Aly" <,shfskfjsf@sliuflky4iuhdf.erl> a &#4294967295;crit dans le message de news: 
TNmdnVfkKKuHBrfbnZ2dnUVZ8vidnZ2d@bt.com...
> Hello,
>
> I am working with the Maxim/Dallas RTCs which specify a requirement of 6pf
> load capacitance crystals.  I can only source 12pf.
>
> It is mentioned in the datasheets that a load (12pf) which is greater than
> what the RTC is designed for (6pf); "If the capacitive load is less than 
> the
> crystal was designed for, the oscillator runs fast. If the capacitive load
> is greater than what the crystal was designed for, the oscillator runs
> slow."
>
> Would anyone have any idea how much difference this will make in practice?
> I think I've read on the net in the past that this is only runs into a
> minute or so over each month but can't find that information now.
>
> Just wondering if anyone has experience of this?  A few minutes a month
> isn't a huge problem, it's really just for date stamping.
>
> Many thanks,
>
> Alison





Hi,

Have a look at DALLAS app note :
"Application Note 58 - Crystal Considerations for Dallas Real-Time Clocks"

Yvan 

On Sun, 22 Apr 2007 00:16:48 +0100, "Aly"
<,shfskfjsf@sliuflky4iuhdf.erl> wrote:

>Hello,
>
>I am working with the Maxim/Dallas RTCs which specify a requirement of 6pf
>load capacitance crystals.  I can only source 12pf.
>
>It is mentioned in the datasheets that a load (12pf) which is greater than
>what the RTC is designed for (6pf); "If the capacitive load is less than the
>crystal was designed for, the oscillator runs fast. If the capacitive load
>is greater than what the crystal was designed for, the oscillator runs
>slow."
>
>Would anyone have any idea how much difference this will make in practice?
>I think I've read on the net in the past that this is only runs into a
>minute or so over each month but can't find that information now.
>
>Just wondering if anyone has experience of this?  A few minutes a month
>isn't a huge problem, it's really just for date stamping.
>
>Many thanks,
>
>Alison
>
Are you sure the datasheet says "what the RTC is designed for"? I
would expect not. The crystal has been specified to run on its
frequency with a certain load, not the RTC.

Taking the first hit I find with google, the motional capacitance of a
watch crystal is 2fF. Putting 12pF in series wil give about 7 seconds
change per day. Putting 6pF in series gives 14. So the difference
between 12p and 6p will be about 7 seconds per day.

If your crystals are specified for 12p, just place 2 times 22p (or
better 22p and 27p) at the crystal. I would be really surprised if the
RTC would not work. You might have to substract the port capacitance
of the RTC itself (order of 3p?).

Cheers,

Joop

"Aly" <,shfskfjsf@sliuflky4iuhdf.erl> wrote in message
news:a8WdncTMytRfBrfbnZ2dnUVZ8sGvnZ2d@bt.com...
> "Aly" <,shfskfjsf@sliuflky4iuhdf.erl> wrote in message
> news:TNmdnVfkKKuHBrfbnZ2dnUVZ8vidnZ2d@bt.com...
>
<SNIP>
>
> Rapid;
>
> 90-3313
> 90-3050
>

For anyone reading this thread in future, also search on similar Maxim part
numbers;

DS1305
DS1306
DS1307

Each will bring back different hits but are sufficiant for sourcing useful
information from a personal DIY point of view.

"Paul Keinanen" <keinanen@sci.fi> wrote in message
news:of3m23tkai4ukpruljifrfklfeo4qd90le@4ax.com...
> On Sun, 22 Apr 2007 00:16:48 +0100, "Aly"
> <,shfskfjsf@sliuflky4iuhdf.erl> wrote:
>
> >I am working with the Maxim/Dallas RTCs which specify a requirement of
6pf
> >load capacitance crystals.  I can only source 12pf.
>
> One reason for low availability of very small capacitors is that the
> stray capacitances will dominate anyway. If everything else fails, put
<SNIP>

Thanks Paul :-)  Acknowledged.

