The FS8205A on Your BOM Probably Isn't the One on the Datasheet
Fortune's FS8205A is TSSOP-8 only. Every SOT-23-6 part sold under that name is a second source with a different pinout.
If you are putting a 1.14" ST7789 panel on a board, check the contact orientation of your FPC connector before you route anything. I did not, and it cost me a board revision three weeks before a 100-unit run.
The useful part of this, the part I could not find written down anywhere, is at the bottom. If you are in a hurry: there is no 8-position top-contact FH12.
The setup
The pen uses an EastRising ER-TFT1.14-2: 1.14 inch IPS, ST7789V, 135 x 240, on an 8-pin 0.5 mm FPC. The board takes it through a Hirose FH12-8S-0.5SH.
Both parts are extremely common. Both are correct parts. They do not work together.
Contact orientation is not a detail
An FPC has copper fingers on exactly one face of the tail. A ZIF connector has its contacts on exactly one side of the slot. If those two faces do not meet, you get nothing, not a wrong signal, not a weak signal. No electrical contact at all.
The panel datasheet, section 2.1, says: "8 Pin 0.5mm Pitch Top Contact ZIF Connector."
Hirose's own 2D drawing for the FH12-8S-0.5SH says: "Do not insert FPC/FFC with the conductor surface face up." That is a bottom-contact part.
Top-contact panel. Bottom-contact connector. In every insertion orientation, that combination is an open circuit.
The obvious fix does not exist
The obvious answer is to swap to the top-contact version. Hirose makes one, the FH12A series. FH12 is bottom contact, FH12A is top contact, FH12S is bottom contact with a strengthened actuator. Same footprint family, same 0.5 mm pitch.
So I went to order an FH12A-8S-0.5SH. It does not exist.
Here is every 0.5 mm FH12A position count in Hirose's own series catalog:
10, 12, 15, 16, 18, 20, 22, 24, 26, 28, 29, 30, 32, 33, 34, 36, 40, 42, 45, 50.
The bottom-contact FH12 goes down to 8 positions. The top-contact FH12A starts at 10. If your design needs 8 pins and top contact, that specific part is not available in this family, and no amount of searching distributor stock will change it, it was never made.
I could not find this stated anywhere. Distributor filters happily show you FH12A parts and let you assume the position range matches FH12. It does not.
What the panel was telling me the whole time
Reading the datasheet properly instead of grepping it for a pinout, two things stood out.
The mechanical section gives the outline as "Outline Dimension with FPC Folded: 17.60 x 31.00 x 1.60 mm." The manufacturer specifies the panel's size in its folded state.
The drawing marks a single-side bend region on the tail, and carries a note (translated from the Chinese) warning that the bend protrudes past the backlight bezel and to leave clearance for it in the enclosure.
Folding is not a workaround for this panel. It is the intended installation. And a 180-degree fold flips which face the contacts present, which makes a bottom-contact connector the correct choice after all.
The catch
A fold flips contact face and reverses pin order together. You cannot have one without the other. Folded, panel pin N lands on board pad 9-N.
Panel pinout: 1 = LEDA, 2 = GND, 3 = RESET, 4 = RS, 5 = SDA, 6 = SCL, 7 = VDD, 8 = CS.
Reversed onto the board, pad 1 has to carry CS and pad 7 has to carry GND. My board had pad 1 on GND and pad 7 on CS, which meant, built as drawn, pad 2 would have driven 3.3 V straight into the panel's ground pin.
Half the fix was not a board change
What surprised me: of the four pads that needed to move, only two were actually copper.
Pads 5 and 6 needed DC and RESET swapped. Both of those land on ESP32 GPIOs, so the swap is two lines of firmware, change which pin the driver calls DC and which it calls RST. Nothing on the PCB moves.
Only pads 1 and 7 involved real copper, because GND is not a GPIO and cannot be reassigned in software.
Final DRC came back identical to the pre-change baseline: 32 warnings, zero errors, zero unconnected. BOM and placement file byte-identical, so the assembly house sees no change at all.
What I would tell past me
- Contact orientation is a spec, not a detail. Check it the same moment you check pitch and pin count. It is a binary works/does-not-work property and it is invisible in a footprint.
- A footprint and a part number are two independent assumptions. The library entry is named after a part. That is not the same as being that part.
- Read the mechanical section, not just the pinout. The words "with FPC folded" were in this datasheet the entire time and would have answered everything.
- Check that the part you are about to spec exists in the position count you need. Series-level marketing copy says "6 to 60 positions." The actual per-variant table says otherwise.
Nothing had been fabricated when this was caught. That was luck as much as process, the manufacturer happened to be waiting on a separate question.
Logan Holby is the founder of Nexus Pen, building an AI-powered pen with a color display. More at the founder page.