Using electronic scales for running gravity reading

[ArgM]

Does anyone here have any experience with using a Ultrasonic Transducer?

"Ultrasonic waves are passed from a source, through the fluid of interest, and into a detector which measures the acoustic spectroscopy of the waves. Fluid properties such as density and viscosity can be inferred from the spectrum." http://en.wikipedia.org/wiki/Specific_gravity

There also seems to be a few Arduino compatible sensors and tutorials on setting one up... It seems like this would be a good way of doing, only downside will be translating the information into SG which Im sure will become easier with time

 

[MastersBrewery]

I would think once you had spectrum differential across a sample of differing gravity "worts" a simple equation would spit out the SG all done on the arduino

 

[ArgM]

True, this is looking like the way to go, found the sensor online for $6 so I reckon it's worth a shot!

Thanks everyone for your ideas and experience!

Going to start a post documenting my process as I go through and put it all together, so if you wanna check out how it goes look that up!

(probably get started in a week or so)

 

[MastersBrewery]

I did see a link to a paper on the subject of ultra sound and liquid density, they wanted $35 for me to read it you may find you will need to play with a range of frequencies and returns to get one reading. Be very interested in your results, don't forget temperature variance.

 

[MCHammo]

I started working on something to do exactly this; measure wort density using ultrasonic sensors. I got most of the way there, but gave up at the end for several reasons (one of which is that it isn't really necessary)

1. You need the longest possible path through the liquid to take measurements, as the time difference between two different densities is incredibly small. On that note,
2. You need very accurate, high resolution timing to even detect the differences.
3. The sensor will need to be immersed in the wort to accurately measure the path through the wort. The sensors need to be resistant to water, alcohol, etc. (such sensors do exist)
4. Some US sensors (particularly these resistant ones) need a high driving voltage to operate. This presents its own problem with regards to driving circuitry (and propagation time through the circuitry). The sensors I tried needed 140Vpp to operate properly (and work at 40kHz). For that, I needed to construct a transformer to step 12V up to 140V.
5. The ultrasonic waves that you intend to send through the liquid will also travel through the container. They do so faster than through the wort, so you need to do some very smart filtering
6. The US waves do not like to reflect off a surface in the FV (at least in a plastic FV), so you need two opposed sensors.
7. Fermentation activity inside the FV will affect the time of flight of the US waves - suspended yeast, CO2 bubbles, turbulence, etc will throw off the readings.
8. As mentioned, temperature will affect the density of the wort. That's a relatively easy thing to measure and calibrate for, though.

I might have missed a few things, that's just off the top of my head.

Masters, can you give us a link to that paper? With any luck, my uni has a subscription to that publication and I can have a read of it myself

 

[MastersBrewery]

MCHammo, it was a late night of trawling a week or 2 back, will see if I can retrace my steps to night

 

[TheWiggman]

My experience with ultrasonic transducers is with solids, specifically thickeness measurements. MCHammo has covered everything I might be able to go on about - the variable you will be measuring is density. Typically for industrial application the speed of sound through a solid will range from 1500 - 9000m/s. You can get pretty accurate measurements off the time it takes to reflect the signal, which for a ceramic 5mm thick can be around 6 E -7 seconds. They will measure accurately to within +/-0.1mm if calibrated correctly. Albeit with a unit that costs thousands of dollars.

Water is around 1400 m/s (temp dependent). You could back-calculate from your result knowing the bulk modulous and time it takes to reflect off something in the wort. The biggest variance, in my opinion, will be the target you will be reflecting the sound off as it this isn't REALLY accurately set up your results will be inconsistent and innaccurate, unless you knew the SG and temp to start with and tell the controller 'this is equal to this SG'. The distance could then be calculated and the rest of the work done from there. Of course like MCHammo said your sensor would be immersed which is not ideal.

Otherwise, for much more effort, you could place the target (say a stainless plate) in the bottom of the wort and have the sensor pointing at the solution from just under the lid. The biggest issue you'll face there is massive reflections from the suface of the water which will have to be somehow ignored, which is typically done using some sory of clever algorithm or graphing the results and determining which reflections should be considered. You then have two things to account for, being air and wort. And krasuen too I suppose.

I think MHB's suggestion is the best. There's more than one way to skin that cat.

 

[MCHammo]

The setup I was going for two sensors (one Rx, one Tx) set in the walls of the fermenter, one on either side. That takes out a lot of the uncertainty of reflections, but reduces the time available to take the measurement (with a reflection off a solid surface, you've got about twice the distance - and hence time - to measure). What I was planning on doing was making a ballpark measurement of the ToF (Time of Flight) of the waves through the wort, and filtering out anything that came outside of those bounds (waves through the fermenter wall should arrive much sooner, reflections through the wort should arrive later). Once you establish a reading of the time, you can use that as an estimate of the time the next measurement should take (as the SG of the wort will change very slowly wrt the frequency of measurement). Setting up a band-pass filter on your Rx amplification circuitry works a treat, too.

