EMRFD Message Archive 12283

Message Date From Subject
12283 2016-02-07 07:13:27 Ashhar Farhan 7 mhz bandpass filter experiments
i needed a better 7 mhz bandpass filter that didn't need tuning. the trimmers available in india are of really terrible quality. so, the design had to have a wide enough passband to allow for 5% tolerance of inductors and caps. i opted for a series tuned triple section band pass. the series inductor worked out to 6uh and the series capacitor was of 100pf, the caps parallel to the ground were of 470pf, 1000pf, 1000pf and 470 pf.

here is a textual representation of the filter:
1. 470 pf parallel,
2. 6uh series 
3. 100 pf series
4. 1000pf parallel
5. 6uh series
6. 100 pf series
7. 1000pf parallel
8. 6uh series
9. 100pf series
10. 470 pf parallel

this was built with SMD capacitors.

1. build this with 39 turns on T50-6 as the inductor. sweeping it with the SNA (Sweeperino) gave 5db loss. after inserting an attenuator to bring the signal down to -30dbm, the signal was tuned into the spectrum analyzer. the signal was measured with and without the bandpass filter. the spectrum analyzer's readings showed only 2db loss. the extra 3db was reported by the SNA due to the square wave nature of the signal. inserting a low pass filter into the SNA's detector confirmed this.

2. the T50-6 inductors were replaced with T30-6 inductors with 40 turns each. it gave exactly the same readings and the same plot as with the T50-6.

3. Molded inductors of 4.9uh were soldered in place of the toroids. they were loosely at right angles to each other on the open bench without any shielding. the measured insertion loss as 4db at 9.6 mhz.

There seems to be no difference of Q between the T30-6 and T50-6 cores. The T30-6 are cheaper and much smaller. 
The real surprise is that the molded inductors showed much better performance than expected. for lower bands where losses are acceptable, the 4db IL bandpass filter can be made without winding an inductor.

Attempts to measure IIP3 failed. the homelab two tone generator's output is -10 dbm. At this level, the filter demonstrated absolutely no observable intermod products at -90 dbc (limits of the lab), putting the IMD at higher than +35dbm probably, it is higher than 40dbm.

- f
12284 2016-02-07 12:19:39 Sandeep Lohia Re: 7 mhz bandpass filter experiments
Farhan Sir,
for BITX as U say & used tripple tuned has an magical advantage over
image frequency.

but I used double tuned BPF, for little less loss, & using both 10 mhz
12 MHz IF for 40 meter, image will be above cut off frequency of LPF.
so LPF is also there to take care of image, M I right?

what we more need is a High pass filter for better RX, mob chargers,
CFL, electronic chokes etc.

Philips PS film 27 PF trimmers have gone absolute long ago from Indian
market. I use 50 PF variable silver coated ceramic.

is it for fixed 7 MHZ only? coz 6uh that high will have better Q but
narrow bandwidth...



12290 2016-02-07 20:30:38 Ashhar Farhan Re: 7 mhz bandpass filter experiments
sandeep, 

i did this for a club project to build a 40 meter bitx. the original TTC doesn't have a good high pass characteristic at 16 MHz. it also had pretty high losses. almost 10 db. 
anyway, let's keep the bitx specific discussions out of this mailing list. u can mail me offline to follow up on that one or better yet, discuss it on bitx group.

- f

12430 2016-02-20 18:03:31 Clutter Re: 7 mhz bandpass filter experiments
Just curious- what was the 3dB bandwidth of your 3-section filter?
Do you have any feeling for what the skirt selectivity might look like?
How about ultimate rejection, assuming excellent shielding?
Was it designed for 50 ohm impedance in and out?

I calculated resonance to be at about 6.5MHz if the 100pF and 6uH
were spot on. Of course your 'trim' could be simply adjusting the
turns spacing on the toroids.

Thanks, 73,

David K3KY




i needed a better 7 mhz bandpass filter that didn't need tuning. the trimmers available in india are of really terrible quality. so, the design had to have a wide enough passband to allow for 5% tolerance of inductors and caps. i opted for a series tuned triple section band pass. the series inductor worked out to 6uh and the series capacitor was of 100pf, the caps parallel to the ground were of 470pf, 1000pf, 1000pf and 470 pf.

here is a textual representation of the filter:
1. 470 pf parallel,
2. 6uh series
3. 100 pf series
4. 1000pf parallel
5. 6uh series
6. 100 pf series
7. 1000pf parallel
8. 6uh series
9. 100pf series
10. 470 pf parallel

this was built with SMD capacitors.
12431 2016-02-20 20:09:09 Ashhar Farhan Re: 7 mhz bandpass filter experiments
david,

first, i must thank you for pointing on an ethical short-coming on my part. I didn't report factually. I should have mentioned the exact values and performance. Here it is :

I had posted the values from my lab book. I went back and noticed how small 'lies' becoming big. The 1000 pf are actually two 470 pfs in parallel. The inductor is with 42 turns on T30-6. I am not sure of how much the inductance comes to, but I have guessed it to be around 6 uH. I should have reported this rather than the 'predicted value'. 

