[NLRS] VHF low pass filters?

[Dr. Gerald N. Johnson]

Doug you work too hard.

Ed Wetherhold wrote up Tchebyshev (some spell it without the leading T) 
standard value capacitor filters for the 1984 ARRL handbook and the 
design techniques and tables were in ARRL handbooks for at least 20 
years. One place that has the tables on line is: 
d1.ourdev.cn/bbs_upload782111/files_26/ourdev_534491.pdf Clearly a 
country that neglects to protect other's intellectual property rights.

The 6m TVI problem was hit solidly by F.E. Ladd, W2IDZ in June and July 
QST 1954. Pages 21-23 + 114 in June and 32-33 + 124 July. The first 
pages are available to ARRL members on line if you can find their 
archive search. Takes clicking on "Technical", then "ARRL Lab", then you 
can find archive search. You must be signed on as a member to get that 
far. He shows both low pass and high pass with extreme details on how to 
build them, including winding the coils.

High pass filters were hit again by Ed Wetherhold (W3NQN) on pages 30 to 
34 of February QST, 1982. Specifically for TVI prevention. He 
understands filters very well and MAKES them work.

On pages 42 to 46 of the September 1969 QST, Ed Wetherhold shows how to 
design elliptical (Cauer) filters with design tables and detailed 
instructions on scaling them for other frequencies. Often the filter 
design tables are for 1 Hz cutoff frequency and so capacitors have 
values of nearly a farad and inductors may be a couple hundred henries, 
but they are intended to be scaled by simple relations. fundamentally 
100 Mhz is 100,000,000 Hz so the capacitor and inductor values are 
simply divided by 100,000,000. Impedance scaling from 1 ohm to 50 ohms 
is a little more complex. There inductor values go up by the ratio of 
the circuit impedance to 1 ohm, and capacitor values go down by dividing 
with that ratio. In his article he mixes them together with one formula 
for inductor frequency and impedance scaling, and another for capacitors.

There are some details important to making filters have a stop band at 
VHF. Primarily inductance of the shunt capacitors prevent a filter from 
working in real life the way it does on the computer. All capacitors 
have inductance. Let me go to a real tale.

Back about 1964, maybe late 1963 I was working in the high power 
transmitter department at Collins, my first year out of school (before 
any graduate school) and my boss assigned me the task of finding or 
making a set of low pass filters to reduce the harmonics of our HP 
signal generators driving an HP amplifier (HP215 IIRC). I dug back into 
a 1956 QST article (QSTs circulated around Collins and were kept in the 
company technical library) that I remembered showing nomographs for 
designing constant K and M derived filters based on math known since the 
early 1930s or maybe even earlier. Not exactly "modern" filter design. I 
sketched out designs, I don't remember what I used for inductances, but 
Collins stock had a wide selection of molded inductors which I probably 
used but used hand wound coils at VHF. I had a lab tech build them, 
probably starting with a mini box and lined the ends and middle with 
scrap PC board or copper foil soldered together. Connectors were mounted 
to the ends. Shunt capacitors I had him give special treatment. Grounded 
end with NO leads, even if that meant cutting away some of the 
insulation of the dipped silver mica capacitors. The floating end I used 
ONLY the capacitor as a stand off and arranged the adjacent coil axes to 
be at right angles for minimum coupling. Up to about 100 MHz these 
filters worked as designed, above 100 MHz they didn't work well, but I 
found some GR and HP adjustable and fixed coaxial filters that worked 
there to complete the set. We needed these filters to put the harmonics 
of the generator and amplifier set down 80 or 90 dB because we had to 
prove our 250KW PA output network would do that because at that power 
level it takes that much harmonic attenuation to pass FCC and government 
standards.

Later in 1964 I and 7 others moved from CID to Richardson and we took 
that set of filters with us and they were used a lot in testing the 
output networks. (about 1/3 of the transmitter that took up 1000 square 
feet of floor space with a 13foot ceiling was output network). Another 
young engineer was asked to make a second set of filters copying mine. 
So he opened them up and was appalled at my construction and built a 
copy set. In the copies he used plastic standoffs to support the 
capacitor to inductor junctions and ground lugs for the grounded end of 
the capacitor leads. When tested his didn't have the harmonic rejection 
mine did because of two things. The inductance of the shunt capacitors 
with perceptible lead lengths and the capacitance of the insulating 
stand offs. The inductance made the caps go series resonant and above 
series resonance the pair is inductive. The capacitance of the 
insulating stand offs parallel resonated that inductance at some VHF 
frequency so the shunt C impedance was open circuited giving the filter 
stop band a very low attenuation pass band. His filters were not tossed 
but kept only for less critical measurements where mine remained the 
standards to be used.

