[time-nuts] Clock Project Help

Hal Murray hmurray at megapathdsl.net
Wed Oct 28 21:27:57 UTC 2009

> Yeah, the LCD idea would be fine.  Better would be OLED I think.
> Digikey is one solution, but I am not even sure what embedded system
> to look into. Eventually I want to package these for marketing, so
> they will become more specialized once we get a prototype built, so I
> wanted to know what makes a regular dedicated clock go (IC wise) and
> then kind of build my controller... but I am definitely open to
> suggestions. 

I don't know anything about marketing.

If you intend to sell lots of them (Wallmart), then price is everything.  If 
you are targeting geeks, they are probably less price sensitive but the 
market is small.

The dumb battery powered wall clock s have only a 32 kHz crystal and an IC 
that drives a stepper motor.  The less dumb ones include a small CPU and a 
radio chip that can pick up WWVB.  The CPU doesn't have to be very smart.  It 
just has to keep track of time (count), parse the radio signals, and insert 
or mask off a few clock ticks to correct the time.  It's probably a few 
thousand bytes of code.

If you want to talk NTP, you need an ethernet driver, IP stack, DHCP or UI to 
get an IP address...  You are talking real software.  So you need a real CPU. 
 There are lots of single chips that do everything available for the embedded 

If I was doing something like you have described, I'd poke around to find a 
chip I liked and see what sort of development boards I could get.  You might 
get lucky and find a board that has both ethernet and LCD.  (I like the Atmel 
ARM7 chips.  They are probably (way) overkill for this, but I've worked with 
them before.)  For low volume, you can just copy the parts of their design 
that you need.  If your volume ever gets big enough you can consider 
switching to a cheaper CPU.

Here is a crazy possibility.  There are open source software packages that 
run on some of the WiFi access point hardware boxes.  I wonder if any of the 
hardware it runs on have USB connectors (or use chips with unused USB pins 
you could wire up)?  If so, you could just hack the software and plug in a 

Soapbox time:

Since you mentioned NTP, be sure to read:
  Dave Plonka's paper (in the References) is very good.
  I'd call it required reading for any networking course.

It's amazing how easy it is to screwup such a simple protocol in a way that you can't find any problems in the lab but it's really really nasty out in the field after is gets widely deployed.

These are my opinions, not necessarily my employer's.  I hate spam.

More information about the time-nuts mailing list