RX filter problem

Post your awesome projects here for everyone to marvel at!
Posts: 25
Joined: Mon May 15, 2017 2:40 pm

RX filter problem

Postby marc » Sun May 21, 2017 7:04 pm


Did someone really measured the IL of HPSDR/Hermes/Angelia/Orion Mk1 filter ?

I’m trying to understand how is working the RX filter of the HPSDR family (and by consequences all the Anan series except the 8000DLE)
HPF are almost perfect, the distribution of all filters from 1.5 to 64 MHz has been choosen with cleverness (the 1.5 MHz and 6.6 are a little bit too high… but it works)

But I really don’t understand the way LPF are build. All “Alex based” filters are using the very same component values, those defined by Graham KE9H and Phil VK6APH. Here are the simulated rolloff frequencies for the LPF branch

2MHz, 4MHz, 7MHz, 14 MHz, 20 MHz, 30 MHz, 64 MHz

- The two first are too close from the highest authorized 160 and 80 m frequencies.
- 7 MHz filter cuts just at the beginning of the 40m band and do not reject the second harmonic of 80m
- 14 MHz filter forbids any use of the 20m band and do not reject F2 of the 40 m band (could be used only for the 30 m)
- The 20 MHz one is usable for 20m and 17m only
- 30 MHz filter is “limit 29.7” (0.5 dB attenuation on a critical band), can be used for 17, 15, 12 & 10m
- The 64 MHz is absolutely perfect at last… far away enough of the end of the 6m band and ate the right place for antialiasing

But simulation is… simulation, a virtual vision of a perfect world. I decided to build 3 sets of filters (the most “critical” one: 4, 7 and 14 MHz using values given by the Alex schematic and all the Anan literature. Results measured with my VNA are, without surprise, strictly those given by the simulation. A little bit worth on 7.2 MHz with a loss of 30 % of power... it's not a filter, it's a bad joke. A tchebytchev would have been less harmfull than the original Cauer.

I’m surprised nobody asked any question about this lpf distribution. People wishing to use high dynamic frontends are generally trying to "hunt for the smallest dB". And Alex is a filter used for more than 9 years.

Or did I missed something ?

Marc f6itu
User avatar
Posts: 10
Joined: Sun Apr 09, 2017 6:22 pm
Location: Victor, NY

Re: RX filter problem

Postby k2ue » Sun May 21, 2017 10:36 pm

Insertion loss due to mismatch, like a filter approaching cutoff, has less effect on system NF than resistive attenuation -- they are not the same thing.  Are you aware that the very lowest NF amplifiers reach their lowest NF with a very large mismatch?  Conjugate matching is not a guarantee of lowest system NF.
Posts: 25
Joined: Mon May 15, 2017 2:40 pm

Re: RX filter problem

Postby marc » Mon May 22, 2017 7:02 am

Hi Clyde
Many thks for your answer.

Totally agreed. Building filters is always a tradeoff between best NF, best match, best phase behavior, IMD, group delay...

But why these frequencies? <sarcasm > why adding a wonderful predistortion mechanism if your F2 is as strong as without filter </sarcasm>

More seriously, choosing, for example, a 40m LPF with a rollof freq. at, say, 8 MHz would help a lot offering a Z input closer to 50 Ohms, no phase change within the band or so close), les I.L. too, and a good rejection even if you change the Cauer topology for a Zobel one (I think the US name is “m-modified”). Originally, this part of the filter was used for TX operation. When your VSWR reaches 2.0 “in band” just because your filter has decided to, you frankly don’t care about .6 dB or the noise factor.

But you’re right, I just intend to use these filters in the RX path.

I was just surprised by
- The total absence of info, explanation, figures and measures concerning Alex compared to PA3AKE, W3NQN, WB6DHW projects (among many others) .
- the lack of curiosity of hundredth of hams using these filters for the last 9 years without measuring them, without questioning about their performances, without considering re-calculating them (I dig in the TAPR archives, in the Yahoo ML… nothing found on this subject)
I’ll spend the next weeks to redesign Alex with different cutoff frequencies. Who knows, it might interest someone

Thanks again for your response and attention
VY 73’

Return to “PC's & Homebrew Hardware”