Low Pass - Looks More Wobbly to me!

Well,

You may recall the basic ideas for a 4 and 6M linear that I discussed back here:

http://g0mgx.blogspot.co.uk/2013/11/now-that-smoke-has-settled.html

I've started to fiddle with the software and have most of the basics working. The software will just about control everything in the amplifier including cooling and alarms. The most important alarm will be the over drive - I suspect even with input attenuation I am going to find it very easy to overdrive the amp.

I started a few days ago to make the low pass filter. My first attempt looked kind of OK and used air cored inductors, but when I hooked it up to the spectrum analyser and tracking generator I didn't like what I saw at all:

The overall response wasn't very flat and at the target frequency I had quite a bit of attenuation. As a consequence and thinking it was my inductors that were suspect I ordered some T50-10 torroids as these would be good at 70MHz and also would create the required inductances with a reasonable number of turns.

The Low Pass Filter design has evolved a little bit and now looks like this in theory:

So after the T50-10s duly arrived a re-wound all the inductors only to find exactly the same result. Hmmm.

So, then I tried:

• Winding Air Cores with a smaller inside diameter
• Reducing the capacitor values and adding trimmers so I could fiddle with the filter
• Going for a curry

None of the above helped much, the results were still rather rubbish at my target frequency i.e. I had buckets of attenuation where I didn't want any.

Then I tried using some different test leads:

Conclusion? The test leads I was using are spectacularly rubbish at 70MHz - Aaarrrghhhh!!!!

I do have a slight bit of attenuation at my target frequency - the slope is starting just a little too soon, and I need to decrease the capacitors either side of the center inductor a tad.

So here's the filter as built right now:

So I can only conclude that the first attempt at the filter was fine, I just didn't know! The moral of this story is to always connect the test leads from the Tracking Generator output to the Spectrum Analyser input without the circuit under test in the way to ensure you have a flat line - do this first!

Here's the ham cat:

Annoying, egh?