What is this world coming to when you spend seven bucks on a digital scale and you have to completely rebuild it to get the functionality you need? Is nothing sacred anymore?
Such were the straits [Jana Marie] found herself in with his AliExpress special, a portable digital scale that certainly looks like it’s capable of its basic task. Sadly, though, [Jana] was looking for a few more digits of resolution and a lot more in the way of hackability. And so literally almost every original component was ripped out of the scale, replaced by a custom PCB carrying an STM32 microcontroller and OLED display. The PCB has a complicated shape that allows the original lid to attach to it, as well as the stainless steel pan and load cell. [Jana] developed new firmware that fixes some annoying traits, for example powering down after 30 seconds, and adds new functionality, such as piece-counting by weight. The video below shows some of the new features in action.
Alas, [Jana] reports that even the original load cell must go, as it lacks the accuracy her application requires. So she’ll essentially end up building the scale from scratch, which we respect, of course. At this rate, she might even try to build her own load cell from SMD resistors too.
[via Twitter]
Why didn’t he just build it from scratch in the first place?
Because the lid would be missing.
7$ is probably less than a scratch built case involving metal.
Spend a little money save a lot of time…. Get burnt by the product, have a jumping off point in the case, cut your losses and save a little time.
Where’s the fun in all that. It’s a gamble, at $7. Could have worked as expected. Even if you spend more, on a better grade, doesn’t mean you’ll get all of the features you really want, or that the work the way you need. Hacking, or building your own, means you get pretty much exactly what you wanted, instead of sending failed units to the parts bin, or landfill. But mostly, you learn a few things along the journey. It’s an adventure. Bonus, the build gets shared online, so others can save the time and money, if they want. Least let people know the $7 scale is crap… Also, the build files, are a huge start, for someone wanting to build from scratch. Time well spent, and appreciated.
Fully agree … easier to:
– edit a speech than write one from scratch
– remix a song than write one from scratch
– rebuild an engine than build one from scratch
etc.
Will have to bookmark for a future project. Got a Pelouze PE5 that’s acting dead, and piece count upgrade would be sweet too. Not sure if the load cell has a good resolution or not though.
A load cell is an analog device.
What’s the equivalent then?
My analog scales definitely lack something because for a relatively wide range of 10kg it’ll read the same weight.
Like a local minima in the spring.
A worn down spot in the mechanism? If you tap it around that spot and it doesn’t return, it sounds more like static friction.
Yes, for some reason I was thinking they came in assemblies with an ADC built in.
Was peeking into it and looking stuff up. Appears to be regular Wheatstone bridged strain gauge type deallie with only 10 bit ADC.
So given that wiring it direct into a NodeMCU or nano clone analog input won’t get me any better, I am thinking I might try to dig a 16 bit DAC out of an old CD drive or player, and then make a ramping voltage comparator type ADC with opamps. I don’t think I need it to be super speedy after all.
Don’t waste your time. The HX711 is a 24bit-ADC for load cells and sells for less than 1€ on Ali.
Poor guy had his entire weekend planned out and you ruined it!
Better than mine since it comes with a calibration weight.
Raid your piggy banks, sort out unworn looking coins, check weights on wikipedia, put piles on scale that you’ve calculated weights for, see how close it is. Also a graduated measuring cylinder can be tared and be filled with specific amounts of water, deionised if it’s super hard from the tap.
It’s been my experience that even with a volumetric flask and water at the right temperature, you can barely provide a close enough mass to match even a cheap scale’s precision.
My cheap scale measures to 10 milligrams. With a 250mL volumetric, and some skill in filling it, my expected precision is exactly 10 milligrams. The good scales are 10-100x as precise.
I’d expect that a graduated measuring cylinder may be accurate to 200 milligrams.
Well yes you’re limited to the precision of your measuring cylinder, I was thinking of a 50 or 100ml one that read in ml, that you’d read parallel with bottom of the meniscus in approved school chem lab fashion. Maybe needing a dropper to get it dead on the mark.
This is sweet because it provides the hardware for an interface. You can move part counting software onto your computer, and add logging and automatic taring after each measurement. Plus you can measure troy, carats, grains, whatever weird unit you feel like you need. One of my physics teachers established, for pedagogic purposes, a prepackaged “little john’s burrito” measurement system, where its length, weight, density, and caloric content were the units we used to do things like calculate how much it heated up every time it was dropped from the top of a ladder and how long it would take to bake it by doing that. (Forever: it lost more heat dropping than it picked up in impact.)
Yeah the feature which interests me most, easily possible on modern mcu, is the mean and SD of some weights. Calculated by adding piece wise.
That will make counting feature much better. You could weight many smaller things with producing a confidence interval.
One sided it’s ideal for a small business. Like 99% sure you provided at least 100 LEDs. Can scale to two sided (to prevent being her generous) easily as you scale up.
Individual items maybe too small and variable for the scale to accurately weight in count
Aww..i got excited and thought that 50g weight was a bullet casing.
I don’t know how he will get that extra precision without protecting the scale from air currents. Even the moisture from your hands evaporating will cause drops at the 3rd and 4th decimal place.
The old MyWeigh i2600 I bought new around 20 years ago measures to 0.1 gram and it’ll react to static forces between the plastic cups I use for mixing resin and the resin cans. Zero it, get the can close enough and with the right weather conditions it’ll pop up to -0.1. I can forcefully push a flat hand down at it and make it register 0.1 gram briefly from the air flow.
The auto power off “feature” alone is cause enough to do this. Damn, that is annoying. I tried baking with a scale like that once and it would constantly shut down while I was using it to measure the ingredients.
I had one like that too… really annoying, especially because the timeout was really short, like 15-20 seconds or so. I got rid of it, and bought a slightly more expensive German Soehnle brand scale. That one just stays on until you press the button again.
I also hate the ones where the load cell is connected to the feet instead of the platform.
THIS RIGHT HERE! With so many cheapo modules and products based on them, where are the open source ones? People could feature creep them out in strange directions.
There’s a reason some folks still use balance beam scales and check the calibration often. Even the electronic scales are supposed to be allowed to stabilize and then checked. All we want is .1 grain accuracy. All the time.
So one of my buddys told me of a trick he said worked for him, he got a cheap digi measure from the corner store and when doing the calibration he added a ten gram weight instead of the one gram tare weight it came with to calibrate and then it gave him .01 instead of .1 resolution, just having to move the decimal place over a digit. I tried it with a cheap one that i had but it just gave me an eror message so its a roll of the dice if itll work for your scale. He mentioned that the cheap .1 and the more expensive .01 measurement models looked exactly the same so maybe its a matter of how your specific scale was designed.
I love the themed PCB art! 😍