Searching \ for 'What caps... Oscillator design' in subject line. ()
Make payments with PayPal - it's fast, free and secure! Help us get a faster server
FAQ page: www.piclist.com/techref/timers.htm?key=oscillator
Search entire site for: 'What caps... Oscillator design'.

Truncated match.
PICList Thread
'What caps... Oscillator design'
1999\02\16@121759 by Barry King

flavicon
face
Section 2 of the PICmicro Mid-range MCU family Reference Manual is
a complete reference on the oscillators, and choosing crystals and
loading caps.  Recommended reading if you are trying to make sense of
the evidence, or go deeper than the usual rules-of-thumb.  I assume
that this book is on http://www.microchip.com for download as a PDF, but I
haven't looked yet.

Also, AN588 has a (very technical) discussion of the electronics
behind crystal and oscillator specs.

The rule-of-thumb value for 8 MHz is 15 to 30 pF.  This is a
starting point that works fine for the majority of situations.

Larger cap values tend to give better frequency stability and
repeatability (strays are swamped by the capacitor), BUT the
oscillator will start up more slowly.  Smaller capacitor values will
give faster startup, BUT less stability.  Another one of those pesky
engineering tradeoffs  :)

Faster startup might matter a lot if you are waking up from sleep.
The oscillator start-up timer makes the PIC very robust against the
odd slow startup, but check that your timing will be OK in the worst
case.  For most projects, that's probbaly all you need to know.

For high reliability across wide temperature ranges, low battery,
etc... (Hmm.  sounds like my product.  Sigh.) ...you will need to
really measure things with your intended components.  Like startup
time at low temp and high temp.  Like startup time at low and high
Vdd.  And like drive level to the crystal, especially if the drive
level spec of your crystal is low, like a strip-cut crystal for 8
MHz.  Overdriving a crystal can cause some intermittant problems and
or rapid aging and destruction of the crystal.

By the way, startup time is often improved by having the cap from
OSC2 larger (high end of the range) than the one for OSC1 (make it at
the low end of the range).  This added complexity (two different
values) is probably only justified if startup time is an issue.

Oh great, now I've gone on and on... I guess you can't get a Ham
radio operator started on crystals without this happening  <g>.

One other interesting tidbit.  The LP oscillator's (probably the
other modes, too) gain is affected by ambient light.  So /JW
prototypes MUST have the window covered with something that is
actually opaque when checking drive level and startup times.  I got
burned by this, and it wasn't until the MicroChip FAE checked with
the factory that we found out why the /JW parts were overdriving and
the OTPs were OK.  Duh.  I should have known that.  FETs are
sensitive to light!

Cheers,

------------
Barry King, KA1NLH
Engineering Manager
NRG Systems "Measuring the Wind's Energy"
Hinesburg, Vermont, USA
spam_OUTbarryTakeThisOuTspamnrgsystems.com
"The witty saying has been deleted due to limited EPROM space"

1999\02\16@130652 by dave vanhorn

flavicon
face
>Larger cap values tend to give better frequency stability and
>repeatability (strays are swamped by the capacitor), BUT the
>oscillator will start up more slowly.  Smaller capacitor values will
>give faster startup, BUT less stability.  Another one of those pesky
>engineering tradeoffs  :)

True, as long as this is not interpreted as putting 50pf on a 20pf xtal.
Thr point is to buy a crystal specified for a larger or smaller loading C.

>For high reliability across wide temperature ranges, low battery,
>etc... (Hmm.  sounds like my product.  Sigh.) ...you will need to
>really measure things with your intended components.  Like startup
>time at low temp and high temp.  Like startup time at low and high
>Vdd.  And like drive level to the crystal, especially if the drive
>level spec of your crystal is low, like a strip-cut crystal for 8
>MHz.  Overdriving a crystal can cause some intermittant problems and
>or rapid aging and destruction of the crystal.

True fact!


>Oh great, now I've gone on and on... I guess you can't get a Ham
>radio operator started on crystals without this happening  <g>.

KC6ETE here :)

>One other interesting tidbit.  The LP oscillator's (probably the
>other modes, too) gain is affected by ambient light.  So /JW
>prototypes MUST have the window covered with something that is
>actually opaque when checking drive level and startup times.  I got
>burned by this, and it wasn't until the MicroChip FAE checked with
>the factory that we found out why the /JW parts were overdriving and
>the OTPs were OK.  Duh.  I should have known that.  FETs are
>sensitive to light!

This one bites with many teeth!  All windowed parts should be covered with
an opaque label (not white paper sticky ones) when operating. White paper
is better than nothing, but opaque is the only way to get repeatable
performance.

1999\02\17@060130 by mlsirton

flavicon
face
I was just having crystal problems myself and did a mini-research on
the subject...

On 16 Feb 99 at 11:28, Barry King wrote:
<snip>
> Also, AN588 has a (very technical) discussion of the electronics
> behind crystal and oscillator specs.

Personally I didn't like this app note.  It seemed to dodge around
the questions I needed answered (start-up times and drive levels).

A very good reference on microprocessor oscillators is Intel
AP-155 (you can get it on their web site).
Fox electronics (http://www.foxonline.com) also have some good app notes on
their site.
Also various data sheets and app notes for National Semiconductor RTC
chips contain some information on component selection for real time
clock oscillators.

> The rule-of-thumb value for 8 MHz is 15 to 30 pF.  This is a
> starting point that works fine for the majority of situations.

As someone else has said, the load capacitance has to match the one
specified by the crystal manufacturer and is the series combination
of the capacitance on both sides of the crystal (and in parallel with
stray capacitance).  At 8Mhz the exact values are not critical.

<snip>
> Vdd.  And like drive level to the crystal, especially if the drive
> level spec of your crystal is low, like a strip-cut crystal for 8
> MHz.  Overdriving a crystal can cause some intermittant problems and
> or rapid aging and destruction of the crystal.

Usually crystals specified as "Microprocessor Crystals" will have
a drive level spec that can be used without a series resistor but I
think emissions and stability would benefit from a series resistor -
possibly around 400 ohm for 8Mhz  (never used one myself).

Just fried some 32.768Khz crystals myself. (EPSON ones, recommended
drive level 0.1uW. yes microwatt).  I have a feeling the method
suggested by Microchip (testing signal level at CLKOUT) does not do
the job properly.  I did find a real time clock chip with a similar
oscillator that has a built in series resistor of 200K for 32.768Khz
crystals. (NSC DP8570)

<snip>

Another practice recommended by various sources is guarding the
oscillator circuit (surrounding the oscillator, caps and pins with a
ground conductor).

Hope this helps,
Guy - .....mlsirtonKILLspamspam@spam@inter.net.il

More... (looser matching)
- Last day of these posts
- In 1999 , 2000 only
- Today
- New search...