[GPS_Standard] 10MHZ SMD TCXO output boost ?

[Dave Platt]

On 08/17/2012 01:52 PM, Dimitri.p wrote:
> Hi everyone.
> I attempted to use the CRYSTEK CVT32 TXCO  (a tiny SMD unit) and 
> although it presents about 1V Pk-Pk to the input of U1, it doesn't 
> seem to be enough since there is no output on pin 1 of U1.
> 
> Regarding U1, (in the interest of full disclosure), I am not using 
> the LTC1485 (Digikey was out of stock) but the SN75LBC176P from Mouser.
> 
> I'll try adding a a 74AC14 as a buffer in between the output of the 
> TCXO and  the input of U1 and in the meantime maybe someone can 
> either shed some more light or suggest a better arranegment :)

I used that same part, and wasn't initially able to latch onto
the output from an Efratom oscillator having a similar P-P
voltage.

The problem turned out not to be the amplitude of the signal, but
the fact that the oscillator's output was DC-coupled into the
SN75LBC176P.  One of the differential receiver's inputs is
grounded, the other is fed from the oscillator.  Unless the
oscillator can swing at least 50 millivolts or so *below*
ground, it won't exceed the receiver's hysteresis range, and
the receiver will stay in one state.

The data sheet for the CVT32 pretty strongly indicates that
this is what's going to happen... the oscillator itself has
no negative supply and cannot swing below ground.

The fix (for me at least) was easy - just add a .1 uF capacitor
between the oscillator output, and the receiver's A input.
It's probably be a good idea to have *some* resistor in the
R1 position (either the 51R termination resistor suggested
in the article, or a 4k7 or 47k pulldown) just to ensure that
the "receiver" side of the DC-blocking cap is at ground potential
on average.

Add such a cap and (if necessary) a pulldown, and your oscillator
will be swinging the differential receiver's pin about 500 mV
below the other (grounded) input, and you should get a good
output.

The CVT32 data sheet "test circuit" shows just this sort of
.1 uF / 10k-to-ground circuit, probably for just this very
reason.

The other option is to leave the oscillator DC-coupled into
the differential receiver, but bias up the second differential
input to about 0.5V so that it's right in the middle of the
oscillator's common-mode range.  This seems like a lot more
trouble... you'd have to cut traces on the board and pooge
in a couple of additional resistors.