I would like to see a clock on some future product. Unless the internal timers have good accuracy over extended periods of time like three years?
TOD, RTC
(7 posts) (4 voices)-
poslathian
LeafLabsFunny you should mention that, we just had a long discussion about RT clocks. The internal oscillators are certainly not up to snuff for long duration sync (~3 years). I'll have to check the exact specs on the crystal we use. The stm32 onboard maple can be configured to accept real time (32.768KHz) crystals. Honestly, doing a shield with a RT clock is a pretty lightweight project, but its pretty low on our stack. We'll keep this thread posted with any updates about a clock shield.
Posted 5 years ago # -
I've used this little i2c RTC module before, expensive ($20) but comes with a battery and there's arduino code out there for it that should be easy to port to our forthcoming I2C library:
http://www.sparkfun.com/commerce/product_info.php?products_id=99
For great stability (seconds over several years) I think you'd definitely require a special built IC/board. I bet you've looked in to it, but you can also hook in to radio transmitted clock signals (http://en.wikipedia.org/wiki/Radio_clock ), the GPS system, NTP over the internet (http://en.wikipedia.org/wiki/Network_Time_Protocol ), or the GSM/Cellphone network. These little devils are pretty good too: http://www.nist.gov/public_affairs/releases/miniclock.cfm
Posted 5 years ago # -
We don't have a 32khz oscillator on our Native prototype... we're trying to keep costs down and simplify the design/layout. Can you make a good argument for it? ;)
Posted 5 years ago # -
edpell
MemberI think a clock is not a universal need so I would not argue for it give the nice RTC with I2C interface that is available (thank you bnewbold).
What concerns me more with high pin out boards is the interconnect and packaging issues. There is no standard. I guess the 0.1 inch pitch is a standard (Arduino and LeafLabs). But no nice standard for holding various boards physically, no power wiring standard, no bus wiring standard. This is for folks like myself that can not afford to build custom boards for each application.
I guess ribbon cable is the wiring standard (do not know what speed that is good to). It would be nice if pins came in 4 bit and 8 bit sizes (not six for instance) and power came in some single standard like vdd and gnd, without either being interspersed among signals (or when speeds get high enough a systematic interspersing of say signal, vdd, signal, gnd, signal, vdd, signal, gnd, etc), using a single type header one row or two rows or three rows but not a mix. As to how to physically hold various board I guess a backplane. Maybe mechanical only. Those screw hole would work nicely with two tall metal pins and some spacers.
Ed
Posted 5 years ago # -
poslathian
LeafLabscomplicating the issue is that we want to make oak and native shield compatible. Also headers are $$, so you have to choose wisely thre. My favorite thing about the maple/arduino headers is that you can jam wires into them without actually setting up a shield, as you would for some of the higher density headers. Ribbon cables are good, but come with bulky and pricey connectors. If we stick with the arduino type header in long rows (there are a TON of pins to break out on native and oak), then monolithic shields (which we like for oak, since things get fast and complicated...camera interconnects etc) are possible, as are wire jamming. I even have a few male-male cables (that i made) that drop right into the maple header, which i find very useful. Also, on oak, we have this plan for a single monolithic bus (with interspersed pwr/gnd) and a CPLD controlled bus arbiter, so you wont need to plug certain things into certain places and can route signals in software on the fpga.
Designing the header scheme is tricky, as you pointed out. Eventually we just have to commit and see what people think, and incrementally improve on another board.
Posted 5 years ago #
Reply
You must log in to post.