[600MRG] LPF - Pi vs Tee form and Constant K impedance match

[Roger Graves]

Frank, thanks for the nice comments. That is a great idea re. using an SWR bridge as an easy check for dirty TX output!
73,
Roger

On May 14, 2017, at 5:31 AM, Frank Lotito <k3dz at live.com> wrote:

> Roger - EXELLENT posting on the filter measurements!  YES indeed, lots of folks forget filters are all to frequently evaluated using a pretty ideal fixed resistive load and not real loads!  About the only antennas I am aware of that present a relatively constant impedance over a wide frequency range are multiband trap antennas, e.g. HF yagis and verticals, and log periodics.   
> 
> Regarding filters and harmonic / non-harmonic related rejection:  In my own backyard for HF I use non-resonant dipoles, fed with home brewed parallel wire feed line, and tuned with home brewed mono-band air-core coil antenna couplers at the transmitter (shack) end.  Its amazing how many classic era and historic era transmitter's output is filthy with harmonic and non-harmonic related output!  I have also found similar filthy outputs on some "modern era" QRP rigs, factory assembled and kits, such as the Small Wonder SW 30 and SW 40 QRP mono-band transceivers.  The dead giveaway I use without firing up my Hameg Spectrum Analyzer and sniffing around:  I tune up the antenna system  using a modern quality transceiver (e.g. ICOM 746PRO), switch over to the classic era etc rig - If the SWR remains the same as the modern rig, I'm OK.  If it climbs, do not tweak the antenna tuning system.  I install an appropriate low-pass filter at the output of the transmitter (transceiver) producing the filthy (high SWR) output.  However, my buddies using multiband H antennas do not see this SWR increase, or see just a little SWR increase which they try to correct by tweaking their simple L-March (all to frequently the Hi-Pass configuration) antenna tuner.  
> 
> Now for 630 / 1700 / 2200 meter reception - UGH, yes indeed, the problem of BCST band rejection, etc. is far more difficult and definitely deserving of some careful experimentation as you have done at your station.  
> 
> Again, EXCELLENT!  To bad we can't have more thought provoking message board articles instead of the all to frequent drivel "you hear me, I heard you" with no technical substance.  That's my opinion - HI.
> 
> 73 Frank K3DZ / WH2XHA
> 
> ________________________________________
> From: 600MRG <600mrg-bounces at w7ekb.com> on behalf of Roger Graves <ve7vv at shaw.ca>
> Sent: Saturday, May 13, 2017 7:31 PM
> To: 600mrg at w7ekb.com
> Cc: Alan
> Subject: [600MRG] LPF - Pi vs Tee form and Constant K impedance match
> 
> With my Pi form Low Pass Filter between the tapped bucket coil 630 mtr tuning inductor and my RX which has a 50 Ohm input, I looked at suppression of BC stns near 950 kHz as a check on potential 2nd harmonic suppression. The difference between no LPF and w/ the filter was only 12 dB. When driven by a 50 Ohm generator the suppression was 26 dB, just as predicted by the design program. I built a 2nd filter, with identical parameters but Tee configuration. This one also showed 26 dB suppression with a 50 Ohm source but now the BC stns were 32 dB down when driven by the tapped inductor, which must be wildly reactive and way different from 50 Ohm resistive at 950 kHz. So, with my antenna system, the Tee configuration LPF provided 20 dB better suppression of 950 kHz. For my SDR RX, 20 dB btr suppression of BC stns is important for eliminating IMD spurs.
> 
> The btr suppression with the Tee form should also be seen going the other way if the filter were driven by a 50 Ohm source TX. TX output circuits will likely not be 50 Ohm resistive, especially on the harmonics, so the actual harmonic suppression of a LPF should be measured, not assumed based on the LPF design plot shown for pure 50 Ohm input and output.
> 
> BTW, I discovered that besides the “Voltage Effective Gain” plot, the “Input Impedance” plot on the filter design programs is very important. The nice sharp cut-off filters I was designing (with 50 Ohm termination) turned out to have non 50 Ohm input Z at 475 kHz, and the Z curve was very non flat around 475 kHz. I found that a “Constant K” design with 50 Ohm reactance of the L and C elements provided a nice flat 50 Ohm match. The lower Q would also reduce voltage on the L and C (which would be of interest to those running more than my 1 Watt output!).
> 
> 73,
> Roger, VE7VV
> _______________________________________________
> 600MRG mailing list
> 600MRG at w7ekb.com
> http://w7ekb.com/mailman/listinfo/600mrg_w7ekb.com