Testing a monolitic band filter (3 pins)
|
Hello,
the story: I made a homemade Ubitx. The reception seemed too weak, I tested the different stages and wondered if the 45 MHZ band filter was well centered and rated a good bandwidth (minimum 15 Khz) if I test the filter with a generator I have the maximum at 45.007300Mhz I am at 50% of the incoming signal (-3db) at 45.008600 and 45.006050 which will theoretically make a width of 2550 htz (!) no good(?) Filter is a motorola 20J26 --> unable to find documentation here on ebay (Come from US to france)! As I have a nanovna, I thought I would test it with this tool. So I did a montage based on the PJ case model. PDF file. the filter is between 2 toroids, each primary connected to the Nano i have this curve (see file) more surprising when looking this video from 2mn30 #554 NANOVNA Measuring Various Bandpass Filters - YouTube And some French ham radio, tell me that a good band filter must have at minimun two band filter like this (see file) here the story on Ubitx forum /g/BITX20/topic/calibration_filters/89337641?p=Created%2C%2C%2C20%2C1%2C40%2C0%3A%3Arecentpostdate%2Fsticky%2C%2C%2C20%2C2%2C0%2C89337641%2Cprevid%25253D1649866481766021915%2Cnextid%25253D1649175119081743046&jump=1 So I am lost! My questions: Does anyone have a way to find the doc of this motorola filter? How well to test it on a Nano? The curve does not appear to reflect reality (????) thank you for your answers and hope you were clear cdt |
|
The fixture you have is for testing individual crystals, NOT for testing crystal filters. Loading the filter with 12.5 ohms as you have done will give you a god-awful passband response, as you have already seen. I would expect that a filter of that bandwidth at that frequency would have a required terminating impedance in the neighborhood of 1000 ohms. Sweep the filter with proper terminating impedances and it will look a lot more like you would expect.
|
|
Also, looking at the roofing filter schematic you posted, I see that the input and output coils are 14 turns, with a 2 turn link. The impedance ratio would be the square of the turns ratio, or 7 squared = 49. Multiplying the input impedance of 50 ohms times 49 gives us 2450 ohms, a fair distance from the 1000 ohms I guesstimated, but still a heck of a lot closer than 12.5 ohms!
|
|
Hello Dave and all,
I put together a small model that makes exactly the assembly of the 45MHZ filter part of the Ubitx. The only difference is to have 2 variable capacitor (36 pf max I think) instead of 22pf fixed. here is the best result when playing with the 2 capacitors. The dish is better by 20 db, but not yet perfect, but the slope is less steep (loss of 20db) I feel like both capacitors are all over the place. I will repeat tests by adding 2 capacitors in// of 22PF your opinion cdt |
|
Hello Dave,
Hello Dave W0LEV, I only followed the instructions of Dave W6OQ. He asked me to test this filter in the configuration of a Ubitx. That is what I did. Several people asked for resistance. So I took a part of the Ubitx diagram for the band filter part and tried to test it. It is sold like that and it seems to work because in the US you are many to use it. Now I can¡¯t find the datasheet of this Motorola 20J26 filter, I thought that addressing a US forum might help to find. There are not many Motorola alumni. Now I can also stop busting my head with this and foolishly buy new filters So I should conclude that on this forum, no one can give a band filter test pattern for a nanovna What¡¯s the point of equipping? Have a nice day cdt |
|
I also tried to locate data for your Motorola resonator. I could not.
toggle quoted message
Show quoted text
Likely it's a custom part Mot. used in one or several of their products. I could run a quick BPF filter S21 plot, but that would not teach anything. Playing with ELSIE is highly educational, doesn't require soldering, and is quite user friendly. It's a wonderful teaching and learning tool. I'll attach a screen capture of a 20 MHz BPF I designed in ELSIE for other purposes just to illustrate the target general pass characteristics of your filter. Of course, it's for a different frequency, but this is how it should look. See the attachment. Dave - W?LEV On Fri, Apr 15, 2022 at 7:10 PM Gerard <kabupos@...> wrote:
Hello Dave, --
*Dave - W?LEV* *Just Let Darwin Work* |
|
Dave,
Thank you for your response I would test it tomorrow, because here it is already the beginning of the night. I had downloaded this software, but not tested yet. It is true that there are many software programs and that we cannot control them all. I¡¯ve been interested in radio for a short time, much like anyone who can do it with a little more time after their career. The subject is vast and we learn and discover every day. Amazing what you can do with a coil and a capacitor. See you soon. thank you for your help cdt |
|
On 2022-04-15 12:10:-0700, you wrote:
So I should conclude that on this forum, no one can give a band filter test pattern for a nanovnaNot really my strength, so I was lurking to learn. But /this/ filter is pretty much right out of the mfg notes. Measured on NanoVNA-F this month, and if my notes are correct, using NanoVNA-App-v1.1.208 x86. Does this fit the bill? ~R~ 72/73 de Rich NE1EE The Dusty Key On the banks of the Piscataqua |