Point is at VHF shunt capacitors aren't pure capacitors.

I presented a paper at maybe the 1980 CSVHF conference in Cedar Rapids 
on using Cauer (Elliptical) filters at VHF where by allowing for and 
adding more inductance in series with shunt capacitor which are inherent 
in such designs I was able to achieve better 2m to 435 MHz isolation for 
dual band satellite operation than the standard cavity filters in use at 
the time. I didn't find that proceedings, probably I need to root for my 
original and put it in my papers on line. Its sure to be here someplace, 
maybe even in a file folder in one of the file cabinets.

That being said, I prefer the constant K filters from ancient designs 
for transmitting. First they are quite predictable using either the QST 
tables (May 1956, pages 31 and 156) or simple formulae published in most 
vintage radio handbooks, amateur or professional. Second, they use 
smaller value shunt capacitors than Tchebychev or Cauer filters and that 
means smaller circulating currents that tend to heat up and damage small 
capacitors in transmitter filters. Third they have a flatter passband 
transmission and input SWR over that pass band. Fourth they tolerate a 
larger error in component value without a severe error in the predicted 
responses. They do need more sections to reach the rate of change of 
more modern filters.

The basic idea of a Tchebychev filter trades off a flat passband and 
continously increasing stopband attenuation for a more rapid transition 
from pass band to stopband. The cost is ripple in both the pass band 
(which shows up at some frequencies as SWR increases seen by the 
transmitter) and the stop band (which shows up as peaks and valleys in 
the stop band. Always Tchebychev filters have higher value shunt 
capacitors which means greater RF current when transmitting. And they 
require greater precision in the parts to achieve the desired response.

Cauer (elliptical) filters achieve even more rapid transitions from pass 
band to stop band at the cost of including critical resonant circuits 
that can be a pain to tune and guarantee much higher component currents. 
They also can have significant ripple in pass band and stop band. Its 
often possible to trade pass band ripple for stop band notches at 
critical frequencies, which would be 100, 150, and 200 MHz for a 50 MHz 
transmitter filter, but those resonant circuits have to hack watts of 
harmonic energy circulating which means the smallest available parts 
probably will detonate. The only load on those notch circuits is the 
component losses so circulating currents and capacitor voltages can be high.

I have a problem with chip inductors with Qs under 30. I've seen 
resistors with higher Q and my Q meters don't work well for such low Qs. 
I'm probably spoiled by ordinary coils having Qs up to 200 if only 1" 
diameter and I've built one (as part of a Collins Study of coil design) 
with a Q over 1000. That one was not small, wound of 1-1/2" copper water 
pipe.

Today powdered iron toroids make decent coils for filters and tend to 
not complicate construction by stray magnetic coupling. ARRL handbooks 
and web pages from vendors of those like Amidon include design details 
for planning on desired inductance values.

73, Jerry, K0CQ

On 1/5/2013 2:51 PM, Douglas H Reed wrote:
>
>
> Based on the WA4DSY filter design web page:
>
> Bandstop filter, 52MHz center, 40MHz wide, 75 ohm impedance
> Selected Chebeshev Tee filter layout.
> 	
> Part Values
> Part	Chebyshev 	Standard value
> L1	0.2530 uH		240nh TK2707-ND
> C1	37.03 pF		36pf
> L2	0.1872 uH		150nh TK3102-ND
> C2	50.03 pF		50pf or 22+27pf
> L3	0.2530 uH		240nh TK2707-ND
> C3	37.03 pF		36pf
>
> Problem with the above filter because standard inductors are not that
> close to preferred value. But the parts are listed as in-stock at
> DigiKey.
>
> Changing to a different bandpass or changing the impedance will make a
> lot of difference in the filter components. You can juggle specs until
> you get reasonable values to use.
>
> 51.5MHz Center, 35MHz BW, 75 ohms looked pretty good for 220nh or
> 240nh inductors. Personally, I'd prefer to stick with unshielded
> inductors for higher Q. SMD parts have even lower Q than small
> shielded inductors. If you download the spec sheet for the part series
> it will probably estimate the tunable range for the inductor you are
> looking at. Also need to look at the self-resonant frequency of the
> inductors and caps used, especially if you are trying to cover the
> whole 50MHz to 2100MHz frequency range in one system.
>
> 73, Dog Reed, N0NAS.
>