Scales and design - Page 6

Want to talk espresso but not sure which forum? If so, this is the right one.
wojtowip
Posts: 64
Joined: 3 years ago

#51: Post by wojtowip »

OK, finally made some progress. It's not pretty, but at least it's a proof of concept and I can order some smaller electronics and make the code a bit more stable :).

I'm using the cheap 5kg load cell I had, created a lower profile platform for it. It's not as stable as I would like. Ideally I would mount it directly to the CT2, but I'm not exactly ready to drill holes in it yet. The whole thing is about 18.7mm tall and I had to slide it in in two pieces. The final load cells will be much smaller. I would like to get a more ridged base and platform, but don't have a way to mill stainless steel, so will need to find maybe one thicker sheet of acrylic instead of two glued together. We'll see.

So here it is inside:


Here is an overview with the display:


And here it is with the drip tray in place:


Testing scale with water (sorry about the vertical video, it's late :))
Testing scale with 100g weight:
LMWDP #694

DamianWarS
Posts: 1379
Joined: 4 years ago

#52: Post by DamianWarS »

Espressoman007 wrote:Hey,

Perhaps there is an answer or discussion on the forum about it, if there is, send me a link. If not, perhaps someone has the answer - why are all coffee/jewelry scales made to be used with AA or AAA batteries, and not with button batteries. I don't know if every kitchen scale is thinner than a coffee scale because of that, but most familiar to me are.

Is there a reason for that or... what?

There are not many espresso machines, levers, that have a gap that small between the base and group/portafilter, but there are some that need the thinnest scale there is. Strietman for example. And what's more interesting is that kitchen scales (unlike coffee scales) are manufactured to measure heavier things.

Thanks!
Cheers!
there probably is not much demand for such tiny scales and as mentioned if you need and want to pay the premium Acaia has a 13mm scale. I have the Lunar and it's a 15.5mm scale and to some a 2.5mm may make all the difference but I would think the Lunar would be a better option as it is a much "smarter" scale and has more options to weigh/time and designed for espresso. For a cheap option there's this generic scale that can be super cheap. I bought 100 of them for less than $5 each and gave them out at an event. I've had it for years and it still works great. I don't know the exact height and there's conflicting information, the amazon link says 0.6 inch and 3/4 an inch. Various ones have different features like accuracy or that one had usb charging. But if is 0.6 inches then it's the same height as the lunar otherwise its 19mm high. if it's still too snug you can try using little shot glasses over cermic cups to keep it as minimal as possible.

Other solutions can be 3D printing a drip tray with a place for your scale to sit in (or buy a custom made one) or if the machine has high enough legs use the scale under the machine and have some sort of "C" shape stand that allows the cup to sit on top while putting pressure on the scale under the machine (another 3D printed option) this option is best if the stand can firming be attached to the scale so it could mean gluing it on. Here's an example of one

BeesKnees
Posts: 33
Joined: 2 years ago

#53: Post by BeesKnees »

I have been researching parts for a built-in scale.

Been looking at this:

https://www.mouser.com/ProductDetail/824-FX29J0-1A-01L

I think this will be easier to mount under a tray. The part costs a bit more but should update faster than other solutions and is already calibrated.

User avatar
Jeff
Team HB
Posts: 6811
Joined: 19 years ago

#54: Post by Jeff »

Nonlinearity and repeatability are 1% and 0.8% of span, respectively. It looks like the "lightest" one is 10 lb, or just under 5 kg. 1% of 5 kg is 50 g.

It might be better over a fraction of its range for nonlinearity. I don't know about repeatability.

Calibration really isn't a big problem, unless you expect to swap out sensors without recalibration. With a reasonably linear transducer, zero it and measure, put a 500 g cal weight on it and measure, and you've got your offset and gain.

For 0.1 g repeatability on a 500 g range, I'd be looking for an 0.1% device.

BeesKnees
Posts: 33
Joined: 2 years ago

#55: Post by BeesKnees replying to Jeff »

Thanks for digging into the data. I kinda liked the idea of using the part to measure the tamp pressure as well as the dose but if is too inaccurate to measure a few grams I'll keep looking. Having I2c out of the part seemed really cool too.

BeesKnees
Posts: 33
Joined: 2 years ago

#56: Post by BeesKnees »

my little scale project continues... step 1 buy parts and monkey around. Thankfully there are libraries for the parts.
Found a video for the "decent" machine + "decent" scale. So with a 0.1g accurate scale they make a shot that is 2.7% (0.9g) off from the set value and they said it was "very good". It looks like the decent scale is updating 10 times a second. Maybe some of that is at play? It is neat to see where the highbar is if I attempt to automate a shot.

https://github.com/sparkfun/SparkFun_Qw ... no_Library

I am going to try out a nau7802 with a 500g load cell. This should update faster than the HX711. Does that matter? I dunno. This scale would be used for beans.

It seems like the bar load cells are not meant to have weight on them all the time since they will warp. So now I am back to looking at the TI FX29K to try since it is meant for this job at the cost of accuracy. If I use an accurate scale to measure the beans that should let me fudge a bit for measuring the water out of the machine. I really hate the look of a scale on top of a shiny espresso machine so this suits me.

https://github.com/eepj/FX29K_for_Arduino

@wojtowip have you had any issues with drifting measurements or needing to re-calibrate over time?

User avatar
Jeff
Team HB
Posts: 6811
Joined: 19 years ago

#57: Post by Jeff »

That might be an old video. I rewrote the stop-at-weight algorithms about a year ago and with a Skale II, the accuracy is typically within a 0.15 sec * flow rate. At typical rates, that's usually within a tenth or two of target. The Skale II, for some reason, updates 10 per second but transmits on a 150 ms clock, doubling up reports.

BeesKnees
Posts: 33
Joined: 2 years ago

#58: Post by BeesKnees replying to Jeff »

Hey, Jeff! Looks like July of last year? So just 6mo.

Ok, so you are saying within 0.1 to 0.2g of the target is possible. That is darn tight! Congrats! So 10 readings per second is enough to get there. I figured with the need to average a bunch of readings more measurements per second would be better.

User avatar
mkane
Supporter ♡
Posts: 1750
Joined: 6 years ago

#59: Post by mkane »

...........

BeesKnees
Posts: 33
Joined: 2 years ago

#60: Post by BeesKnees »


My little scale journey continues....

I ended up with the display above and an Adafruit Feather nRF52840 Express for the smarts. So battery charging, bluetooth, and color leds are in the mix. It will be cool if I can emulate a consumer BT scale and use some coffee apps. It looks like it would be easiest to copy the "decent" scale's settings from the api (thanks decent!).

The part of the project I am excited about is having a display that can update hundreds of times a second. I am curious about flipping between graphs and displays at a fast rate. The sensors are not as fast as the leds so this could work out. Or, maybe not, so I got 2 displays just in case.

side note: I saw a scale app that let you set a measuring cup weight in grams and that would tell a scale to auto-tare. I am 100% going to poach that idea. I use the bottom of my manual grinder to weigh beans. With a set weight auto-tare, I could just set it to slightly below the measuring cup weight and have one less button press. If anyone has any automatic scale features they like please share the process so I can steal that too.

more later! My fingers are crossed that this is a friction-free hobby project.