This is the test jig I made to test the 40M filter.

I used T37-2 (red) toroids and old junkbox capacitors. L1 has 17 turns and L2 has 15 turns. The filter is built with capacitors as calculated from the design and sometimes may need to be 2 or more in parallel. It is necessary to accept the tolerances created by the capacitor variances and this will show as small variations on the calculated filter response. The inductors are wound to formula from Elsie or mini ring core calculator etc. The Inductors are adjustable and due to core variances may need a turn added or removed it spreading or compressing the turns reaches the end of the range. Step 1 is to adjust L2 (The inductor in parallel with the 120 pF capacitor) for resonance (14.408 MHz in this case) by which ever method is available i.e. sig gen and RF probe or VNA etc. Here is the transmission after adjusting.

As you can see the loss is 0.18 dB at 7.2 MHz so it looks pretty good and is probably an indication that L1 is pretty close

Next change to reflection mode using a VNA and look at the passband region. Shown below is what I saw after tuning L2. Note that there is a shallow dip at around 4.8 MHz and a phase transition from plus to minus 180 deg. What you have to do here is adjust L1 so that 2 defined dips appear and this adjustment is fairly sharp but really does show up when you hit the spot.

Here is the reflection after adjusting L1. See how the two dips have appeared and the SWR curve is quite closely following the design prediction. The return loss is around 30 dB or better in the 60M and 40 M range which is very good for a filter and still better than 26 dB anywhere else in the passband which makes it very suitable for cascading with an HP filter to make up a bandpass filter.

The dips did not come out where calculated exactly but are in tolerance for 10% capacitors which is what I was using. The LC ratios affect the frequency of phase transition in the transmission mode VNA reading and I feel it would over complicate the filter tuning if you strive for the nth degree of adjustment to correct this. Just accept that the filter most likely won't be perfect but will be close enough. When you look at the VNA plots this filter is pretty good really and better than many commercial designs and certainly will do the job. It worked fine on 50 watts run through it. Another puzzling anomaly is the lack of the second dip at 22.586 MHz as predicted by Elsie. I have had several filters exhibit this phenomenon and Jim N2ADR has noted the same effect. I have no explanation for this and would dearly love to find the answer. Another area to investigate is the effect of termination impedence and stray capacitance caused by relay contacts etc. so more to come in the future. All these results and conclusions are experimental and it is quite possible I have it wrong in parts but this is what works for me.