Listen to Latest SolderSmoke Podcast

Saturday, January 12, 2013

How to shape audio in simple DSB gear?

How should we handle the need for frequency response shaping in simple DSB transmitters?  If we don't roll off the lows and the highs, we risk wasting a lot of energy on RF that will be outside the passband of the SSB rigs on the other end.  This is especially worrisome if we use those cheap electret mic's that seem to have response curves from DC to daylight (well, maybe not that high, but you know what I mean). 

This is not a real concern with SSB rigs, because that crystal filter keeps our signals on the straight and narrow (!) but with DSB rigs, what is the best SIMPLE way to keep the audio between 300 and 2500 hz?    Is there an alternative to the 741 op amp configured as an audio bandpass filter?  

73  Bill N2CQROur book: "SolderSmoke -- Global Adventures in Wireless Electronics" Our coffee mugs, T-Shirts, bumper stickers: Our Book Store:


  1. Hi Bill,

    An audio passband filter would probably not be what you want for voice audio shaping as it is difficult to make a very "wide" bandpass filter that has a flat response in the middle - this, because you need to cover nearly a 10:1 frequency range (300 Hz to 2500 Hz).

    What you do instead is to cascade a highpass filter to pass 300 Hz and above a lowpass filter at about 2200Hz or so and leave everything in the middle alone.

    One of the easiest to design and build - and among the more forgiving in terms of parts tolerance - is the Sallen-Key configuration. I would suspect that a 2nd or 3rd order filter would give you the sort of response you wish.

    A book that is handy for explaining and designing simple filters of this sort is the "Active-Filter Cookbook" by Don Lancaster, but there are similar books as well.

    For the heck of it, I plopped down a somewhat cluttered LTSpice file here:

    This bandpass filter, which can be built using about any type of op-amp, is a 3rd order low-pass followed by a 3rd order high-pass with a bit of RFI protection on the input (e.g. R1/C1.) It has about 12dB of gain and should operate from whatever supply voltage range makes the op-amp happy. According to LTSpice, it will roll off 3dB at 300Hz and 2.9 kHz and be down by 10dB by 200Hz and 4 kHz. The only caution is that none of the capacitors used in the filter itself (the 0.1 and 0.01uF) should NOT be ceramic since those go all over the map with temperature - use metal-film or other non-ceramic units for these!



  2. In a followup, I meant to include another comment in my post to better answer your question.

    It's tricky to get a "good" response without having to resort to a slightly complicated "wide-band" bandpass circuit (e.g. lowpass and highpass cascaded) as described in the previous post.

    A simple R/C can't get better than 6dB/octave response - which isn't really sharp enough to be much help - and you really can't cascade those simple R/C filters to get a better response without ending up with something that is "haystack" shaped - that is, a "rounded" passband with broad peak in the middle that sounds pretty bad on the air as it has neither good highs or good lows!

    To make matters worse, such a circuit can be source/load impedance sensitive and if something is changed on either side of it, you must make sure that the filter response isn't affected much.

    73 again,


  3. Hello Bill,

    Here, G/EA3-GHS Eduardo, from Bristol, UK.

    I think the filter has two objectives

    1 improve the spectral efficiency
    2 improve the power efficiency

    1: The voice has more power in some frequencies than in others. Then you need to measure the power ratio between the total voice power versus the 3..20khz power and see if the optimization is valuable, or is a waste of money. I think it is not necessary.

    2: Filtering voice between .3 and 3kHz (or a bit less) is enough for a voice communication. In the same shortwave band you can put more voice channels in a optimally way.
    I think it is necessary for this reason.

    73 eduardo

  4. In my opinion, there are two big advantages in a DSB transmitter: 1) Low part cound, 2) High quality audio, free from filter distortion. (By definition we're talking about QRP here.) The great majority of voice power is in the .3-to-3 kHz band, so the power-saving benefit of filtering is slim. Not worth giving up the simple design and clean sound of DSB.

  5. From Simon GW0NVN N1XIH

    Hi Bill,

    As Eduardo wrote and Mike said.
    But we also forget our Regulators ( FCC, Ofcom, ERO etc)requirements for amateur radio transmitters. If not curtisy to other radio amateurs up to 20 KHz either side of where our supressed carrier would be. ( I know there is little voice power above 3.4 KHz and in some cases the antenna system bandwidth can have a marked effect )

    But filtering may also improve inteligebility at the receiving end. Giving more contacts. In the same way some filtering on receive may make reception of some stations less fatiguing.

    If you are using a large case these additional circuits can be built, tested and evaluated. Using an 'Altoids' tin with a minimum of components. Every component has to have a function. So a carbon or dynamic microphone would be a better.

    A simple low power minamalist transceiver does not have to be a low performance one. The art is in the design and making each component count.

  6. Hi again, Bill.

    Your post got me to thinking (as it should!) about ways that a simple filter to shape the audio for DSB gear. Initially, I'd rejected the use of inductors and capacitors since this would likely involve hard-to-get and/or expensive components that couldn't be easily replicated.

    It turns out that this isn't so: Rummaging around my collection of transformers I checked the inductance of those cheap ($2.99) Radio Shack 8 ohm to 1k (center-tapped) audio transformers and found that they were just suitable for the purpose and a useful filter can be built with just two of these things!

    The insertion loss is sort of high (6dB-20dB) - mostly due to the fact that one needs to precisely source/terminate their impedance and this is most easily done with (lossy!) resistors, but a simple amplifier can recover this.

    Details are here on my blog:
    L/C audio bandpass filter using cheap audio transformers




Designer: Douglas Bowman | Dimodifikasi oleh Abdul Munir Original Posting Rounders 3 Column