|
Hello Rich and others
Thank you for your response The good thing is that your PDF gives a measurement diagram. As said, I never found the documentation of this Motorola filter. I¡¯ll look in my inventory if I find one, I know I have ceramics on an old satellite decoder and do a test to see the curve. I wondered, if I use the same circuit with an input generator from 44.8 to 45.1 Mhz and I draw a curve in relation to injected signal, output signal would be a solution (?) by monitoring that the injected signal is always the same voltage. ---> Ratio input/output curve. In the worst case there are some on Ebay, but it¡¯s a long time to have them from China, if came US or Greece some shipping costs Unless a good soul can send me two, from Europ will be the ideal. there are also special shops but high shipping costs or minimum shopping of €50 for free shipping cdt |
|
Hello Dave and all
I didn¡¯t say it was a ceramic filter. I was actually talking about a test I could do with possibly recovering a ceramic filter on an old sattellite decoder Only to see the form of curve. If you look the sch¨¦matic of the Ubitx, there are 2 coils of 720Nh it's give 200 ohm resistor value at 45 MHZ. I made a measurement directly on the Ubitx with a generator and an oscilloscope This curiously gives a dish from 44.996000 to 45.004000. which would give a dish of 8Khz and a center frequency at 45Mhz. (I am lucky?) with the curvature one can imagine falling to about 15Khz of bandwidth. I checked my generator at frequency, it¡¯s good at a few hertz. Then the measurement may not be in the standards but gives an indication. On a Ubitx the filter is a "WTL 69 45,000" mine's So my idea is to modify the ubitx software and set the clock values in this direction USB --> 57003400. LSB--> 33010400 with BFO 11996500 for SSB filter. But new story with pb's of compilation see here :/g/BITX20/topic/help_programming_arduino/90507972?p=,,,20,0,0,0::recentpostdate/sticky,,,20,2,0,90507972,previd%3D1650166014008181026,nextid%3D1649404434270126304&previd=1650166014008181026&nextid=1649404434270126304 cdt |
|
The schematic from the ubitx of the filter in question follows:
toggle quoted message
Show quoted text
[image: image.png] This is embedded in a 50-ohm system according to the complete schematic. So, for your test fixture, I would not change anything. C210 and C216 show "0" which I presume means "no load" or nothing. Leave them out. C211 and C215 are just DC blocks so 0.1 ?F is appropriate. For C212 and C214, I'd install something like a 3 to 30 pF variable set about 80% to 90% of the max to start with. For L5 and L7, don't bother with ferrite coils from some "recipe". Instead of that, use an air wound inductor: Diameter = 0.5-inches, length = 1.0 inch, = number of turns = 12. That should pretty much replicate what is in the schematic of the ubitx. Your 50-ohm impedance of the NANOVNA is appropriate for a through measurement (after a complete cal. of the NANOVNA). Test only one resonator at a time. Once you are satisfied with that result, add the second with a small ( several pF) coupling capacitor) or, better yet, a 3 - 30 pF variable. Dave - W?LEV On Sun, Apr 17, 2022 at 9:09 AM Gerard <kabupos@...> wrote:
Hello Dave and all --
*Dave - W?LEV* *Just Let Darwin Work* |
|
Hello "the" Dave,
On the Ubitx forum, a friend tell me the reference of the filter on an original unitx. It's "WTL 69 45,000" I couldn¡¯t find the datasheet on the internet. On E.. I find: There are also some on Digikey, farnell etc, but the cost of transport is high 18€, except for a minimum purchase of 50€ The weird thing is, I did a measurement directly to the oscilloscope out of the filter and injected a signal into its input with a generator. I have a nice "flat" from 44.996000 to 45.004000 (or 8Khz) and battery centered at 45MHZ! It drops to the extremities of these frequencies, then nothing. So that leaves about 15 Khz of band. The test with nano, will give at the extremities, 44.995130 and 45008089, with a center (Despite the curvature down) of 45.002036.( Or 13Khz of bandwidth). I don¡¯t know why I have these two different types of behavior. Then when we look at this, there seems to be only one capacitor and 2 filters (3 welds below) and 2 toroids, unless 2 capacitors are hidden. Unable to read references Dave, tu as dit que es Motorola 20J26 pour faire un test ¨¦ventuel Mes toroids sont bien des T50 /6 (yellow) cdt |
|
Hello,
Making a model reproducing the 45MHZ pass filter of the ubitx (but with variables capacitors) and connecting at each end a toroid transformer I managed to improve the tracing of my filter. Or a slightly flatter shape at -2db and 16.15 khtz width at -3db with best adjust of capacitor If I want a perfect dish, is there a manipulation to do? I was wondering if the middle point must be absolutely connected to the ground, or if there is a possibility of connecting it to the ground via a small capacitor? your suggestions. cdt |
Gerard
Jim Allyn - N7JA
Rich NE1EE
