Searching \ for 'Am I overdriving my crystal ?' 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/index.htm?key=overdriving+crystal
Search entire site for: 'Am I overdriving my crystal ?'.

Truncated match.
PICList Thread
'Am I overdriving my crystal ?'
1999\03\16@232954 by Eric Oliver

flavicon
face
Hi guys,

Well, I finished my PIC16C77 proto-board tonight. Haven't done anything
exciting with it except to test it by blinking an LED.  I ran into a
problem though that took me several hours to discover and I wanted to get
some feedback to make sure I'm doing the right thing.

After the feedback from the list regarding whether crystals or resonators
are better, I decided to stick with crystals <g>.  Even though I am in
Texas and blue jeans and beer ain't that bad <g>!  So I stuck a 20Mhz
crystal on my proto-board and the time comes to fire it off .. nothing.  So
through a series of tests, I became convinced that it was my crystal
circuit that was causing me grief.

I then built a small PIC circuit on my bread board to test various crystal
setups.  On the proto-board I was using two 33pf caps.  On the bread board
I went all the way down to 15pf with no joy.  In a last ditch attempt, I
tried the circuit with no caps and, lo and behold, it worked !  I then
bypassed the caps on the proto-board and the proto started working also.

Now the Microchip data sheet ( for the PIC17 .. I haven't printed the C77
data sheet all the way out yet ) recommends the following :

       4mhz            15 - 68pf
       8mhz            15 - 47pf ( with a 330 ohm resistor )
       16mhz           no recommendation
       25mhz           15 - 47pf
       33mhz           no caps, only the board capacitance was present

So I chose the 33pf because it was in the middle of 15 - 47pf.  Now I've
heard about possibly overdriving crystals if the correct caps aren't used.
Unfortunately, I don't know much about crystals and I don't have a scope (
I'm working on my wife for that one ) so I'm a little concerned that even
though the board works, I may be overdriving the crystal.

BTW, I'm using an ECS crystal which is what Microchip uses in most of their
examples.

So, what do you think ?

Eric

1999\03\16@234026 by Dave VanHorn

flavicon
face
>Now the Microchip data sheet ( for the PIC17 .. I haven't printed the
C77
>data sheet all the way out yet ) recommends the following :
>
>        4mhz            15 - 68pf
>        8mhz            15 - 47pf ( with a 330 ohm resistor )
>        16mhz           no recommendation
>        25mhz           15 - 47pf
>        33mhz           no caps, only the board capacitance was
present


>BTW, I'm using an ECS crystal which is what Microchip uses in most of
their
>examples.




First, is it a paralell resonant crystal?  It must be, or you will
never get to sing the right tune, if it sings at all.
Second, what loading capacitance value does the datasheet on the
crystal say to use?  I don't know how uC could pull those c values out
of the air like that. The loading cap values are determined by the
CRYSTAL, not the chip..  In digiKey, a paralell rock is one that has a
loading C specified, if there is no C speced, it's series mode.

Assuming you have a paralell rock:
Take the load C that is speced, subtract about 5pF for parasitics and
chip capacitance, and try that value.
If it runs, listen on a shortwave receiver and see if it's on
frequency. It should be dead on, as close as you're likely to be able
to tell. Tune in WWV on 2.5, 5, 10, 15, and 20 MHz (Whichever you can
get and is closest to the freq you're using) to make sure your
receiver is in cal.

1999\03\17@051342 by mlsirton

flavicon
face
Hi,

On 16 Mar 99 at 23:36, Dave VanHorn wrote:
<snip>
> First, is it a paralell resonant crystal?  It must be, or you will
> never get to sing the right tune, if it sings at all.

This isn't exactly correct.  If the crystal is specified for a serial
oscillator it should run fine in a parallel one but slightly off
frequency (typically about 0.02% above its specified frequency).

> Second, what loading capacitance value does the datasheet on the
> crystal say to use?  I don't know how uC could pull those c values out
> of the air like that. The loading cap values are determined by the
> CRYSTAL, not the chip..  In digiKey, a paralell rock is one that has a
> loading C specified, if there is no C speced, it's series mode.

Microchip gives recommended typical values.  (I'm guessing uC stands
for Microchip 'cause it could also be micro-computer,
micro-controller or something else...)

> Assuming you have a paralell rock:
> Take the load C that is speced, subtract about 5pF for parasitics and
> chip capacitance, and try that value.

