[time-nuts] Next Generation Time/Frequency Standards May Require Provisions Preventing Vertical Displacement

Oz-in-DFW lists at ozindfw.net
Thu Sep 30 15:50:47 UTC 2010

On 9/30/2010 8:43 AM, jimlux wrote:
> how stable?  
The parts are generally  ~ 1 ppm over temp and another ppm or two aging.

> I'm sort of curious, I wonder what sort of temperature range
> cellphones are expected to really work over..
depends on the vendor to some extent.  Not all standards spec an
operating temp range.  The bottom end is usually -20 or -40 C though not
at full spec, and -40 is pretty rare. 
> (not necessarily what they're specified for, but what the designers
> see as the "sweet spot").. It's not like people carry their phones in
> pocket on the back of a backpack in -40 weather.
Most phones are not specified to operate this cold.  Even standards that
specify operating temperature range are often not fully complied with.
> I wonder if they're like pager receivers in some sense (e.g. they're
> on all the time, waiting for a call)
No, a lot of effort is spent in letting them spend most of their time in
standby and only wake up every second (or few seconds.)  Most modern
pagers do the same thing.  They indicate to the receiver when they will
be sending data, and when they will be sending device addresses
(generally called the paging interval for cellphones and pagers both.)
> And, as the phone heats up as you transmit, how much does the
> frequency change?
Once the phone is participating in the network it's locked to it.  At
that point the only thing we care about in the TCXO is short term (~
1sec) drift and that's WAY better than 1 ppm. If you are "transmitting"
you are still receiving several times a second and getting frequency
offset updates at the same rate.  It's closed loop.
> It's a real cost sensitive huge volume market, so the specs for a
> cellphone reference oscillator could be highly tailored to a specific
> application.
Yup, and stamped out in the millions per month so they are really
cheap.  Well under a buck. Most are just a slab of silicon, a slab of
quartz, and a package.  All oscillator functions and compensation are
provided in a custom bit of CMOS with EEPROM or fuse programmable
compensation memory.  I didn't think this was possible, but I saw one
yesterday in a GS that was in an SSOT-23 package.  Sheesh.

mailto:oz at ozindfw.net    
POB 93167 
Southlake, TX 76092 (Near DFW Airport) 

More information about the time-nuts mailing list