Hardware Sorting Test
May 19, 2025
Getting ready for the stress test
The calibration tool we built earlier isn’t suited for this task — which is expected.
Now it’s time for a new tool: a simple test app focused purely on mechanical stress-testing.
No need for fancy visuals — speed of development and minimal comfort for the developer is what matters.
At this stage, we’re testing only the mechanics. The recognition module won’t be used — instead, we’ll insert a short 0.5-second pause to simulate processing time.
Naming the slots
Our card sorter has seven slots in total, and it would be helpful to refer to them by name. We could use numbers — but the last slot is reserved for unrecognized cards, and calling it “Slot 7” doesn’t feel quite right.
So let’s use letters instead: [A–F] for recognized cards sorted by specific traits, and [X] for unknown or unrecognized ones.
Basic motion testing
Before we can run a full-scale stress test, we first need to define the basic motions.
For the dispenser, that means:
– feeding a single card.
For each sorting platform:
– swiping the card left or right,
– dropping the card onto the next platform,
– and using the sensor to determine the shortest timing that still ensures reliable movement.
First "Magic: The Gathering" card sorting test
We’ve added Start/Stop buttons to control continuous card feeding.
To simulate recognition, the system uses a fixed 0.5-second pause and returns slot IDs in a loop from [A–F, X].
This makes it possible to fully test the mechanical sorting process — and finally see the big picture: how the system performs under load, how it looks in action, and how stable it is.
The first tests weren’t perfect. Sometimes the dispenser would feed two cards at once — and tightening the separator would occasionally stop it from feeding any at all.
After some mechanical tweaks, we reached stable performance with MTG cards, averaging around 28 cards per minute — well within our expected range.
Testing "Pokémon" card sorting
When testing Pokémon cards, the results were almost identical to those with MTG — the dispenser fed the cards reliably, and they moved across the platforms without issues.
That said, Pokémon cards initially showed some occasional sticking. This might be caused by static buildup on the surface or possibly stronger surface adhesion found in brand-new cards — but that’s just a theory for now.
We’ll see if the effect repeats consistently before jumping to conclusions — and if needed, we’ll find a fix.
Testing "Lorcana" and "Flesh and Blood" card sorting
Flesh and Blood and Lorcana cards were also put through sorting tests — and the results were very much in line with MTG.
The dispenser fed the cards reliably, and the sorting platforms handled them without any issues.
No noticeable problems occurred during these sessions, but we’ll continue to monitor future tests to ensure consistent performance across different card types.
Testing "One Piece" and "Yu-Gi-Oh!" card sorting
Unlike previous cards, One Piece cards are thicker and stiffer, while Yu-Gi-Oh! cards are smaller and have a glossy finish.
In earlier versions of the dispenser, these cards simply wouldn’t feed reliably — at least not one at a time.
But after recent mechanical tweaks, we gave it another try — and it turns out there is hope, as long as two conditions are met:
– The cards must be perfectly flat (manually flattened if needed),
– Additional weight must be applied to the pressure roller.
We’re not big fans of placing extra objects (like weights) on top of the card stack — it’s awkward and feels like a workaround. Ideally, we’d love to keep the top of the stack completely clear.
Either way, to avoid stalling progress, we’ll stick with this setup for now — and possibly explore better solutions later on.
Results and Takeaways
During testing, we managed to reach 33 cards per minute — so a steady 30 cards per minute looks perfectly achievable.
The card dispenser requires careful tuning of the separator gap, but once adjusted, it performs quite reliably. Later on, we’ll definitely want to explore mechanical solutions that make this adjustment easier — and improve long-term consistency.
All in all, the design has proven to be viable — and honestly, it feels good to see it working.
What’s Next
With mechanical testing complete, it’s time to move on.
The next step is designing the actual user interface — the one that will run on the device and be used day-to-day. Once the design is in place, we can start developing the app itself.
In parallel, we’ll begin collecting card data for each game and building our own database.
This also means defining meaningful traits and classification rules — things like rarity, color, mana cost, or game mechanics — so we can create sorting presets later on.
As for the device app, one idea is to allow full remote access — so you can view the screen and control everything from your phone or PC.
But more on that as it takes shape.
In the meantime, don’t forget to subscribe to the blog so you won’t miss what’s coming next. ;)
To be continued...
Want to support the project?
See how you can help →