1/CL = 1/C1 + 1/C2

CL = Load C, C1 and C2 loading caps on both sides of the crystal.


The 20Mhz crystal should have run nicely with *all* the cap values
you described.  I would look for the problem somewhere else:
1. Bad connections/solder joints.  Is this on some kind of a
breadboard?  Check all GND and VCC connections with an ohm-meter.
2. "Dirty" power supply? Decouple with a high value
(10uF-100uF) and a low value (0.1-0.01uF) with the low value as close
to the IC gnd/vcc as possible.
3. Place crystal and caps as close as physically possible to the IC.
4. Make sure you satisfy the start-up requirements in the datasheet
(Vcc rise time, reset etc.)

I don't think you need one but a series resistor on the order of 100R
from the oscillator output pin would keep the drive current a little
bit lower.

Guy

1999\03\17@102650 by Dave VanHorn

flavicon
face
>> First, is it a paralell resonant crystal?  It must be, or you will
>> never get to sing the right tune, if it sings at all.

>This isn't exactly correct.  If the crystal is specified for a serial
>oscillator it should run fine in a parallel one but slightly off
>frequency (typically about 0.02% above its specified frequency).


We may be saying the same thing here, but I'll re-state. A series rock
WILL NOT run at the specified frequency in a paralell circuit. It MAY
operate, but it will be at a different frequency, and it may refuse to
start at all.  I have seen series rocks that couldn't be made to
oscillate with a microprocessor. (that was the major clue that someone
bought the wrong rocks)

>Microchip gives recommended typical values.  (I'm guessing uC stands
>for Microchip 'cause it could also be micro-computer,
>micro-controller or something else...)


Typical values are better than nothing, but I hear SO many people with
similar problems.. Trying C after C, yet loading the xtal with the
specified load cap (and having a paralell resonant xtal) has NEVER
failed me, even in production of millions of units.


The 20Mhz crystal should have run nicely with *all* the cap values
you described.  I would look for the problem somewhere else:
>1. Bad connections/solder joints.  Is this on some kind of a
>breadboard?  Check all GND and VCC connections with an ohm-meter.
>2. "Dirty" power supply? Decouple with a high value
>(10uF-100uF) and a low value (0.1-0.01uF) with the low value as close
>to the IC gnd/vcc as possible.
>3. Place crystal and caps as close as physically possible to the IC.
>4. Make sure you satisfy the start-up requirements in the datasheet
>(Vcc rise time, reset etc.)


These are all good points, you have to get it all right, or it don't
go! :)
I'd add to #3, to return the xtal caps ONLY to the uP ground pin,
allowing no other device to share that path. It's not critical to the
osc, but it will dramatically increase your EMI if you don't do it
this way.

>I don't think you need one but a series resistor on the order of 100R
>from the oscillator output pin would keep the drive current a little
>bit lower.


I agree, I wouldn't go there until I was sure everything else was
right, and I had checked the drive level against the xtal specs.
Reducing the drive level also reduces the loop gain, and adds phase
shift, which are both directions that will take a "right" circuit away
from oscillation.

Oscillation is the condition where you have a feedback path, gain >= 1
and phase shift of 180 deg.
If you alter the phase shift, then you attempt to alter the operating
frequency.

1999\03\17@122816 by mlsirton

flavicon
face
Hi,

On 17 Mar 99 at 10:24, Dave VanHorn wrote:
<snip>
{Quote hidden}

I have not researched this issue personally but all the literature I
have (from crystal and silicon manufacturers) says the series rock
will oscillate just as well but a tiny bit off frequency (the series
resonant frequency).  There is no physical difference between
those crystals.  It's just that parallel resonant crystals are
specified for a certain load and frequency.  I can give you specific
references if you like.

<snip>
> Typical values are better than nothing, but I hear SO many people with
> similar problems.. Trying C after C, yet loading the xtal with the
> specified load cap (and having a paralell resonant xtal) has NEVER
> failed me, even in production of millions of units.

I am willing to bet 99.9% of these people didn't get oscillations for
other reasons.
I used to work with this guy who would just pull out caps from his
"small pf" drawer without knowing their values and stick them on
the various circuits we made (generally 4Mhz-50Mhz).  Never once did
any of these circuits not oscillate.  We also never used a series
resistor..
.
However, when you work with low freq. watch crystals it's a whole
different ballgame...

