This project is the LED clock based on the STM8S003 microcontroller. This clock uses firmware 24-bit digital controller oscillator to achieve fine-tuning tolerances. After calibration, no external synchronization is required.
Features:
-6-bit 7-segment LED display
- RTC backup with supercap.
- Time, Date, Summertime
- Low power: 5V 20mA
I chose to use ordinary 12MHz crystals as the frequency source, not the regular RTC modules used by other items. Be
If the clock is used in a temperature-controlled home environment, the temperature variation may be small enough to remain reliable without using TXCO. For example, set the heater to 21c, set the air conditioner to 25c, means that the change is about 4C.
The 16-bit timer in STM8 generates 20 Hz IRQ. Then, more accurate 10 Hz can be generated by using the 24-bit number-controlled oscillator (NCO) implemented in the firmware. Be
For example, if your 20Hz is +50 ppm fast = 20.001Hz, NCO can be used to divide it at 2.0001 to get 10 Hz.
NCO can perform decimal division by tracking the phase in the accumulator, the following three lines of extra code is very fine crossover! Be
The RTC code uses 10 Hz to update the time, date and achieve the summer day (DST) of North America.
The RTC code is implemented in the firmware, so the odd clock of different time units can be suitably used. For example, metric time or Mars time.
accuracy
This test was carried out from November 25th to December 25, and less than +/- 1 second after 1 month. This includes a tolerance, long-term drift due to temperature, power, and the like. 4 weeks 6 days later, the clock speed was reduced by 1 second. It can be seen that this is as accurate as some of the better levels of RTC after calibration.
In contrast, the ready-made RTC module is mixed, and some common crystals with external crystals are not accurate.
No need for any calibrated RTC module accuracy:
> However, in our experimental project (using the RTC with Arduino is used for data recorder, we found that these DS1307 modules have differences in timing - increase / decrease a few seconds, while others Add / decrease in a few seconds per day / 3-5 minutes per day. Be
> The DS3231 chip on this module is 2 ppm (one million), which means less than a second for every 5 to 6 days. So far, we have tested the units of less than 1 ppm, so the most lost or increased a few seconds per month.
When using GPS to calibrate high quality RTC modules:
. 1882. This (DS3231) RTC is in this week's performance is -0.038ppm +/- 0.085 ppm, fasting about 1 second annually
> This week's PCF2129 performance is 0.462 ppm +/- 0.239 ppm, which is about 1 second per month.
These two parameters are currently satisfied.
calibration
This clock can be calibrated without external frequency counters, GPS, or the like. I just use the binary search algorithm to reduce the value of the clock's run / slow to the atomic clock website. Each of these tests can be noted for a long time. As the test progresses, as the clock becomes more accurate, the running time of each test will be longer. One of the advantages is that long-term effects / drift is part of the test, so this is the accuracy in real life.
So far, I have reduced this comparison by 7-8 times, and my accuracy is less than +/- 1 second / month. What surprised me is that very inaccurate crystals can be so good. Of course, we can also make more adjustments, but I think other factors (such as temperature drift, etc.) will become restrictions, unless I use a more stable clock source (eg, TxCO or OCXO) instead of crystals.
User Interface
Clock and display time:
The arrow key is switched between time and date display. For example, January 16 (Thursday) is as follows. Be
7 Duan LED can only display a set of limited characters.
You can use 3 buttons to set the time (and similar dates).
Hardware description:
Initially, I plan to use VFD, but 4 digits of user interfaces have too much restrictions. (I have to limit the UI on the STM8 timer with 16x2 LCD)
I use STM8S003, 74HC595 and several cheap green 7-segment display. In fact, the display efficiency is not bad because the entire clock runs at ~ 20mA at 5V voltage.
RTC is implemented in the firmware in the firmware to perform fine frequency adjustment, and the supercapacitor allows STM8 for a few minutes during the power failure, and the LED driver is located on a separate rail.
The PCB is connected to the original PCB to use the existing button. In addition, GPIO for public cathode drivers is used for polling buttons.
I can use STM8 alone, but 74HC595 split some GPIO lines, UART, I2C, timer pins, power and debug interfaces to connect to connector connectors on the I / O panel PCB using existing RJ11 slits.
(Since then, I change the power connector to MicroSB because it has a better latch function.)
External I / O
There is an I / O connection using existing apertures (a pair of RJ-11 for caller ID). They are 5V power inputs for hardware debuggers SWIM, 3.3V serial port: TX, RX + power, 1 GPIO, I2C.
Change the old traveler's Cassio clock because it is difficult to read at night.
Be
Be
Be
Article Source HackADAY
Our other product:
