses ml/pulse
Low flow rate 205 ml/min 106.0 grams 57 1.85ml/pulse
High flow rate 472 ml/min 102.4 grams 31 3.3ml/pulse

These results doesn't appear to validate the Gicar specs of +-2% accuracy with better than 0.2% repeatability (as per the Gicar catalog above). Also, looking at the Nozzle specs, it would seem to me LM chose a nozzle size of (probably) 3.7 mm which would be strange since Gicar recommends to a value centered around the expected flow. At 500ml/min one would assume Gicar would recommend and 0.7 to 1mm nozzle.

A few questions:
Did anyone try running similar tests and can corroborate the results (or hopefully) get better results?
Does anyone know the nozzle size for the Gicar Flowmeter in the GS/3? Are these nozzles replaceable?

Thanks,

Assaf
Scraping away (slowly) at the tyranny of biases and dogma.

User avatar
AssafL (original poster)

#2: Post by AssafL (original poster) »

Flowmeter back:


Gicar part number seems to be 9.0.95.05G.

NSF have it listed as an RoHS compliant part:

Nickel Plated Flowmeter without LED[1] [3]
9.0.95.(1)G

[1] Suffix (1) is a two-digit code which denotes features of electrical connection, output
configuration, nozzle diameter and/or type of thread (G1/4 or G1/8).
[3] Suffix G indicates compliance to 2011/65/EU RoHS II Directive.

What does the 05 part of 9.0.95.05G represent?

A hunch: Looking at other Gicar part numbers I think 0 is G1/4 and 5 is 1.8 mm (assuming 0-0.7mm; 1-1mm.... 5 would be 1.8mm).

NB - I do apologize for the amount of dirt that seemed to have accumulated at the bottom of the espresso machine. One should learn never to focus one's camera at soiled parts of one's kitchen.
Scraping away (slowly) at the tyranny of biases and dogma.

Shife

#3: Post by Shife »

What I've found with the flowmeters in my machine is that they are very accurate provided the machine hasn't sat for a long period. I don't trust them for my first shot because if I forget to flush the group, the resulting weight in the cup will not be the same as the second shot. The other factor that becomes a problem is changes in the beans will produce changes in end weight. The flowmeters are plenty accurate, but the medium we are working with is quite variable. Once I set the volumetrics up, if I pull 5 shots in a row they will all be extremely close in weight (usually within a gram.) If I only pull one or two shots per day, I'm better served by using a scale under the cup and weighing my shots.

If you were pulling multiple shots per day, then setting up a volumetric machine to do as you suggest is rather easy. The addition of a scale would be redundant, however it could be used as you suggest with software to augment the flowmeter. If water had drained from a line or evaporated due to inactivity the software could command the pump to continue running until desired weight is reached. I suppose the software could use the scale to react to changes in shot weight and adjust the volumetrics accordingly. Although the more I think about this the more I'm left wondering what real use the flowmeter would be if you had a scale reading the direct weight of the cup and feeding that info to a computer controlling the pump.

I pull 40 grams of liquid weight from a 20 gram dry puck. It obviously takes more than 40 grams of water to produce that result. How much? I don't know and I'm not sure I have the desire to try and find out. I've not yet gotten to the point where I'm taking TDS readings, but this method has allowed me to have real data when making decisions on what to change and why. It is neat that a lot of these newer machines have enough sensors already on them that hooking up a pc and getting creative could really be interesting and fairly easy.

goodboyr

#4: Post by goodboyr »

This type of flowmeter is sensitive to inlet pressure variations. So you should repeat the experiment with inlet pressure measurements. I believe your method of varying flow also varies the inlet pressure at the flowmeter. That could account for your errors.

User avatar
AssafL (original poster)

#5: Post by AssafL (original poster) replying to goodboyr »

Good catch. (and really the sort of catch I was hoping for posting here).

Indeed that is what might be happening since I adjust the flow before the flowmeter. Indeed have to rerun the tests when I get a chance.

Does challenge the usefulness when preinfusion is involved, since the pressure during preinfusion is at least 6 bars less than during the pull.
Scraping away (slowly) at the tyranny of biases and dogma.

