EMRFD Message Archive 11981

Message Date From Subject
11981 2015-12-10 03:33:33 Eduardo Alonso CW filters: trade-offs and group delay
hello friends,

with the help of a friend, I am building a simple telegraphic
superhetereodyne transceiver, with only with few transistors.
Reviewing receiver chain, I see I have a no clear idea the design
criteria of CW audiofrequency filters of a reasonable complexity. My
starting point is

- 3 inductors?
we will use passive components only: 3 inductors and 4 capacitors?
our goal is simplicity, not a high performace.

- Fcentre:600Hz
i think it is a physiologically pleasant frequency, not the most
sensible point of the human hearing system (1500Hz?)

- BW:500Hz
i see this bandwith is of common use, but i do not know why.
I am using a varicap+crystal variable oscillator VXO, with a 10 turns

- ringing (group delay?)
i am not CW operator, but I read the filters have ringing
effects(wikipedia:spurious signals near sharp transitions).
And this is related with the group delay.
Could I find in internet any recording/video to understand this problem?
Could someone give me a short summary of key ideas or a link to some
lecture about the subject?

- Bessel filters?
I did a small search and I see Bessel filters generates low ringing,
but maybe exists another kind of filters with a good compromise

- magnetic materials
Could someone recomend magnetic materials to build audio inductors for
this kind of filters?
Preferable if they are available in mouser/farnell/digikey/partandkits...

thanks for your help
11982 2015-12-10 05:30:19 Russell Shaw Re: CW filters: trade-offs and group delay
12057 2015-12-20 15:47:03 tttpppggg Re: CW filters: trade-offs and group delay
I think you said performance is not that important so I wouldn't go into all those details. At 600 Hz nearby signals 200Hz away could normally overload your receiver but that won't matter so long as you are receiving a strong signal. I see you are treating this as an exercise, but I think as an exercise it might be wisest to start out without any filtering. You won't need a multipole crystal filter since you'll only be receiving strong signals with a few transistors. It's probably even instructive to witness the double signal that you'll have with a low Q IF amp. With one tuned stage you might get some attenuation of the other sideaband, but not much. Probably if you are looking for a best filter in minimal parts would be a single crystal IF filter ahead of the first IF stage, and skip the audio processing puzzle. It will work fine without any filtering if it is just a bench exercise. You'll get your 600 Hz tone.

12061 2015-12-21 09:48:15 Eduardo Alonso Re: CW filters: trade-offs and group delay
hello all,

thanks for all feedback received.
thanks also to EA4FUK for his detailed personal mail.

effectively, my problem is just an exercise.
i am exploring the "very simple TRX" solutions space, (and learning
about group delay).

it is true, a multipole filter could be not needed in a low gain RX chain:
only big stations will be listened.

lesson learned: a QRP transceiver, should be equipped with a QRP receiver.
for an optimal configuration, both chains, TX and RX should be balanced.

anyway, phase-group delay effects are interesting problems, because it
is appearing their impact in human hearing physiology.
(i cannot find a good study of this topic applied to ham radio, only
brief sentences).


It was useful to me the N7VE's UNICHIP-PLUS report.
He recommends filter stages with low Q, to avoid big peaks in group delay.


Fast PH/GD variations make fatigue in the operator.
But who generates "ringing"?
The band white noise and/or a fast noise peak?


Another important question rises in this point:
Should I use a bandpass or a lowpass filter in the audio baseband?

I think my initial starting point could be erroneous: I do not need a
bandpass filter. It is a complete non-sense.

A downconverter_mixer+lowpass_filter, is in fact, a bandpass filter
centered around LO with twice the BW of audio filter.

Again, noise contribution in each configuration (LPF or BPF) and the
complexity could be a trade-off decision.
The analysis could be a bit more complex, if we consider the effect of
the "IF crystal filter" in the total passband noise
(in my particular case, only one crystal).

(the rig will be operated in the field with batteries, mains noise is
not a concern)

thanks for the discussion.
73 from Guildford,
Eduardo, G/EA3GHS