[sdiy] crystal bandpass filter / amplifier with CD4049UBC

[Czech Martin]

Your circuit suffers from high gain/parasitic coupling due
to proximity, thus feedback. The pins are close together,
the circuits share the same substrate and the power pins,
which is a very bad idea @ 10MHz.

Single transistors or (if it must be inverters at all cost) Tiny Logic
single gate CMOS would cure this.

quarz filter: remember that a quartz has two basic resonances:
one parallel (very high impedance) and one serial (very low impedance).
These two spots are close together.
Also there are higher modes at the odd "harmonics".
So building a good quartz bandpass is not that easy.

m.c.

-----Original Message-----
From: Magnus Danielson [mailto:cfmd at swipnet.se]
Sent: Mittwoch, 14. Mai 2003 00:04
To: bert.schiettecatte at esat.kuleuven.ac.be
Cc: synth-diy at dropmix.xs4all.nl
Subject: Re: [sdiy] crystal bandpass filter / amplifier with CD4049UBC


From: "Bert Schiettecatte" <bert.schiettecatte at esat.kuleuven.ac.be>
Subject: [sdiy] crystal bandpass filter / amplifier with CD4049UBC
Date: Tue, 13 May 2003 23:14:37 +0200

> hi all,

Hi Bert,

> i've been trying to build a very narrow and steep bandpass 
> filter / amplifier using a couple of hex inverting buffers, 
> some 1M resistors and 10Mhz crystals. 
> 
> I need to amplify a very small 10Mhz sine signal, but it is 
> burried in noise.

But, but, but... this sounds like the arch-definition of applications where you
really, really, _REALLY_ should turn to analog PLLs. A carrier burried in noise
is just what many radio/satellite links use. GPS with it's not so very
directional antennas (anything not being DOWN is OK). OK, for such a PLL you
need a phase detector which doesn't degrade with noise, a loop filter and a
10 MHz VCO or VCXO. The phasedetector of choice would be a 4-quadrant
multiplier and a 1495 should work really well (this is what it was designed to
do and not the LF stuff), the loop-filter should be an active filter, possibly
even with 2 poles.

Think of the PLL as a bandpass-filter with lots of gain. The loop-gain will set
the bandwidth of the filter and the center frequency will track the signal -
hey, that's a useful bandpass filter!!!

A good GPS receiver using tricks like both carrier and code loop tracking and
topped with data-wipeoff can reach Q-values of several hundred milion to over
a miljard (Giga). Deep burried in noise - sure thing!

> I've put a 1M resister in parallel with 
> each buffer and a crystal between buffer inputs / outputs. 
> I've made 6 stages of filtering / amplifying using a single
> 4049. The circuit idea comes from one of the elektor circuit 
> books (306 circuits or something). 

I think I recall it even...

> the problem is that the circuit is oscillating :( i can't 
> stop it.. anyone have an idea why?

Parasitics eating you up. Much gain and just a little stray capacitance gives
you a good feedback-loop. You might want to separate things up and isolate
the stages so you break feedback loops.

Cheers,
Magnus