locked Measuring and matching 10.7 MHz ceramic filters


David Platt
 

I did an experiment this weekend, throwing together a simple impedance-matching test circuit which allowed me to measure a few 10.7 MHz ceramic IF filters using my new NanoVNA-F analyzer.

The circuit was easy to build.  I took measurements of four filters I purchased from Bill Ammons a few years ago.  The numbers and curves I got, let me to conclude that Bill sold me just what he said he sold me :-)

I've posted the schematic, a photo of the test jig, and the curve plots over in the NanoVNA Users group.

https://groups.io/g/nanovna-users/message/11457

Using a NanoVNA and a test jig of this sort, it ought to be possible to measure and sort these filters into matched sets pretty easily, just as Bill used to do (has anyone heard from him lately?)


newaag
 

Dave;
Nice work, and thanks for the heads up on that new analyzer.
I'll post my comments on specifics of the test fixture over there after I join. What are the filter markings, usually something like E10.7X...
As a reminder, we have lots of filter data in files here.
Bob


newaag
 

There are a few ways to match 50 ohms on the analyzer. Using resistors is the easiest, of course, but not convenient if you have more in mind than testing passive filters. If you add buffers on both sides of the filter, you can use it to drive and measure an IF chain overall response in an actual tuner.

The lesson I have learned is that the filters raw response is only a starting point when swapping filters. How the filter responds in the actual circuit, and in conjunction with the other filters (IF wide and narrow match) really needs you to evaluate the whole IF, if possible. To do this, you need I/O buffers to increase dynamic range and allow high impedance on the measure side. This nanovna unit may be able to sidestep that requirement if you use an active probe on the measure side, and it has enough drive capacity to do the job on the stimulus side.

I actually built such a fixture for Ed Hanlon way back along time ago, to go with his Meguro IF sweeper. It could test both filters and the entire IF chain. It was optimized for solid state tuners. You have to be very careful with voltages in tube tuners, especially with high AC chassis potentials (that are not grounded using typical 2 wire plugs). I also built and gave him the probes to do both in-situ IF stimulus and response.

My favorite non-contact stimulus technique was using a small coil of wire (built on a pencil) in series with a 50 ohm resistor. The coil is slipped over the mixer mosfet device, (antenna input shorted). You can then make the measurements, stage by stage. You probe with a high impedance probe, which is properly terminated and than followed by a buffer with 50 ohm output.
If you use an active probe with 50 ohm output, it makes it even easier to go directly into a spectrum analyzer or VNA. I posted a while ago about a fairly inexpensive one for sale ($30) on Ebay from a guy in Europe.

The other variation I built uses a scope display and HP 8601 sweeper, and AD 8307 log amp. This allows for log amplitude displays on the scope screen.
But the nanoVNA is even better. So I just bought one, the nooelec "sanctioned" version on Amazon. It has a 6 month warranty, so there must be some kind of quality check done (I hope). 
Bob       


David Platt
 

On Tue, Mar 3, 2020 at 12:20 PM, newaag wrote:
What are the filter markings, usually something like E10.7X...
The two 280 kHz filter are marked E10.7A, with red dots in the upper left and upper center.

One 180 kHz filter is marked SKM3C.  The other is marked SKM3& where "&" is a trapezoidal blob.


David Platt
 

On Tue, Mar 3, 2020 at 01:18 PM, newaag wrote:
There are a few ways to match 50 ohms on the analyzer. Using resistors is the easiest, of course, but not convenient if you have more in mind than testing passive filters. If you add buffers on both sides of the filter, you can use it to drive and measure an IF chain overall response in an actual tuner.
Yup... lots of range for experimentation here, and different levels of complexity.  I'm probably going to play around with several different passive versions, and then (if the urge is still strong) cobble up a more flexible powered/buffered system.

The lesson I have learned is that the filters raw response is only a starting point when swapping filters. How the filter responds in the actual circuit, and in conjunction with the other filters (IF wide and narrow match) really needs you to evaluate the whole IF, if possible. To do this, you need I/O buffers to increase dynamic range and allow high impedance on the measure side. This nanovna unit may be able to sidestep that requirement if you use an active probe on the measure side, and it has enough drive capacity to do the job on the stimulus side.
I can well believe it.  Considering how much the source and termination impedances affect the passband amplitude and phase, an "in place" selection process would give better results than a cook-book recipe, since every model of tuner (and I suppose every individual example) will present slightly different impedances.  Lots of interactions possible.

I do have an active FET probe or two that I can probably make work for the sort of probing, I think.

I actually built such a fixture for Ed Hanlon way back along time ago, to go with his Meguro IF sweeper. It could test both filters and the entire IF chain. It was optimized for solid state tuners. You have to be very careful with voltages in tube tuners, especially with high AC chassis potentials (that are not grounded using typical 2 wire plugs). I also built and gave him the probes to do both in-situ IF stimulus and response.
Nice!  Having the right goes-inta and goes-outa connections is always a good thing!

I've got another project in the prototyping stage... an FPGA/DAC/modulator composite-stereo signal generator with pretensions of eventually being a sweeper.  If it works out (TBD) it would complement a VNA-based IF analyzer fairly well.

