Putting two RF filters in series

[CNe7532294]

Out of curiosity I decided to check what this looks like on the spectrum analyzer. Why is it that a peak occurs before dropping off? See picture. Using only one filter produces the expected sharp cutoff/dropoff of higher frequencies. Here I'm using two 4.4 GHz LPFs. This same peak occurs for two 1.8 GHz LPFs as well.

[yl3akb]

Hi!

Its probably due to impedance mismatch at frequencies of filter transition region. You could try to add small value attenuator (3dB or so) between the filters to see if it solves the issue. Of course, loss probably will not be acceptable. Also isolator or amplifier with good directivity could help

[RadioNerd]

Adding a short length of coax between the filters can help reducing these resonance artifacts by shifting the phase of the reflected signal...

[CNe7532294]

@yl3akb I'll have to see where I put my pads, cables and adapters. I was thinking about this explaination too but the cabling is an issue see below.

@RadioNerd See picture responses. Seems to be the case. I forgot about the cable length and corrected that. I still get a "step like" pattern but definitely better.

For everyone here is the info.
Spectrum analyzer: HP 8563E Opt. 026 (male 3.5mm connector)
Tracking Generator: HP 85644A (female N connector)
Two 4.4 GHz low pass filters: HP 11689A (male and female N connectors)
Two 1.8 GHz low pass filters: Two Daden Associates, INC LA1800-10SS or one RLC Electronics L-1636 aka HP 0955-0491 in combination with the former. (all female SMA connectors)

First test was a sloppy setup of one N male to male cable with a 3.5 mm female to N female adapter using two 4.4 GHz LPFs.
Second and third tests had a 3.5 mm female to female adapter with 2 identical SMA male to male cables using any two of the 1.8 LPFs above.

PS: If you're wondering why the tracking generator is uncovered, I just recently fixed it's low band assembly (the microcircuit amp was damaged but is easily replaceable). It works though so that is not part of the issue here. Especially if I go above 2.5 GHz.

[CNe7532294]

Identical 1.8 GHz LPF test.

[CNe7532294]

Mixed brands but identical 1.8 GHz LPFs.

[CNe7532294]

Ok so some interesting things. Balancing the cables and making them shorter does not work for the identical 4.4 GHz LPF. Making a weird looking 6dBm/adapter pile in between smoothes it out. Picture here is the before. Next response is after.

[CNe7532294]

After adding the attentuation contraption. Attenuator is a Tektronix 011-0069-02 2x aka 6dBm pad with N to BNC adapters.

[borjam]

Remember that most filters are reflective, not absorptive.

In a critical application with reflective filters you would need to insert a circulator and a terminator between both filters.

Doing that, the RF energy reflected by the second filter will go to the next port in the circulator, where it can end being fed to a load.

Mini Circuits, for example, sells some "non reflecting" filters. I guess these filters dissipate RF energy as heat. There shouldn't be a problem to cascade several of them.