h a l f b a k e r yCrust or bust.
add, search, annotate, link, view, overview, recent, by name, random
news, help, about, links, report a problem
browse anonymously,
or get an account
and write.
register,
|
|
|
Please log in.
Before you can vote, you need to register.
Please log in or create an account.
|
Commercially-available real-time clock chips today generally incorporate a wide range of features including battery-backed RAM, automatic month/day/leap-year handling, automatic power switch-over (so as not to draw from the backup battery when the main system has power), etc. While all these features
are well and good, they push the price of RTC chips beyond that of small microcontrollers.
What I would propose would be a five-pin chip (VBatt,VSS,Crystal1,Crystal2,Output) which incorporates the bare minimum functionality: a 32Khz crystal oscillator, 47-bit counter, a multiplexor, and a weak open-collector output. The output would transmit 16 frames/second of 60 bits each. Each bit consist of either a 30.5us "on" time followed by a 213.6us "off" time (for a "1" bit), or a 213.6us "on" time followed by a 30.5us "off" time (for a "0" bit). The output frame would consist of the current time (top 36 bits of the counter) and an optional laser-programmed serial number.
Because the chip would switch its output transistor independent of anything else in the system, it would not need fancy "enable" hardware. Since the power consumption would be unaffected by the rest of system operation, and because the output would be open-collector, it would not need any internal power switching circuitry. While there would be no facility to set the counter on the chip, nor to store any data in an on-chip RAM, both of these problems could be easily mitigated via the use of an inexpensive EEPROM chip.
While existing RTC designs tend to be somewhat pricey, I would expect that a minimal-RTC design like the above could sell for well under $0.25 if not under $0.10. The design and engineering should be fairly minimal, so it should not take too long to recover those costs.
[link]
|
| |
Existing RTCs are cheap enough that I cant think of an application where they need to be cheaper. And you need to add the crystal and the EEPROM. You might as well build it on the same die as a small micro and be done with it (and, I'd be suprised it that wasn't baked!) |
|
| |
Some PIC microcontrollers include a 32KHz crystal oscillator which runs independently of the main CPU and can awaken the main CPU every 2 or 4 seconds to increment a clock. Unfortunately, the marginal current draw from doing this seems to be in at least the tens of microamps. Additionally, it becomes necessary to have the main CPU powered up while any attached peripherals are powered down; this can necessitate additional circuitry. At minimum, this requires adding a "diode-or" arrangement to the CPU power supply, with the associated voltage drop. |
|
| |
As for existing RTC chips, I'm not aware of any I2C-style chips for $0.25 or less. Small EEPROMs are already extremely cheap, and some CPU's that are needed for other reasons (e.g. 16F6x) already have EEPROM built in. While throwing the RTC die into the same package as the CPU may be a cute notion (such things are already done with EEPROM in e.g. the 12CE518 and 12CE674), I don't think it would buy a whole lot when there would have to be three external pins on the package devoted to it (VBat, X1, and X2). |
|
| |
BTW, if you know a source for RTC chips with a two-wire interface for under $0.25 each (1K quantity) I'd really like to know about it. |
|
| |