The other variation I built uses a scope display and HP 8601 sweeper, and AD 8307 log amp. This allows for log amplitude displays on the scope screen.
But the nanoVNA is even better. So I just bought one, the nooelec "sanctioned" version on Amazon. It has a 6 month warranty, so there must be some kind of quality check done (I hope).
I dealt with Nooelec a couple of times, buying their versions of an RTL-SDR dongle.  One dongle I ordered was badly damaged in shipment... it must have gotten caught and shredded by a conveyor-belt drive.  Nooelec immediately replaced it, no questions asked.  The replacement works very nicely... a high quality "build".

The NanoVNA family really is an amazing value for the cost.  It won't scare Keysight or R&S, but they can't get anywhere near its price-point!


Paul
 

On the subject of the NanoVNA, I don’t have one yet but wonder if it has enough dynamic range for our needs at 10.7 MHz  for sweeping an IF chain and matching ceramic filers. I must admit when i see software for it like nanovnaSaver it is very tempting. But i think a good signal sweeping source and at the measuring end a 10x probe and the AD8307 feeding a scope in XY makes me feel a little better.
https://mikeharrison.smugmug.com/Tips-and-stuff/SWEEPING/i-6khJFVC/A


newaag
 

I would say yes, as I have tested a perhaps a hundred or more filters on the NanoVNA so far. Make sure you get the H version, which has a higher output level than the earlier hardware. That link to the site you reference is the same as what I had been using for a while (after losing access to my HP demo gear - SA w/tracking source).
The nanoSaver software is the ticket - it has automated measurements for center frequency and bandwidth, as well as fairly accurate group delay plots, all on your PC screen.
But have not used it to sweep a tuner active IF yet. I test filters using passive resistors to match impedance, and also include the 10 pf cap on the output side (per Murata test diagrams). Let me know if you need the exact values. 
I'll try that next, but yes, it does require an active probe or buffer as a minimum. I personally would not be hooking it up to a tube tuner yet without a really good way to make sure high voltages cannot make their way back to the nanoVNA, i.e. coupling transformers, or back to back diode protection / cap coupling / large series resistors on both sides.     


newaag
 

Attached is a screen shot of the nanoSaver testing a 150 kHz ceramic filter. This is the only way I know of to get easy prints from that program. I just did a "print screen" from a windoes PC, then pasted in Microsoft Paint, and saved as a jpg. This uses the analysis capability to measure center frequency and bandwidth. The response is on top, group delay on bottom.
Bob


Paul
 

Nice thanks for sharing. Seems for that task the 40db range is plenty to see the skirts of the filter.
that software sure makes it easy to match filters. Instead of fiddling with markers and doing some math.
if your up to it, maybe print screen it used on a tuner IF.


David Platt
 

Very nice, Bob!

What was your test jig like?  Based on the S21 amplitudes shown I'd guess it was a straightforward two-resistor termination?

Another way I've found to get images out of NanoVNASaver is to right-click on the individual plot in the main screen.  Use the pop-up menu to create a separate window containing this one plot.  Resize that window as desired.  Then, right-click on it again and save the image to a file (e.g. "amplitude.png").  PNG format is a good choice for individual plots because it's both compact and lossless.


newaag
 

Yes, the software, which is free and public domain, is an incredible bargain :-)
This package does what I used to do with an HP Spectrum Analyzer (that had an optional built in tracking generator), but it's even better, as that SA solution was a scalar analyzer (no phase).
The nanoVNA is a vector analyzer, so with phase it can do group delay. And group delay side saddle evenness can vary from filter to filter, even when magnitude (top response) response may not. That's worth knowing when sorting and selecting filters.
No in-situ measurements yet, but will post them when I get there.
Bob


On Thu, Mar 26, 2020 at 08:34 PM, Paul wrote:
Nice thanks for sharing. Seems for that task the 40db range is plenty to see the skirts of the filter.
that software sure makes it easy to match filters. Instead of fiddling with markers and doing some math.
if your up to it, maybe print screen it used on a tuner IF.


newaag
 

David,
Thank you for that tip on saving images. It will save a lot of hassles getting images out directly.

Yes, it is a 2 resistors (each side) termination, using 4 resistors (2 x 300R, 2x 56R) and one 10 pf cap total. Add to that 2 right angle board mount SMA connectors, and a small perfboard with a 3 pin SIP socket, and you gave the complete bill of material. The 56 ohm resistors go to ground at each SMA signal lug, and the 300 ohm go in series from the SMA signal lug to each end of the 3 pin socket. The center pin goes to ground. The cap goes to ground on the output. 
I used a little sheet of adhesive backed copper foil for a ground plane underneath. 
And so of course I included a couple pictures, below.
Bob 


Paul
 
Edited

And this  cost how many thousands of dollars say back in 2004 when this group started, then came used gear, vintage that brought down to few hundred dollars still ....and today less than 150$ sum total. Now that no one cares about FM tuners and that no one can buy bags of ceramic filters anymore to sort. Fun for us however.


newaag
 

I see a stateside source is selling the NanoVNA for $49/$59, which is a great price. Equivalent quality ones on Amazon sell for about $30-50 more. And you don't want a cheap fake with this one.  