The filter is centered around 7.050 and it is about 800 Khz wide with a reasonably flat response of about 1 db ripple. The 3db points are 900 khz away.  My sweeperino has an 80 db range, the sweep hits the floor at 13 MHz on the higher side and it is about 40 db down at 5 MHz. The sweeper, based on Si570 doesn't go any further down. The loss is about 1.8db. All this was measured on an open bench without any shielding. I cut a strip of small pads in the middle of a copper clad board to solder the capacitors and inductors across them. All the caps that go to ground were soldered on the same side of the strip. 

The EMRFD' GPLA shows almost the same predicted values. I am quite surprised at a number of things. First, this is the lowest loss tripel tuned band pass filter that I have ever made. The skirt on the higher side though expected was quite a surprise.  I remember that I substituted the toroids with store bought inductors of 3.9uh (they look like 1/2 watt resistors), the frequency moved up to 9 MHz but the shape remained almost the same. I didn't copy anything in the lab notes though.

Increasingly, the plots and the values are coming off the computer screen from sweeps etc. The lab notes are the 'old way'  to document. One fails to take screenshots of what goes on, losing much of data in the process.

- f

12432 2016-02-21 00:04:42 Clutter Re: 7 mhz bandpass filter experiments
Hi, Farhan-

Thanks for the very detailed description. I appreciate your sharing
these results with the group. I plan to duplicate your filter here, in
due time, and compare it to a few other designs I know of. The
low measured losses of your design are interesting.

I did a couple of quick calculations. Your T30-6 core with 42 turns
would be expected to produce 6.35uH, according to the W8DIZ
kits and parts site:

http://www.kitsandparts.com/toroids.php
http://toroids.info/T30-6.php

His online calculator yields the 6.35uH value. Of course tolerances
on these inductors are sloppy due to variations in the permeability
from batch to batch and also individual winding technique. The
value can also be 'rubbered' around quite a bit by squeezing or
stretching the turns around the ring, as you know.

I calculated the 'ideal' inductance value, assuming 7.00MHz and a
perfect 100pF capacitor, and I get 5.17uH. Indeed, your coils may
be well under 6uH, or your caps may all be on the low capacitance.
side.

You are better instrumented than me, Farhan. I hope to have a
spectrum analyzer here eventually. Anyway, I'll take your word on
the filter performance. When I build a circuit like this one, I'm usually
playing with someone else's design, and I simply make an effort to
carefully measure all the caps and inductors to make sure I'm
likely to duplicate the design results. My AADE L/C meter comes in
quite handy from day to day for that, and I also have an old Boonton
260A Q meter when I get serious about characterizing parts at
actual RF frequencies rather than at a few hundred kilohertz.
Frankly, I use the little digital L/C meter far more often. It's usually
close enough for what I'm working with.

Thanks, 73,

David K3KY




david,

first, i must thank you for pointing on an ethical short-coming on my part. I didn't report factually. I should have mentioned the exact values and performance. Here it is :

I had posted the values from my lab book. I went back and noticed how small 'lies' becoming big. The 1000 pf are actually two 470 pfs in parallel. The inductor is with 42 turns on T30-6. I am not sure of how much the inductance comes to, but I have guessed it to be around 6 uH. I should have reported this rather than the 'predicted value'.

The filter is centered around 7.050 and it is about 800 Khz wide with a reasonably flat response of about 1 db ripple. The 3db points are 900 khz away. My sweeperino has an 80 db range, the sweep hits the floor at 13 MHz on the higher side and it is about 40 db down at 5 MHz. The sweeper, based on Si570 doesn't go any further down. The loss is about 1.8db. All this was measured on an open bench without any shielding. I cut a strip of small pads in the middle of a copper clad board to solder the capacitors and inductors across them. All the caps that go to ground were soldered on the same side of the strip.

The EMRFD' GPLA shows almost the same predicted values. I am quite surprised at a number of things. First, this is the lowest loss tripel tuned band pass filter that I have ever made. The skirt on the higher side though expected was quite a surprise. I remember that I substituted the toroids with store bought inductors of 3.9uh (they look like 1/2 watt resistors), the frequency moved up to 9 MHz but the shape remained almost the same. I didn't copy anything in the lab notes though.

Increasingly, the plots and the values are coming off the computer screen from sweeps etc. The lab notes are the 'old way' to document. One fails to take screenshots of what goes on, losing much of data in the process.

- f