Kegbot (flow meter and digital taplist) project

[Moad]

I got tired of kegs kicking at inconvenient times, I had some arduinos laying around and I am a bit of a nerd so thought I would make a kegbot system. It is a small server running on either a raspberry pi, linux machine or a tablet. An arduino board connects to the server and the flow meters connect to the arduino. Basically the arduino reads the signal from the flow meters and sends that to the server.

So far I have soldered the flow meters onto some cable and a 3 pin connector per flow meter (Jaycar)

I've attached a USB connection to the project box to go from the Raspbery Pi (kegbot server) to the arduino. The sockets for the connectors have been fitted onto the box, this should make soldering easier too. The arduino sketch has been uploaded and tested, the raspberry pi loaded with kegbot and tested.

Left to do is finish soldering the sockets and then hook them up to the arduino in the box. Then I need to screw on the barbed fittings to the flow meters and fit them into my keezer. Then run the connectors out of the keezer into the project box and I will be up and running.

I'm hoping to have it all running this weekend with a keezer tidy up happening, more to follow.

 

[megabyte]

Nice! Keep us posted.

 

[contrarian]

Where do the flow meters sit in your system? Any issues with foaming from them? It certainly is one of the most annoying things about kegs! When mine go they tend to go in clumps too!

 

[Moad]

Where do the flow meters sit in your system? Any issues with foaming from them? It certainly is one of the most annoying things about kegs! When mine go they tend to go in clumps too!

I haven't actually poured beer through them yet. I've read that they need to be as close to the keg post as possible so that's the plan. Hoping to get some time this arvo to finish it off

 

[Moad]

Success!

I finished a uni exam early tonight (this morning?) so had some time while the family was asleep to test a flow meter and finish the keezer rebuild.

I connected up the lines, beer poured fine so I grabbed one of the finished flow meters. I hooked up some water in a mini keg and poured through but it wasn't registering in the kegbot server. hmmm.

Hooked up another one I had prepared earlier and success! Registering pours. I still need to tweak the ticks/ml setting to calibrate the flow meter but it works a treat. No foaming issues although I am running an extra .5m on my lines so it could counter the foaming. In any case, I am stoked! Will finish the build off tomorrow night and fix the flow meter to the keezer a bit better.





Need to finish the soldering of the flow meters and sockets on the project box and then attach them to the keezer and the beer lines.

 

[malt junkie]

Buggered if I can find the flow meter on JayCars website. Might just have to ebay it.

 

[user 35653]

Buggered if I can find the flow meter on JayCars website. Might just have to ebay it.

I couldn't find it either. I did research these a while ago and found the accuracy wasn't too flash. Excited to hear how how these actually perform!

 

[Moad]

Buggered if I can find the flow meter on JayCars website. Might just have to ebay it.

Jaycar don't sell them any more, I got mine from AliExpress. They sell on eBay as well if you just want one or two.

http://m.ebay.com.au/itm/1-30L-min-1-75MPa-1-2-quot-Hall-Effect-Flowmeter-Control-Water-Flow-Sensor-Arduino-/201693323507?txnId=1509142147010

 

[Moad]

I did some quick calibration and got it to 450ml for a schooey, I'll play tonight and report back on consistency once I dial it in.

It was 2.30 and I sank 3 schooners in half an hour calibrating hahaha

 

[malt junkie]

I did some quick calibration and got it to 450ml for a schooey, I'll play tonight and report back on consistency once I dial it in.

It was 2.30 and I sank 3 schooners in half an hour calibrating hahaha

Dialing these things in has ruined many a liver, good luck!

 

[klangers]

I'd be interested to hear about the longevity of these meters, and expecially how well they endure beer line cleaning.

The flowmeters appear to be a turbine type, ie, positive-displacement. As the turbine turns, a hall effect sensor is tripped for every (x) volume units. This should help with the low-flow accuracy. Unfortunately it'll also add a bigger pressure drop, and thus I'm interested to hear if foaming becomes an issue. I doubt it, as you have placed them near the kegs and they have a 1/2" bore.

 

[Moad]

I had zero foaming issues from tests last night, I'm planning to finish it all off tonight.

For the price, I'll happily replace them once a year

 

[Moad]

Accuracy seems a bit off, will do some more playing tonight. Last night I finished the soldering and wired up resistors. All tested I just need to mount the flow meters in line and attach them to the freezer wall, following that I'll play some more with the calibration. Then I need to mount a screen near the taps and I am done! Until I want to add solenoids....

 

[mr_wibble]

Accuracy seems a bit off, will do some more playing tonight.

Off by how much?
Various sources cite either 3% or 10% accuracy.

How are you watching for the "ticks" (pulses) from the sensor?
Could you be missing some of them?
Do you use rPi interrupts for detection? (As opposed to simply polling the state of the pin).
Have you tried using an rPi asynchronous (rising/falling edge) type interrupt?

I've only really dabbled a bit with my rPi, so I can't speak from experience here. But on arduino with a similar kind of reading (a rotary encoder) without using hardware interrupts it would easily miss signals - even something simple like turning the encoder knob anything but dead-slow would cause it to mess up.

 

[Moad]

The arduino handles the interrupts and passes the information to the Rpi. Thats about the limit of my understanding.

edit: with regards to the interrupts, there is a version which runs on the Rpi that uses the Rpi for the interrupts but I didn't look into it too much. I'm using the stock arduino sketch for kegbot. It seems more likely to be over by 3% rather than under so it isn't mising ticks. I'll add more info on this as I use it a bit more.

I am seeing about 3% variation (testing with a 2L erlenmeyer) which is acceptable for this application IMO. it means over the keg I would be out by a schooner max. My starting volume is probably going to be out by more than that so I am happy with that. I'll see over time if it varies more or less and report back.

The build is now functional, connected everything up and am happy to report it is all working as expected.

Next up is mounting a tv on the deck for taps and media streaming/tv, I am trying to code up a custom interface to display all 6 of my taps on one screen rather than the 2 by default.

 

[Moad]

A little more info after playing with these for a bit.

The frequency of the pulses to measure the volume of liquid is linear based on the flow rate. I think what this means is that the flow rate needs to be consistent for the volume measurements to be accurate.

I realised this when I ran out of gas and my pour was slow, the volume was way off.

Real world result is that the readings during a party where lots of people are pouring may be off if you have a check valve or some other reason for the flow rates to change (like flow control).

I am going to remove the check valve I have to help with this, for normal usage though the yf-s201 sensors are pretty good. Even if its within 5-10% I am happy

 

[Moad]

Happy to report I now have the meters accurate to 1-2%

The key is consistent pressure and no air/co2 in the lines

 

[Moad]

I rebuilt the server software onto a linux machine as the raspberry pi couldn't handle all the python.

Finally resolved the last issue I was having last night, next is to mount a TV and tablet on the deck.

http://smartykegs.thickasthievesbrewingco.com

You can check out my night of calibration... 4.5L wwweeeeoooowwwww

 

[dblunn]

Hi Moad,
that looks great but how do you stop SWMBO from seeing that web page?

 

[mr_wibble]

Hi Moad,
that looks great but how do you stop SWMBO from seeing that web page?

What about when SWMBO has a higher score than you :blink: