开云体育

-----Message d'origine-----
De?: [email protected] <[email protected]> De la part de Shabaz Yousaf via groups.io
Envoyé?: mardi 1 septembre 2020 02:45
??: [email protected]
Objet?: Re: [N2PK-VNA] Matching for a Xtal filter

I should also mention, I did all measurements with the RF-IV sensor. It seems to work so well, I didn't use a directional coupler.

Thanks,

Shabaz.


________________________________
From: [email protected] <[email protected]> on behalf of Shabaz Yousaf <shabaz_yousaf@...>
Sent: 01 September 2020 01:34
To: [email protected] <[email protected]>
Subject: Re: [N2PK-VNA] Matching for a Xtal filter

Hi Marc,

Thanks for the information, it's great to hear about the procedures used. I've not used the touchstone format yet, that's something I'm looking forward to. I didn't realize VNWA had a matching tool to solve this.
I guess I just needed to persevere, I spent hours today on it, and finally have something I'm happy with for now.
Basically, after thinking about it last night, I figured I must have missed the matched spot due to too few points or something.
Also, my setup was crude. I re-did it all today, replacing the ferrite toroids with more stable T50-6 iron cores, and removing the trimmer capacitor (which was hard to trim) with fixed ceramic capacitors. Although the online calculators (based on the QMF filter datasheet value of 910 ohm//25 pF) had calculated 71 pF, I found that 100 pF worked better to shift some of the peaks more closer to the center of the Smith chart. Also, I had 23 turns of wire on the T50-6 cores for 2.4 uH, and I found I needed to squeeze them to occupy half of the toroid area, i.e. an inductance slightly higher than the calculated 2.4 uH was needed.

Now the filter result looks like this:

The passband peaks are -2.5 dB which is within spec, although the troughs are a little out of spec, but this was a filter bought from ebay so probably to be expected. The stop-band is only -50 dB, which is not good, but I'm guessing it is due to lack of shield between the input and output of the filter, although I did try to position the toroids far apart. A photograph of the construction is here:


I terminated the filter with 50 ohm and did a S11 measurement. The VSWR is 1.4 at the filter peaks, the graph is here:


The other port is only slightly worse, at VSWR of 1.6, graph here:


I placed a sketch of the design here:


Many thanks,

Shabaz.


________________________________
From: [email protected] <[email protected]> on behalf of Marc olanie <Marc.olanie@...>
Sent: 31 August 2020 09:03
To: [email protected] <[email protected]>
Subject: Re: [N2PK-VNA] Matching for a Xtal filter

Hi Shabaz

I'm a lazzy guy and definitely not a "tech pundit". But when I'm playing with xtal filters, I first run a transmission measure in both direction, store the datas in an S2P touchstone file format with at least 1000 to 2000 measurement points (due to the high Q of these devices)

Then I use the VNWA software, and more particularly it's "matching tool" to simulate the input and output impedance transformer, an eye on the ripple of the amplitude trace, an eye on the C// value.
It's a pure "visual" tuning approach and the final "real life result" is identical to the simulated Z match.


VNWA is fully compatible with the N2PK VNA and could be downloaded from the following url

I indifferently use vnwa and myvna, depending on the abilities and qualities of each software.

Cheers
Marc f6itu

-----Message d'origine-----
De : [email protected] <[email protected]> De la part de Shabaz Yousaf via groups.io Envoyé : lundi 31 ao?t 2020 05:20 ? : [email protected] Objet : [N2PK-VNA] Matching for a Xtal filter

Hi Everyone,

I hope you're well. Although I have some RF background, I don't do this for a day job, I'm very much a beginner so apologies if I'm asking something basic.

I've never had to match such a filter before, but long story short, I have a 8-pole filter model "QMF 10725" which according to the PDF datasheet<> looks like 910 ohm in parallel with 25 pF.
I wish to match to 50 ohms input and output.

According to calculators, that means that I can use a series capacitor of 72 pF and a parallel inductor of 2.4 uH.
I didn't get good results when I tried it (not sure how precisely I need to trim the capacitance and inductance - I would say I had the inductance and capacitance to within a few percent since I measured them at 10.7 MHz).
With the matching circuit attached to the QMF filter, on a Smith Chart I could not get the 10.7MHz point anywhere near the centre, I was just circling not far from the perimeter!

So, I decided to measure the filter on its own (i.e. without any matching) using the RF-IV sensor, to confirm if it is 910 ohm//25pF or not, and I see Rp and Xp as shown in this screenshot:


I'm fairly confident I'm using the RF-IV sensor and MyVNA correctly (I tested with known resistances to confirm the OSL calibration), so I have confidence in the measurements.

As can be seen, in the filter passband, Rp varies considerably. Xp also varies a lot, in some places inductive, and some capacitive.
How can I match it do you think? Should a series C and parallel L be fine, and if so, what values would you suggest?

The CSV file of the data is here:


The Extended CSV is here:


Many thanks,

Shabaz.