Re: Missing Schmitt Gates??

On Wed, 05 Dec 2007 08:44:43 -0600, John Fields
<jfields@xxxxxxxxxxxxxxxxxxxxx> wrote:

On Tue, 04 Dec 2007 21:53:49 -0500, Spehro Pefhany
<speffSNIP@xxxxxxxxxxxxxxxxxxxxxxx> wrote:

On Tue, 04 Dec 2007 19:54:53 -0600, the renowned John Fields
<jfields@xxxxxxxxxxxxxxxxxxxxx> wrote:

On Tue, 04 Dec 2007 16:47:14 -0800, John Larkin
<jjlarkin@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx> wrote:

On Tue, 04 Dec 2007 16:45:54 -0700, Jim Thompson
<To-Email-Use-The-Envelope-Icon@xxxxxxxxxxxxxxx> wrote:

On Tue, 04 Dec 2007 15:40:47 -0800, D from BC
<myrealaddress@xxxxxxxxx> wrote:

On Tue, 04 Dec 2007 22:17:36 GMT, Rich Grise <rich@xxxxxxxxxxx> wrote:

On Mon, 03 Dec 2007 19:07:08 -0800, John Larkin wrote:
On Mon, 3 Dec 2007 17:57:28 -0800, "Joel Koltner"
"D from BC" <myrealaddress@xxxxxxxxx> wrote in message

A crystal needs a good linear amp.

Everything is linear if you look closely enough...

I am being a little obtuse here -- the kind of oscillator I was thinking
of was your canoncial microcontroller/FPGA clock that doesn't need to be
particularly accurate -- it's common to use 50 or even 100ppm rocks in
such systems; this is a completely different league of oscillator than
those you build for, e.g., fancy RF applications where you're after
2.5ppm or better.

I was never able to get the Schmitts to oscillate anywhere near the
supposed crystal frequency.


Maybe it's a little late in the thread to bring this up, but I'd
think that with the Schmitt characteristics of the input, the crystal
would have to be drastically overdriven, just to get the gate to
notice that there's a feedback signal.

But I wouldn't have any qualms about an HCU inverter or 3. ;-)

Cheers!
Rich

I think Ht for Logic with Schmitt inputs is about 1V @ 5V.

A crystal..well... isn't it just tiny jiggling piece of rock?
Ooops...I might be thinking piezo..
Damn..forgot all my crystal theory...cuts, shapes, modes and all that
jazz.
Anyways.. I can imagine that one has to be kind to a tiny piece of
crystal and not bash it with lots of drive.
However....depends on the precision required..
As someone posted, for clocking an uC or CPU ...who cares about some
drift..


D from BC

A crystal oscillator using an inverter with hysteresis WILL NOT
self-start.

...Jim Thompson


Of course it will self-start. It just won't run anywhere near the
crystal frequency!

---
Nope.

There's no guarantee that it'll self-start because you've only got
one delta V (on turn-on) to cause the crystal to ring, and if it
doesn't ring hard enough to get to the opposite switching threshold
it'll just sit there, squeezed.

At what input voltage? Of course it's assumed you will also have a
high-value bias resistor across the ST inverter.

---
OK, but then I think the ST has to be configured to self-oscillate
in order to pump the crystal up to where it has enough output to run
the ST instead of the ST running itself.

The resistor and some stray capacitance will do that. It can't have a
stable state with the resistor.

The right way to do it is to use an inverter which can be biased so
that the input and the output are both at about Vcc/2 and then let
noise tickle the crystal until it takes off.

The MCS48 used a ST in the clock oscillator IIRC. It would oscillate
at some tens of kHz before the crystal got going. Maybe a "feature"
kind of a limp-home thing if the crystal failed (usually, not always,
open).

---
My 1990 Intel "8-Bit Embedded Controllers" data book states, on page
1-8:

"OSCILLATOR

The on-board oscillator is a high gain parallel resonant circuit
with a frequency range of 1 to 11 MHz. The X1 external pin is the
input to the amplifier stage while X2 is the output. A crystal or
ceramic resonator connected between X1 and X2 provides the feedback
and phase shift required for oscillation"

Then, on page 4-28, they show:

|
+--[C1]---+------+-----2|XTAL1
| | | |
| [C2] [1-11MHz] |
| | | |
+--[C3]---+------+-----3|XTAL2
| |
GND

C1 = 5pF +/- 1/2 + (stray < 5pF)
C2 = (CRYSTAL + STRAY) < 8pF
C3 = 20pF +/- 1 pF (stray < 5pF

Which looks pretty much like a Pierce oscillator.

Finally, in the figure on page 9, they say: "For XTAL1 and XTAL2
define "high" as voltages above 1.6V and "low" as voltages below
1.6V."

I found no mention of self-oscillation, and with that tightly
defined trigger point and no hysteresis, it doesn't seem likely
that's a Schmitt trigger.

No, it's not clear just from the data ***, however IDRC (I Did
Recall Correctly-- hey, it happens..), if you look further back in
your book to AP-155, _Oscillators for Microcontrollers_ (at least I
think it's bound into that edition) and you'll find the following:-

----
MCS-48 Oscillator

The NMOS and HMOS MCS-48 oscillator is shown in
Figure 21. It differs from the 8051 in that its inverting
amplifier is a Schmitt Trigger. This configuration was
chosen to prevent crosstalk from the TO pin, which is
adjacent to the XTAL1 pin.
....
----
(they go on to discuss the thoroughly nasty "relaxation mode" at
~50kHz and the possible associated problems transitioning to proper
crystal operation, with some nice oscilloscope photos)

http://download.intel.com/design/mcs51/applnots/23065901.pdf

(see PDF pages 20-24).

Although the MCS48 NMOS chip is LONG past obsolete, this is still a
good application note and I recommend it to anyone working with micros
and using the internal oscillator.

Best regards,
Spehro Pefhany
--
"it's the network..." "The Journey is the reward"
speff@xxxxxxxxxxxx Info for manufacturers: http://www.trexon.com
Embedded software/hardware/analog Info for designers: http://www.speff.com
.