From: Jonathan Kirwan
Subject: Re: Querry, Kick Starting Crystal Oscilator.
X-Newsreader: Forte Agent 1.92/32.572
NNTP-Posting-Date: Fri, 08 Nov 2002 20:55:25 GMT
Organization: AT&T Broadband
Date: Fri, 08 Nov 2002 20:55:29 GMT
On 8 Nov 2002 12:29:49 -0800, firstname.lastname@example.org (Tim
>email@example.com (Roy McCammon) wrote in message news:<3DCA905A.59FD928B@mmm.com.DELETETHIS>...
>> I've got a processor with a 32KHz crystal. The problem
>> is that the crystal oscillator can take a l_o_n_g time
>> to start and stabilize.
>> My though is to build a 32KHz cmos schmidt trigger RC
>> oscillator and feed that through a small capacitance into
>> the input of the crystal oscillator to get it going.
>I think you'd be better off building your own 32kHz crystal oscillator
>and feeding it to the processor, rather than using the processor's
>oscillator. You suspect and/or know that your existing oscillator is
>flaky; ditching it (since you cannot change it) is the correct
>I originated a thread a few months ago and got some very good
>about excellent 32.768 kHz oscillators. See
That's a nice reference. I'd add that the Intersil (was Harris)
HA7210 controller (data sheet FN3389.9) might be an alternative
to consider or, if nothing else, a small source of ideas
regarding the microcontroller's oscillator. There is also an
application note, AN9334.2, which discusses the start-up time of
From reading, it appears that the HA7210, in the lowest power
mode, can take around 1.5 seconds to start with a good crystal.
They show how to shorten this to around 0.5 seconds and still
get the low power operation. The HA7210 also permits a 'keep
alive' mode but where the output driver isn't soaking up
current, for those cases where fast start is desired and the
1/2-normal current can be tolerated. (Normally, it draws about
10uA at 5V -- in shutdown, this is 5uA.)
Also, the appnote gives a 'hand-wave' at ESR, saying that 32kHz
crystals with more than 50k ohm of ESR should be avoided.