By using submerged sensors, you take things like krausen out of the equation, assuming that the sensors are sufficiently submerged in the wort. The only problem that I see with MHB's solution, is that a decent load cell (last time I checked) was very expensive. You also don't know how the krausen and top-cropping yeast are going to affect the measurement.

Another approach we thought of was very similar to a refractometer - having a prism built into the side of the fermenter, and using a laser and sensors to make readings. The main problem for us, was one of resolution. We would have needed a very high resolution ($$$) sensor to pick up our laser beam, or a prism of very high refractive index (nothing we could find was high enough - and not opaque). We scrapped this idea, as it is pretty much impossible to do on a budget. The US approach is much more dfficult (and in the end may not even work), but is a much much cheaper option. And of course, as this thread has shown, there are many other options out there, too.

 

[ArgM]

So, the sensor sounds like its more hassle than its worth, probably still gonna by one cause it sounds pretty fun to play with.

Ive got a really rough concept of what I came up with in my musings today about it.
This is kind of what I'm thinking of doing but not savvy with different sensors and how to use them so if someone could help with that it would be appreciated.

So, you have a float gravity reader like the hydrometer with an open end, this is then held and floated in the keg, it is able to move up and down. Into the open end of this hydrometer
is a rod, this road is fixed to the roof of the vessel and cannot move, as the gravity changes the floating vessel will move up and down this rod which through the use of a sensor will be able to
provide a gravity reading once calibrated.

also,

 

[ArgM]

No Rod fans around?

 

[MHB]

Actually its nearly what I suggested above, measuring the change in bouncy, except that you have added the extra complexity of moving parts where liquid could get in and change the mass of the float, you have to account for friction between the rod and the tube making up the stem of the hydrometer and any mass of dried on krausen that sticks to the tube...
You haven't suggested how you will measure the location of the "hydrometer" in relation to the rod and how any change in the total volume of the liquids will be accounted for (yes the volume will change - you cant take our a couple of Kg's as CO2 without something changing) as that will change the reference height from where you zeroed your readings
A more complicated and less accurate way to achieve the same thing - no not a fan of the Rod.
Mark

KISS

 

[MHB]

In line digital Refractometers have come down in price dramatically, last time I looked at them they were around $14K now they are only about $3K.
If you had the budget they are really would be a brilliant choice, offering the ability to measure the mash, boil and with a bit of output manipulation the ferment gravity, here are some https://www.misco.com/refractometers/inline-process-Refractometer
Mark

 

[ArgM]

3k is far beyond my reach at the moment haha, I'm having trouble scraping together for a burner!

I put my head together with a friend over a few beers with this the other night and we had an idea that seems like it's worth testing:

if the movement is consistent or a pattern is noticeable against gravity readings, it could be calibrated to turn the movement from initial position and change in velocity through liquid to give SG

An adaption to this would be to add a Flex sensor to the arm.

Now, there will be volume change due to C02 however, keeping in mind this is a domestic brew with volumes never larger that 25L, I was wondering how well a bubble counter could account for this change?

 

[TheWiggman]

Not sure how this would work. An accelerometer will measure acceleration ie rate of change in velocity. To provide a reading it would need movement. What's going to make it move? Also accelerometers only read in one direction, and as it rotates the centripetal force will play around with readings and make them non-linear.
Additionally, viscosity is the resistance to movement. Not sure if viscosity is consistent between brews but if different this is another external variable to account for.

 

[hotmelt]

What if you were to measure the conductivity of the wort.Wouldn't that change as it ferments?

 

[ArgM]

Hey Wiggman,

It would be in a casing which would be balanced to float correctly, as the wort changes this balance will change, the accelerometer should move and this movement should give information on speed and distance and that data is what Im hoping to match against gravity readings... That incorrect that they on read in one direction, they read x,y,z axis simultaneously.
Yeah, viscosity could be an issue not only between brews but also in a single brew with heavier sediment falling to the bottom.

I'm not sure about conductivity, Id need to look into it, Id be afraid of electrocuting my beautiful yeast!

 

[Kodos]

I use a hydrometer.
And sleep in a bed with my wife.

Nah, just kiddding. Sounds interesting. I'll be interested to see if this goes anywhere.

Liar!

You're not married.

 

[ekul]

Why not just leave the hydrometer inside the fermenter? I used to do this when i was too paranoid that draining some liquid out would pull 'infected' air into the barrel. Use gladwrap and just look in to monitor fermentation.

 

[ArgM]

Hey ekul,

theres a few issues with that, the first being I want constant readings so I can observe the relationships between the different forces acting on the beer, with a traditional hydrometer it requires human interaction to get the readings, I played around with the idea of a hydrometer held by a sensor which could read the movement, however as MHB pointed out, it doesn't address change in volume from C02 loss and interfered readings from the Krausen. However, assuming from his last post he was picturing use in a much larger scale production where the changes in volumes would be much greater than for domestic purposes, so there may still be something in there if I can create an accurate bubble counter and find a way to account for C02 in between the wort and roof of vessel that may not have escaped, again, for domestic purposes this should be much easier.

