E-Ink Clock: Hardware
This page gives an overview of the hardware used. You can find the KiCAD files in the GitHub repository . The KiCAD schematic is filled with part numbers and can be exported using KiCAD's BOM functionality.
Attention: The PCB layout in the repository is out of date, since I had some issues in the first PCB. But it should be easy to change the PCB layout in such a way that it fits the schematic again.
The schematic is divided in four parts: the voltage generation circuits, the E-Ink display, the microcontroller and the DCF77 module.
The circuit needs six different voltages, of which five are generated here. The sixth voltage is the contrast voltage of the E-Ink display, which is generated in E-Ink part of the circuit.
- +3.3 V: This voltage is generated using a low-dropout linear regulator, namely a TPS71533 (U5; upper left). There is not much else to say about this, because this is quite standard. This voltage is used as supply voltage for all chips.
- +15 V & +22 V: These voltages are both needed by the E-Ink display and are generated using two independent TPS61041 switching boost converters (U6 & U8; middle row). The circuit was designed using the application note in the datasheet of the chip. Since I have to be able to completely turn off these voltages I added a load switch to turn them off.
-15 V & -20 V: These voltages are again needed by the E-Ink display and are generated using two independent LM2611 Ćuk converters (U7 & U9; bottom row). Building these blocks was, once again, done by strictly following the datasheet. Unfortunately, these two converters draw a high amount of current — I was able to measure about 1.3 A — in the startup moment. If I am going to revise this board, I might want to try to use charge pumps instead of Ćuk converters.
In a first version I tried to control the Ćuk converters using a MOSFET, but this did not work out. The first MOSFET I tried, not considering the high inrush current at this time, was obviously destroyed. The second MOSFET I tried was not able to supply the required startup current fast enough, leading to current limited short circuits. This was a very hard issue to debug, but luckily I found a post on mikrocontroller.net (German) that helped me to come to this conclusion.
All this can be achieved much simpler: In my first prototype I used the mother of all switching regulators, the MC33063A , to generate +22 V and -20 V and two linear regulators to obtain the +15 V, respective -15 V. While this is highly inefficient, it is easy and works well, even on a breadboard.
Not much to say here, the connections are exactly the same as Sprite_tm uses. You can also see that I use his construction to generate the contrast voltage, which happens to be -1.49 V in my case, but can be changed using a trimmer.
To connect the display with the PCB I used a one-sided Hirose FPC-connector with 39 pins and a pitch of 0.3 mm.
The LPC1227 was primarily chosen because of the built in real-time clock that is able to wakeup the microcontroller from its Deep Power-Down mode. There is nothing more to say about the microcontroller itself, because this is the standard wiring.
The SRAM module, a Microchip 23LC512 , has a capacity of 512 kbit and is used as a 1 bpp image buffer (800 px · 600 px · 1 bpp = 480 kbit). Rasterizing to this buffer is much faster than rendering the image on the fly while updating the E-Ink display. This leads to huge power savings, since the E-Ink display update is — by far — the most power intensive phase.
This is a standard wiring for most DCF77 modules including a simple filter.
Just keep in mind that some
DCF77 modules do not work if
ON is directly connected to ground because they
need a high-to-low edge to start up.