"Anthony Fremont" <spam-not@nowhere.com> wrote in message
news:132mcb1jhspiq58@news.supernews.com...
> Gene S. Berkowitz wrote:
>
> > When I used a 12 pf crytal with a DS1202, it gained a minute a DAY.
>
> That's off allot more than a few pF of capacitance is going to make.
These
> things are prone to picking up stray AC noise and counting it.  At 60Hz,
it

<SNIP>

Also picked up off the net is that grounding the can case can limit odd
behaviour.

Gene S. Berkowitz wrote:

> When I used a 12 pf crytal with a DS1202, it gained a minute a DAY.

That's off allot more than a few pF of capacitance is going to make.  These 
things are prone to picking up stray AC noise and counting it.  At 60Hz, it 
could add as much as 158 seconds per day, 132 seconds at 50Hz.  When I was 
tinkering with some on a breadboard, I had to ground the crystal case to get 
it to keep good time.  Using smaller caps probably increased the amplitude 
of the oscillator perhaps drowning out the AC noise. 

On Sun, 22 Apr 2007 00:16:48 +0100, "Aly"
<,shfskfjsf@sliuflky4iuhdf.erl> wrote:

>I am working with the Maxim/Dallas RTCs which specify a requirement of 6pf
>load capacitance crystals.  I can only source 12pf.

One reason for low availability of very small capacitors is that the
stray capacitances will dominate anyway. If everything else fails, put
to 12 pF capacitors in series to get 6 pF nominally, but again the
strays may cause problems.

If there are some free PCB space, making a 6 pF capacitance from just
the stray capacitance between the PCB tracks is not hard. This could
even be adjusted by cutting out or adding sections, which can be
useful if the required load capacitance varies from one crystal to an
other. 

If the capacitance is from crystal terminal to ground, a multilayer
PCB with a ground plane, the capacitance can be formed between a PCB
track and the ground plane through the PCB dielectric, but
unfortunately some PCB materials are sensitive to humidity and the
capacitance may vary.

A DC blocking capacitor and a capacitance diode could also be used if
voltage control is required, but there are temperature coefficient
issues with capacitance diodes.

Paul

larwe wrote:

> On Apr 21, 9:15 pm, Vladimir Vassilevsky <antispam_bo...@hotmail.com>
> wrote:
> 
> 
>>>As a point of interest, note that PC motherboards used to have a
>>>trimcap on the 32768Hz RTC xtal to tweak the timebase. Several minutes
>>>per day were available through a small (<20pF) trimcap.
>>
>>Never seen that on the PC AT and higher. XT didn't have the RTC at all.
>>Will you provide a reference?
> 

I just depopulated an 8MHz XT-clone M/B which had the trimmer; I was
earlier going to suggest specifying a rock high in freq at 12pf and
add the trimmer.

Most XT-clones with RTC and AT clones had the trimmer.

Regards,

Michael

On Apr 21, 9:15 pm, Vladimir Vassilevsky <antispam_bo...@hotmail.com>
wrote:

> > As a point of interest, note that PC motherboards used to have a
> > trimcap on the 32768Hz RTC xtal to tweak the timebase. Several minutes
> > per day were available through a small (<20pF) trimcap.
>
> Never seen that on the PC AT and higher. XT didn't have the RTC at all.
> Will you provide a reference?

Take a look at any 486-class or earlier Pentium motherboard. I am much
too lazy to take photos right now!

> Besides, it is lot easier to correct the time count digitally, rather
> then tweak the trimmers.

I'm not aware that the RTC macrocells embedded in super I/O chips
include such tweaking functionality.

larwe wrote:


> As a point of interest, note that PC motherboards used to have a
> trimcap on the 32768Hz RTC xtal to tweak the timebase. Several minutes
> per day were available through a small (<20pF) trimcap.

Never seen that on the PC AT and higher. XT didn't have the RTC at all.
Will you provide a reference?

> These trimmers have disappeared completely, and I don't think it's
> because the circuits or components are vastly more stable/better
> matched; I think it's because most people are now running OSes that
> connect to an NTP server frequently to correct the RTC.

I think this is because nobody really cares about the RTC accuracy. 
Besides, it is lot easier to correct the time count digitally, rather 
then tweak the trimmers.

Vladimir Vassilevsky

DSP and Mixed Signal Design Consultant

http://www.abvolt.com