Topics

12MHz xtals and QER Filter

Rahul Srivastava
 

Hi!,

I have been scratch building a uBitx and had been experimenting with the 12 MHz QER filter. I have noted few things and adopted some from various postings on group.

1) With the set of xtals that I have with 100pF caps and 200Z I/O filter BW is around 1.85Khz for HC49S xtals and 3.3Khz for HC49U xtals. 
    With filter based on HC49U xtals lower BW is achieved using 150pF caps but impedance drops down to 150-160 ohms. Lowering the
    BW also de grades the shape factor to around 1:2. This can be well understood considering that HC49U xtals have a lower motional inductance Lm.
2) As suggested by Alsion 82pF is working just fine in the filter. With BW of around 2.2Khz. Capacitor value is bit touchy between 82 to 100pf.
3) Q of capacitors used has a major impact on filter response. Parallelling 2 values to arrive at desired value results in a better response.

I am attaching sweeps of few different combination's on my SNA.

I would love to hear from others regarding their experience in this regard.

73

Rahul VU3WJM  

Ashhar Farhan
 

wow, rahul.
meticulous work! i will try this out soon.

- f

On Fri, Aug 3, 2018 at 10:47 AM, Rahul Srivastava via Groups.Io <vu3wjm@...> wrote:
Hi!,

I have been scratch building a uBitx and had been experimenting with the 12 MHz QER filter. I have noted few things and adopted some from various postings on group.

1) With the set of xtals that I have with 100pF caps and 200Z I/O filter BW is around 1.85Khz for HC49S xtals and 3.3Khz for HC49U xtals. 
    With filter based on HC49U xtals lower BW is achieved using 150pF caps but impedance drops down to 150-160 ohms. Lowering the
    BW also de grades the shape factor to around 1:2. This can be well understood considering that HC49U xtals have a lower motional inductance Lm.
2) As suggested by Alsion 82pF is working just fine in the filter. With BW of around 2.2Khz. Capacitor value is bit touchy between 82 to 100pf.
3) Q of capacitors used has a major impact on filter response. Parallelling 2 values to arrive at desired value results in a better response.

I am attaching sweeps of few different combination's on my SNA.

I would love to hear from others regarding their experience in this regard.

73

Rahul VU3WJM  

ajparent1/KB1GMX
 

If the sweep rate is too fast the filter will exhibit tilt.

Also when I went t 82PF I used ATC 100B 5% parts I had so the result was less random.
I also found that the terminating impedance was better and the filter ripple was better with a
differently constructed transformer (ft43-2402 2 hole cores).  I would later design a 100 ohm
transformer that gave better results (2.7khz and flatter).

Keep in mind we have no data across more than a small sample of crystals and maybe only
one vendor.  So variation is expected.

Allison

Rahul Srivastava
 

Very true Allison,

Xtal prams will differ from manufacturer to manufacturer hence it is of out most importance that we characterise the xtals that we are going to use.

Here in India I assume that the xtals available in mkt are same with Farhan as well as with me. 12 MHz HC-49S xtals that I measured came around 17.xx Lm. I guess Farhan also had the values in similar range. Surprisingly 11.0592 MHz xtals were way too high with Lm of around 25 hence not an easy drop in for uBitx, I tested these as wanted to move away from nice looking round numbers  every where ie mainly IF and  products from 16MHz in Nano board. Wes Hayward in one of his papers gives a detailed description of difference between HC49U and HC49S xtals in terms of motional inductance. People doing scratch build should be careful about choice of xtals.

ATC caps may not be easy to come by hence parallel caps warraants further investigation. Effect of Q was more pronounced on U type xtals.

Interesting to note your  observations on BLN cores, another point to look into.

Fun we had in school during Physics lab sessions is back...!! 

73

Rahul VU3WJM


ajparent1/KB1GMX
 

Rahul,

Parallel caps also offer lower ESR best is the values are near the same but
it helps.

The biggest is crystals ground for series resonance vs parallel.  IT seems to get me a
lower ESR and Lm.

When I'm doing a lattice or half lattice filter I look for both parallel and series as the 
mode difference for a given frequency tends to work in favor for that.

My preferred frequency for filters are under 10mhz with 9.6, 9 and 8 being most common
but other like 5, 5.0688 have been done.  There is less drift with temperature.
I know that from doing a wider than SSB roofing filter at 36mhz,  it had to be or heated!

Allison

Rahul Srivastava
 

Allison,

Here paralleling of cap arose  due to sheer unavailability of good quality caps in pF range in local mkt. I wanted to optimise the filters to best possible extent on local components so that people assembling uBitx in future can do so with whatever in hand or in junk box. It is very helpful to know the impact of such substitutions on filter response.

I did try out some 8 and 10MHz filters based on half bridge lattice following the work of DL6ZB , OM Rolf here:



I found them relatively easy to make, matching non critical and impedance's  easily manageable. 

Also found some HC49U 4.43MHz xtals in my old component bin they must be at least 25+ years old from Philips from my post student and struggle days, matched them now  a lot of about 25 xtals and all within 80Hz,  Talk about Quality..!! I made one QER SSB filter its plot is attached. Planning to give a try to some QER filters on 4.915MHz saw a picture on net for an add on board for K2 based on this topology, but I too would prefer xtals in 8-9MHz range with former being more common value.  

As you recommended to avoid ringing sweep rate is kept very slow by keeping the sample rate high on SNA at 1000 samples it sweeps at about 20KHz per minute. 

More as experimentation goes on ...

TNX and 73

Rahul VU3WJM