It is R and L Electronics, a good source for ham radio stuff. These are the good ones.
If you are considering testing ceramic filters, I would suggest you take a peek at this thing. Certainly buy the 4" display unit for $10. You may also want a couple sma to BNC adapters if you don't have any. This source is not a pure sine wave, it's a square wave where the harmonics are used as the source above 300 MHz. But at least two of us have tested it, as posted above, and it works great with the free PC program for sorting ceramic filters.
This is way easier than buying it from China and waiting 3-4 weeks. It is currently in stock, and ships from Ohio.
I have zero affiliation with these guys selling it. I don't think they have an infinite supply, or the price will stay this low, so I would not wait on this one (if you are serious)...
Bob 


newaag
 

On Wed, Apr 22, 2020 at 11:15 AM, newaag wrote:
It is R and L Electronics,
here is the link to the R and L listing for nanoVNA
https://store2.rlham.com/shop/catalog/advanced_search_result.php?keywords=nanovna&osCsid=uppeqc6n3p44t1hpovgh126r27&x=0&y=0&search_in_description=1

support group for PC software to do bandpass filters (pics above)
https://groups.io/g/nanovna-users/topics

Bob


newaag
 

On Wed, Apr 22, 2020 at 12:48 PM, newaag wrote:
here is the link to the R and L listing for nanoVNA
I forgot to add that the nanoVNA is also great for analyzing antennas and the feedline to them. It is supposed to be for a 50 ohm antenna, but works OK for 75 ohm units as well.
Bob


Joseph Strickland
 

Bob;

Thanks indeed for this suggestion! i bought one of the 4" display type today. I am looking forward to using it for a number of things, but using it to evaluate IF filters will be a plus. It will also be useful for checking the parameters of various antenna designs. I believe it will satisfy a number of applications.

Joe
KC5LY


Radu Bogdan Dicher
 

Bob and David,
Thank you both very much for sharing your trials this far. I grabbed one NanoVNA myself - Bob's source - and plan to get to filters testing next week (or whenever I get some diy time...). 
It looks pretty amazing this far. This would be pretty groundbreaking for tuners' work. Thank you for sharing.
Radu. 


Virus-free. www.avg.com


On Thu, Apr 23, 2020 at 4:14 PM Joseph Strickland via groups.io <strijw426=yahoo.com@groups.io> wrote:
Bob;

Thanks indeed for this suggestion! i bought one of the 4" display type today. I am looking forward to using it for a number of things, but using it to evaluate IF filters will be a plus. It will also be useful for checking the parameters of various antenna designs. I believe it will satisfy a number of applications.

Joe
KC5LY


Radu Bogdan Dicher
 

Bob,
Reading your circuit - it seems you have a 300ohm resistor in series with the filter socket on the output, plus a ~50ohm in parallel (and a 10pF cap). Reading the Murata filters datasheet, the required loading on the output (shown as a parallel R) is 330ohms (+ the 10pF cap). Any reason you've slightly steered away from that loading configuration? 
Thank you, 
Radu.

On Sun, Apr 26, 2020 at 7:51 PM Radu Bogdan Dicher via groups.io <vondicher=gmail.com@groups.io> wrote:
Bob and David,
Thank you both very much for sharing your trials this far. I grabbed one NanoVNA myself - Bob's source - and plan to get to filters testing next week (or whenever I get some diy time...). 
It looks pretty amazing this far. This would be pretty groundbreaking for tuners' work. Thank you for sharing.
Radu. 


Virus-free. www.avg.com

On Thu, Apr 23, 2020 at 4:14 PM Joseph Strickland via groups.io <strijw426=yahoo.com@groups.io> wrote:
Bob;

Thanks indeed for this suggestion! i bought one of the 4" display type today. I am looking forward to using it for a number of things, but using it to evaluate IF filters will be a plus. It will also be useful for checking the parameters of various antenna designs. I believe it will satisfy a number of applications.

Joe
KC5LY


David Platt
 

It looks to me as if Bob has arranged a "minimum loss" pad.  This matches the 50-ohm output impedance of the NanoVNA to the 330-ohm input impedance of the filter, and (similarly) the 330-ohm output impedance of the filter to the 50-ohm input impedance of the NanoVNA's second port.

As an example - working backwards from the NanoVNA's second port - this has a 50-ohm input impedance.  Add about 50 ohms in parallel with it, and it's a 25-ohm load.  Add the 300-ohm series resistor and it's 325 ohms, very close to the filter's expected load.

The same process works on the other side of the filter - the filter "sees" a source impedance of just about 330 ohms.

From the other direction - the filter has around 330 ohms of input impedance.  Add the 300-ohm series resistor and it's around 630 ohms.  Add a bit more than 50 ohms in parallel with that, and you end up with 50 ohms... so, the NanoVNA "sees" a 50-ohm load.

This sort of arrangement is called a "minimum loss" T, because it's the least-lossy sort of purely-resistive bidirectional matching you can do between different source and load impedances.  "Minimum" is a bit of a joke, though, as it's still a very lossy match!