Some years ago I've made a design around a TI C50 DSP which would not
run.  After finally getting through to TI's tech support the guy told
me to check all GND and VCC connections.  Turned out some via's to
the ground plane were disconnected.  Since the IC has lot's of GND
and VCC connections, some of them internally connected you can't find
this problem with just a continuity tester.  With a good ohm-meter
I've found about 0.05 ohm difference on these disconnected pins and
after testing an unassembled PCB I found they were actually
disconnected.  This taught me not to assume anything is actually
connected without testing it...  (That board had a TTL crystal
oscillator, these babies always work like a charm if you can afford
the cost...  +5V, GND and a really nice clock).  When I soldered some
wirewrap wires from these pins to GND it ran fine (But I had the wind
knocked out of my sails...)

Guy

1999\03\17@124515 by Dave VanHorn

flavicon
face
>I have not researched this issue personally but all the literature I
>have (from crystal and silicon manufacturers) says the series rock
>will oscillate just as well but a tiny bit off frequency (the series
>resonant frequency).  There is no physical difference between
>those crystals.  It's just that parallel resonant crystals are
>specified for a certain load and frequency.  I can give you specific
>references if you like.



I know that series rocks CAN operate in some paralell circuits at
different frequencies, but they don't always.  I don't know the
specifics myself, but apparently some cuts don't work in paralell
circuits.


>I am willing to bet 99.9% of these people didn't get oscillations for
other reasons.
>I used to work with this guy who would just pull out caps from his
>"small pf" drawer without knowing their values and stick them on
>the various circuits we made (generally 4Mhz-50Mhz).  Never once did
>any of these circuits not oscillate.  We also never used a series
>resistor..


In onesies, you can be a lot less critical.  This is symptomatic of so
many hobby projects. The author did it this way and it worked, but
when others try to duplicate his circuit, they get varying results,
because his circuit as described was on the edge of one or more
parameters.

I agree though, there's "a million ways to die" :)

>However, when you work with low freq. watch crystals it's a whole
>different ballgame...

Those guys are a lot more fussy about drive level too!  I saw one
whole production run blown when an engineer insisted on using a 3
inverter circuit for AT cut xtals, because "It works on all my
prototypes".
50% in production didn't work, the huge drive snapped the little
forks.  (Unlike other xtals, these are actually piezoelectric tuning
forks, and too much drive will literally snap the arms off!)

1999\03\17@130112 by John Esposito

flavicon
face
But what are the differences between a serial and parallel but crystal?

--John






Dave VanHorn <spam_OUTdvanhornTakeThisOuTspamCEDAR.NET> on 03/16/99 11:36:13 PM

Please respond to pic microcontroller discussion list
     <.....PICLISTKILLspamspam@spam@MITVMA.MIT.EDU>

To:   PICLISTspamKILLspamMITVMA.MIT.EDU
cc:    (bcc: John Esposito/gg/DadeInt)
Subject:  Re: Am I overdriving my crystal ?




>Now the Microchip data sheet ( for the PIC17 .. I haven't printed the
C77
>data sheet all the way out yet ) recommends the following :
>
>        4mhz            15 - 68pf
>        8mhz            15 - 47pf ( with a 330 ohm resistor )
>        16mhz           no recommendation
>        25mhz           15 - 47pf
>        33mhz           no caps, only the board capacitance was
present


>BTW, I'm using an ECS crystal which is what Microchip uses in most of
their
>examples.




First, is it a paralell resonant crystal?  It must be, or you will
never get to sing the right tune, if it sings at all.
Second, what loading capacitance value does the datasheet on the
crystal say to use?  I don't know how uC could pull those c values out
of the air like that. The loading cap values are determined by the
CRYSTAL, not the chip..  In digiKey, a paralell rock is one that has a
loading C specified, if there is no C speced, it's series mode.

Assuming you have a paralell rock:
Take the load C that is speced, subtract about 5pF for parasitics and
chip capacitance, and try that value.
If it runs, listen on a shortwave receiver and see if it's on
frequency. It should be dead on, as close as you're likely to be able
to tell. Tune in WWV on 2.5, 5, 10, 15, and 20 MHz (Whichever you can
get and is closest to the freq you're using) to make sure your
receiver is in cal.

1999\03\17@130743 by Dave VanHorn

flavicon
face
>But what are the differences between a serial and parallel but
crystal?


>From a practical point of view, they are each built to perform in a
specific type of circuit.
Series xtals have low impedance at resonance, paralell have high.