User avatar
AssafL (original poster)

#6: Post by AssafL (original poster) »

Shife wrote:If you were pulling multiple shots per day, then setting up a volumetric machine to do as you suggest is rather easy. The addition of a scale would be redundant, however it could be used as you suggest with software to augment the flowmeter. If water had drained from a line or evaporated due to inactivity the software could command the pump to continue running until desired weight is reached. I suppose the software could use the scale to react to changes in shot weight and adjust the volumetrics accordingly. Although the more I think about this the more I'm left wondering what real use the flowmeter would be if you had a scale reading the direct weight of the cup and feeding that info to a computer controlling the pump.
True if I dose accurately. But in the mornings, I prefer to not dose accurately. So in that case having shot weight and total weight (volume) of water should allow me to calculate Brew Ratio (to some degree of accuracy).
It is neat that a lot of these newer machines have enough sensors already on them that hooking up a pc and getting creative could really be interesting and fairly easy.
Yup - and in the case of the flowmeter, a rather expensive one that does rather little for anyone except a busy café...
Scraping away (slowly) at the tyranny of biases and dogma.

ira
Team HB

#7: Post by ira »

I don't know that it applies here, but in a lot of measurements, the only thing that matters is repeatability, not accuracy. I don't think that in the context of a commercial machine anyone cares about the actual volume per click, they only care about the number of clicks to make a cup of coffee. So the 2% might be accurate if it always flows the same amount of liquid at the same rate of flow.

I've measured and calibrated a large number of mass air flow meters for cars using a home built reference flow meter. No idea what it's output actually means, but I do know that it repeats really well which is all I need for what I do.

Ira

Shife

#8: Post by Shife »

AssafL wrote:True if I dose accurately. But in the mornings, I prefer to not dose accurately. So in that case having shot weight and total weight (volume) of water should allow me to calculate Brew Ratio (to some degree of accuracy).
This is where I think it will get tricky. If I set up my machine to deliver a 40 gram shot of espresso, it will do that all day long.... for that day. The problem is as my beans age, the output changes. It stays consistent, but it is no longer 40 grams in the cup. I have to periodically reset my volumetrics to maintain that weight as my beans age. In a busy cafe the bean freshness is stable. In my home I'm using beans that could be anywhere from a few days post roast to a few weeks. I'm not sure how you would account for that as brew water in/out is influenced by factors that would be difficult to control for.

User avatar
AssafL (original poster)

#9: Post by AssafL (original poster) »

ira wrote:I don't know that it applies here, but in a lot of measurements, the only thing that matters is repeatability, not accuracy. I don't think that in the context of a commercial machine anyone cares about the actual volume per click, they only care about the number of clicks to make a cup of coffee. So the 2% might be accurate if it always flows the same amount of liquid at the same rate of flow.
True - but if I want to use the flowmeter and scale to calculate the weight of the water absorbed by the puck I have to convert ticks (pulses) to milliliters (or Ounces).
Scraping away (slowly) at the tyranny of biases and dogma.

User avatar
AssafL (original poster)

#10: Post by AssafL (original poster) »

Shife wrote:This is where I think it will get tricky. If I set up my machine to deliver a 40 gram shot of espresso, it will do that all day long.... for that day. The problem is as my beans age, the output changes. It stays consistent, but it is no longer 40 grams in the cup. I have to periodically reset my volumetrics to maintain that weight as my beans age. In a busy cafe the bean freshness is stable. In my home I'm using beans that could be anywhere from a few days post roast to a few weeks. I'm not sure how you would account for that as brew water in/out is influenced by factors that would be difficult to control for.
Yes - that is for single point measurement (volume or weight). If I have two (volume and weight) and time of pull, then everything else can only be detected by flavor, observing blonding (or perhaps TDS to calculate Extraction Yield).

Weighting the dose and weighting the shot is rather easy.

But it is easier to weigh nothing (actually have it weighed automatically) and have an Arduino calculate everything for you (and perhaps stop the pull. It might be possible, which is as good a reason to try it to see if it does work.
Scraping away (slowly) at the tyranny of biases and dogma.