How accurate are smart scales?

Are smart scales accurate weighing instruments or just simple bathroom scales with expensive connectivity features? To find out, I tested the Withings WS-30, Fitbit Aria and A&D UC-324NFC using class M1 weights.

A&D UC-324NFC Fitbit Aria Withings WS-30
Creative Commons LicenseThis image by digitalscalesblog.com is licensed under a Creative Commons Attribution 4.0 International License

Preparation

Each of the scales had not been used before and was configured as instructed by the manufacturer. This included signing up for an account with Withings and Fitbit. In the case of A&D’s UC-324NFC, I took advantage of a function available in the Wellness Connected  app which let me “calibrate” the scale by entering the local gravitational acceleration.

The scales were tested under similar conditions with a stable temperature of around 22°C and were placed on a flat, hard surface. The Withings and Fitbit scales were left alone for a night after the initial configuration as I wasn’t quite sure how they determine the zero value: unlike older electronic scales, you don’t have to switch them on and wait for them to show ‘zero’, instead, you step right on (the UC-324NFC determines the zero point after you step off).

Test procedure

I used 20kg weights to test the scales at 40, 60, 80, 100 and 120 kg. This procedure was repeated 3 times for each scale (I did a lot of weight lifting that day).

test
Creative Commons LicenseThis image by digitalscalesblog.com is licensed under a Creative Commons Attribution 4.0 International License

Results

The first column gives you the nominal weight (in kg) and the next three are the values as shown by the scales.

Withings WS-30

4040.240.240.2
6060.360.360.3
8080.480.380.3
100100.3100.4100.4
120120.4120.4120.4

Fitbit Aria

4040.240.240.1
6060.260.260.2
8080.280.280.2
100100.3100.3100.3
120120.3120.3120.4

A&D UC-324NFC

4039.9539.9539.95
6060.0059.9560.00
8080.0080.0079.95
100100.00100.00100.00
120120.00120.00120.00

As you can see, A&D’s UC-324NFC was the most accurate scale, never deviating by more than 50g from the true value. Its 50g resolution and gravity compensation function certainly paid off.

However, the Fibit and Withings scales did a great job, too. Looking at the measurement results in each row, you’ll see that they are either identical or off by just one scale interval (100g). We can therefore say that these scales were slightly less accurate, but still very precise* (or that their repeatability was still excellent). As far as your body weight is concerned, you’ll generally be less interested in absolute values and more in tracking changes (“am I loosing or gaining weight”) and these scales would be perfectly suitable for this purpose. Nevertheless, I was a bit surprised that Fitbit and Withings didn’t use the location I had entered during account creation to adjust the scale for the local gravity acceleration. This would most likely have improved their accuracy.

Overall, the results are very good. As far as I can tell from this test, you’re not just paying for the “smart” features, you get accurate scales, too.

Further information / where to buy:

Disclosure:

  • Some links above are affiliate links.
  • My company used to sell A&D’s UC-324NFC in our German scales shop, but we don’t anymore. The Withings WS-30 and Fitbit Aria were purchased at retail prices, A&D’s UC-324NFC was purchased with a dealer discount.

*Meaning of accuracy and precision (as used in this article):
Accuracy and precision explained


[related_posts_by_tax]

RS-232 troubleshooting: fake chips

RS-232 is still the most popular interface for balances and scales. It is often described as “simple”, however, when things don’t work as expected, finding the cause can be difficult.

This article does not aim to be a comprehensive RS-232 troubleshooting guide (for this purpose, please refer to this PDF document from Agilent Technologies or the troubleshooting section on our 232key website). Instead, it is supposed to raise awareness of an issue that is often ignored: counterfeit ICs.

When you’ve tried everything and still can’t reliably communicate with your scale via RS-232, there’s a chance a fake chip may be the cause.

Just a few weeks ago, the thought of encountering counterfeit ICs in digital scales had not crossed my mind. Thanks to FTDI’s recent attempt to “brick” counterfeits via Windows Update, fake chips are now a hot topic on the web. While I don’t agree with the way FTDI tried to punish the end user, I wish the controversy had occurred a few weeks earlier. This would have saved me a lot of time.

Back then I was doing the final QC for several scales which were about to be shipped to a customer. The last item on my checklist was “bidirectional communication using RS-232”, something I had done many times before with this exact model. What should have taken a few minutes ended up taking me several days and nearly drove me crazy because the problems I encountered were difficult to replicate. Eventually, I arrived at the conclusion that something was very wrong with the MAX232CPE+ chips, which are responsible for converting TTL signals to RS-232 levels. After doing some research on the internet*, I started to suspect those chips were counterfeit. It seemed like a far-fetched idea** at the time, but I still desoldered them, took a few pictures and sent them to Maxim Integrated.

Fake MAX232CPE+ and MAX232EPE+

Counterfeit MAX232CPE+
Fake, fake and fake.

Thankfully, I received a reply in less than two hours:

“Yes these parts are counterfeit, they do not match markings of lots we manufactured.”

Now extremely suspicious of all MAX232s***, I disassembled a few more scales from 4 different vendors. 3 contained ICs belonging to the MAX232 family, so I sent the pictures to Maxim Integrated, too. In addition to the chips used by the scale manufacturer which had prompted me to start this investigation, one chip used by another manufacturer was also flagged as counterfeit.

Counterfeit MAX232EPE+
Counterfeit MAX232EPE+

To be fair, my sample size is too small to draw meaningful conclusions regarding the entire weighing industry. However, if you’re in the business of making weighing instruments and were blissfully unaware of this issue, I hope this article serves as a wake-up call.


* I found documents like this one (16 MB PDF presentation by SMT Corporation) or this one (100 KB PDF, University of Conneticut), this article by Maxim Integrated and even videos of YouTubers walking through huge electronic component malls in Shenzhen where almost everything is counterfeit.

** Though not quite as far-fetched as the manufacturer’s idea that “static build up from the polystyrene packaging in road transportation” was to blame.

*** And also seriously angry at having wasted so much time doing something the manufacturer should have done. I won’t do any naming and shaming here, though.


Update October 3, 2017: Several articles concerning this issue have appeared since I originally wrote this blog post.

Update October 15, 2020:

Update February 13, 2022:

Solution: Err 8.2 on Ohaus Defender scales

If your new Ohaus Defender shows “Err 8.2”, you have probably not yet removed the shipping spacers. These protect your scale from damage during transport, but have to be taken out before use.

1. Remove the stainless stainless steel platform pan. You’ll see 4 red plastic shipping spacers wedged between the overload protection screws and the upper frame of the platform:
Ohaus Defender 3000 shipping spacer

2. Remove these spacers by hand or with an adequate tool (without turning the overload protection screws):
Ohaus Defender shipping spacers removal

3. Place the stainless steel pan on the platform. Your scale is ready to weigh.