 

[hotmelt]

Have a look at homebrewtalk.com http://www.homebrewtalk.com/f235/can-even-done-measuring-specific-gravity-using-arduino-raspberry-pi-385352/ someone
there is trying to do he same.
Or there's these http://www.brewballstore.com/

 

[TheWiggman]

Hey Wiggman,

It would be in a casing which would be balanced to float correctly, as the wort changes this balance will change, the accelerometer should move and this movement should give information on speed and distance and that data is what Im hoping to match against gravity readings... That incorrect that they on read in one direction, they read x,y,z axis simultaneously.

An accelerometer will only read measure along a single axis. The most common types have peizo-electric sensor in them connected to a small mass though a dampener. When something is accelerated along the axis of the dampener, it generates a voltage. Multi-axis accelerometers are available but they consist of multiple individual accelerometers and will have 3 outputs. Smartphones (and similar devices like a Wii remote) have 3 inside them to measure each direction.

The movement in reality will be essentially zero, hence why I was a bit sceptical. You won't find an accelerometer that will measure what you're after. It's more likely to pick up vibrations from walking than movement from the arm rotating a few degrees over a few days.
When you say "balanced to float correctly" remember that unless there are two dissimilar fluids of different density in the wort, your float will either sink or swim. To address that you'd need a spring to stop the arm from rotating vertically, but again you can't really use an accelerometer. If you're going to use this method the floation force would be far simpler (and more accurate) to measure which puts us back to MHB's suggestions. Or maybe a circular potentiometer like a fuel tank, but then we're back in overcomplicated territory.

 

[ArgM]

hmm, my thoughts were the same way displacement works, as the gravity of the liquid changes the location where the vessel holding the accelerometer will change, being on an arm controlling the radius this moves means two outputs are required, I wont be measuring the distance moved, but rather the rate in which it moves. By keeping it somewhere stable (so as to not be affected by passing footsteps or trains as I live right next to a train line). I simply need it to write to readings, an initial and a second reading to compare this against.

So, say we have a reading of 300 initial, then it moves as the wort changes with the yeast we may see it change to 310, then return to 300. by observing these changes it may be possible to see a correlation between that and the SG.

I personally advocated the use of a flex sensor over the accelerometer, but the general consensus was that an accelerometer would be more fun to test haha

sorry, you are right, a multi axis accelerometer is what I meant when referring to multiple readings.

 

[MHB]

Lets call it a thought experiment, what makes the accelerometer move up and down? It is either 1/ lighter than the wort so it floats to the top of the arc and stays there, 2/ heavier, so it sinks and stays there, or 3/ at neutral bouncy so it might float in the middle of the ark, until the gravity changes then it sinks = no data

 

[ArgM]

How would there be no data from 3/ the gravity doesnt change immediately from OG-FG it will fall slowly, the moments of falling will create velocity and will give a data reading which the sensitivity will need to be calibrated to or am I missing something?

Those brewballs looking interesting and look like a solution that could work, I would only get stepped readings with those, but it could be possible to estimate the rate of change depending on the time between each ball falls

 

[MHB]

Explain how this works and I will agree that you understand buoyancy

 

[dent]

Took me a while to work out you were talking about buoyancy "wtf is bouncy?"

An accelerometer can measure a static angle well enough, based on the relative pull of gravity along an axis. The common ones available in digital form cheaply nowadays can produce an accuracy down to 0.3 degrees or so, in the units I have built.

 

[glenos]

A very quick skim of the Brew bug webpage gave me this idea, what about a mass (ideally large surface area and small mass) suspended in the wort attached to a load cell? As the density of the wort changes the mass will float.

Or flaot your hydrometer in the wort and use a camera to capture regular images and process them to interpret the scale.

 

[MHB]

Never claimed I could spell, specially later in the day spell checkers tend to get more an more retarded
Mark

 

[ArgM]

I watched the video, I have no understanding of buoyancy... from the top rated comment I gather, whilst the sand is in the top it want to fall side ways which is inhibited by the tube, when sand falls it is no longer top heavy allowing it to rise, the one with sand already at base sinks because it is top heavy.

Now, since we both agree that I dont understand buoyancy it would be nice for an explanation how an object that remains bottom heavy will not work give readings of velocity as the liquid around it changes? I don't know about you but I dont regularly turn my fermenter upside down...

I think Im going to continue with this and the brewballs (going to try and make my own)

Remember, I'm new to all this, I'm studying Industrial Design, not engineering or science, I'm simply trying to find a way to do this an proposing experiments I think may work. I'm here to learn, not to say I have the answers to everything.

 

[schrodinger]

...if I can create an accurate bubble counter and find a way to account for C02 in between the wort and roof of vessel that may not have escaped...

Only around ~0.2% of the total CO2 from fermentation will be in the headspace (assuming wort with 4 kg DME, 5L headspace, 18 deg C, 100% CO2 in headspace, 80% attenuation).

- 4kg DME = 133 mol CO2... 80% attenuation = 106 mol CO2
- n/V in headspace = 0.04 mol/L of air... 5L headspace, 100% CO2 => 0.21 mol CO2