All crystals have both a series and a paralell resonant point, not too
far apart, but their behaviour is only specified (guaranteed) at one
of them. Apparently some compromises are made on one point to enhance
the other. There are also different ways to cut the material from the
crystal that work better in one or the other mode.

1999\03\17@131405 by Harold Hallikainen

picon face
       I don't know if you're overdriving the crystal, but I sure like
the resonators I get from muRata!  They have built-in capacitors and are
characterized for different controllers.  They've done all the
calculations and testing.  You've just gotta hook them up!  Have a look
at their web site in Japan (there's a link there from the US site).
There they have a script that searches for resonator part numbers based
on the controller you're using.
       I've now used thousands of 16 MHz resonators on 16c74a.  They
work great!  They're cheap!  They take less board space than a crystal
and capacitors!

Harold



Harold Hallikainen
.....haroldKILLspamspam.....hallikainen.com
Hallikainen & Friends, Inc.
See the FCC Rules at http://hallikainen.com/FccRules and comments filed
in LPFM proceeding at http://hallikainen.com/lpfm

___________________________________________________________________
You don't need to buy Internet access to use free Internet e-mail.
Get completely free e-mail from Juno at http://www.juno.com/getjuno.html
or call Juno at (800) 654-JUNO [654-5866]

1999\03\17@153903 by Eric Oliver

flavicon
face
Thanks Harold, I'll check them out.

Eric

-----Original Message-----
From:   Harold Hallikainen [SMTP:EraseMEharoldhallikainenspam_OUTspamTakeThisOuTJUNO.COM]
Sent:   Wednesday, March 17, 1999 12:11 PM
To:     PICLISTspamspam_OUTMITVMA.MIT.EDU
Subject:        Re: Am I overdriving my crystal ?

       I don't know if you're overdriving the crystal, but I sure like
the resonators I get from muRata!  They have built-in capacitors and are
characterized for different controllers.  They've done all the
calculations and testing.  You've just gotta hook them up!  Have a look
at their web site in Japan (there's a link there from the US site).
There they have a script that searches for resonator part numbers based
on the controller you're using.
       I've now used thousands of 16 MHz resonators on 16c74a.  They
work great!  They're cheap!  They take less board space than a crystal
and capacitors!

Harold



Harold Hallikainen
@spam@haroldKILLspamspamhallikainen.com
Hallikainen & Friends, Inc.
See the FCC Rules at http://hallikainen.com/FccRules and comments filed
in LPFM proceeding at http://hallikainen.com/lpfm

___________________________________________________________________
You don't need to buy Internet access to use free Internet e-mail.
Get completely free e-mail from Juno at http://www.juno.com/getjuno.html
or call Juno at (800) 654-JUNO [654-5866]

1999\03\18@042130 by mlsirton

flavicon
face
Hi,

On 17 Mar 99 at 12:54, John Esposito wrote:
> But what are the differences between a serial and parallel but crystal?

This is from an FAQ on crystals from Fox (They make crystals).  You
can find the whole FAQ and some more data on their web site at:
http://www.foxonline.com/

What is the difference between series resonant and parallel resonant
crystals?

All quartz crystal resonators have a series resonant frequency (f s ,
frequency of lowest impedance). At this frequency, the crystal
appears resistive in the circuit. Crystals can be "pulled" from this
series frequency by adding reactance (capacitance) in series with the
crystal. When operated in combination with an external load
capacitance (C L ) the crystal oscillates in a frequency range
slightly above its series resonant frequency. This is the parallel
(load resonant) frequency. When ordering a parallel crystal, always
specify the nominal parallel resonant frequency and the amount of
load capacitance in picofarads (pF). Alternatively, a standard value
of C L (such as 20pF) can be ordered; the capacitor values are then
calculated to match the crystal C L . Note that a series type will
oscillate in a parallel oscillator, and vice versa. The observable
difference is a frequency offset of less than + 1000 PPM (0.1%) from
the nominal frequency.

<snip> During the final plating the crystal is dynamically tuned. The
crystal is oscillated in a test circuit and the frequency is read via
a counter. While oscillating, a fine stream of silver is applied via
depositioning process in a vacuum chamber. As the target frequency is
approached the stream ( plating thickness) is turned off. As you can
see from the process above the only difference between series
crystals and parallel crystals is that tuning is offset to a lower
frequency in the parallel crystal to compensate for the system
frequency shift cause by the parallel load of C1 & C2.

Hope this helps,
Guy - KILLspammlsirtonKILLspamspaminter.net.il

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