Efficient telephone audio (300-650Hz) fedback mechanism !!!

Discussion in 'General Electronics Chat' started by vinodquilon, Feb 23, 2010.

  1. vinodquilon

    Thread Starter Member

    Dec 24, 2009
    234
    0
    Can anybody suggest any efficient audio fed-back mechanism through telephone line
    so that the telephoning person can hear audio beeps, by which he can check the
    status of remotely controlled device (teleremote control).

    I have tried two circuits (see two attached files)


    But cannot hear any sounds at remote calling end. I have tried all frequencies in the range 300-650Hz by varying
    POT in the 555 astable audio beeps generator. But no result.

    I have selected this range as in this range there is no possibility for any DTMF tones.
    Thus audio frequency correctly fed-back to telephone line without detected by DTMF IC.

    Audio beep relay turns on only after off-hook relay get activated.

    I know that, in normal telephone line our vocal audio is captured by Mic, in turn converted
    into AF electrical signals. To check further fed-back mechanism ( & to imitate in my project),
    I opened my telephone set. But nothing can find out on a small, but complex circuit board
    inside the telephone set.

    Does Anyone knows any circuits (that should be of workable) ??? :mad:

    I forgot to say, I am feeding square pulses of audio frequency.


    [ I SAY SORRY TO ALL FORUM MEMBERS FOR POSTING THREADS ON THE SAME SUBJECT BASED ON TELEPHONE ]
     
    Last edited: Feb 23, 2010
  2. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    Do you realize that the CM8870 is a DTMF receiver, not a tone generator?

    A 555 timer generates a square wave output, which consists of the fundamental frequency, plus all of the odd harmonics of the fundamental frequency.

    If you want to convert it to a sinewave, you'll need to first generate a 50% duty cycle output, and then filter out the odd harmonics of the fundamental frequency.

    Getting an exact 50% duty cycle from a 555 is somewhat problematic, but it's fairly easy if you clock a D-type F-F at twice the desired frequency with the Q\ output tied to the D input. Frequency stability is another issue; don't expect more than about 3% accuracy even if you're using low tolerance components.
     
  3. vinodquilon

    Thread Starter Member

    Dec 24, 2009
    234
    0

    I think all these problems can be solved by using RC or wein bridge oscillator
    at audio frequency. Thus replacing square pulses with sine waves.
    10k will be replaced with 470 ohm at coupling path. What about my prediction ???
     
    Last edited: Feb 23, 2010
  4. Audioguru

    New Member

    Dec 20, 2007
    9,411
    896
    C4 in your second schematic blocks DC and prevents the telephone line from being off-hook.
     
  5. rogs

    Active Member

    Aug 28, 2009
    279
    37
    It's getting a bit difficult to keep track of your project on more than one thread!

    As Audioguru points out, you cannot loop the line as you've drawn it!

    I have pointed that out in my reply on your other thread --before I saw you had started a new one!

    Couple of other points. The gain resistors R1 and R2 need to kept at 100K, as per your original schematic. The amount of gain you have introduced into the 8870 input opamp will make the signal clip, and the group filters will have no way of decoding that signal reliably. You'll get false decoding, and/or no decoding!
    R5 at 10K is way too high -you're trying to drive back into a (nominal) 600R telephone line -make it more like 1K.

    You need to select frequencies that the 8870 can reject easily. Simply generating single frequencies that are not actually DTMF may not be enough. The third tone rejection circuit in the 8870 will see all 3 frequencies, and reject any DTMF tones as a valid pair, with the third tone added.

    Take the crystal output from pin 8, and use that to drive a divide by 8192 binary counter, to generate a 437 Hz tone. That's close enough to one of the two frequencies that the 8870 rejects easily (350 Hz and 440Hz)

    You'll need to put that output through a bandpass filter, to remove the squarewave harmonics from the digital counter output, otherwise you'll end up confusing the 8870 input opamp again!

    And make sure your final (near) sine wave filter ouput doesn't exceed about 4vp-p, or once again you'll overdrive the opamp, and cause false decoding.
     
    vinodquilon likes this.
  6. vinodquilon

    Thread Starter Member

    Dec 24, 2009
    234
    0
    You'll need to put that output through a bandpass filter, to remove the squarewave harmonics from the digital counter output, otherwise you'll end up confusing the 8870 input opamp again!

    [/QUOTE]
    Why didn't you recommend a LPF (low pass filter), it can easily integrate in the circuit by connecting a capacitor across R5 at top end and other end of capacitor to ground. Capacitor value choose to get 600 Hz cut-off.
     
  7. rogs

    Active Member

    Aug 28, 2009
    279
    37
    A simple passive LPF as you suggest would certainly help.
    Problem is, it will also attentuate some of the signal from the line as well. And it would only be -6dB per octave - not particularly effective.

    Why not try a bandpass filter along the lines of the circuit suggested here ? :

    http://www.captain.at/electronics/active-filter/


    You can then easliy have control over the gain of the signal, and the 'Q' of the filter (and thus the quality of the sine wave) much more easily.
    You can also make sure you are not affecting the DTMF input signals at all.

    I would suggest something like R1=220K, R2=220K, R3=1.2K and C=22nF as a starting point. Using a ST TLO61 opamp, that should give you an approximate sinewave output signal of 2.5V p-p at 437Hz, with a THD of less than 4%. Should be good enough for what you need.
    You might like to use a more modern opamp -the ST TLO61 is somewhere near it's lower limit with a 5v supply. The LMV538 springs to mind, although it is only available in a SMD package, so maybe not quite suitable for a hobby project.

    You could then connect the left hand end of R1 directly to your counter output, and the opmp output directly to your K2 relay input.

    Two further points.
    If you intend to use the 5V supply to power the opamp, remember to make sure your opamp will work with a 5V supply. As I mentioned above, the TLO61 will work, but is near it's limit.

    Secondly, take the + opamp input (pin 3 on the filter drawing) to pin 4 of your 8870 to derive a reference 'half rail' voltage for the opamp.

    Sketch attached of the kind of thing I'm thinking of.

    Just a suggestion.......
     
    Last edited: Feb 24, 2010
    vinodquilon likes this.
  8. vinodquilon

    Thread Starter Member

    Dec 24, 2009
    234
    0
    It is a multiple-feedback narrow band-pass Delyiannis-Friend filter .
    I will use 741 as Op-Amp. I cannot understand why you connect non-inv
    input of op-amp to 8870. I think connecting non-inv input to ground through R2 is enough.
     
  9. rogs

    Active Member

    Aug 28, 2009
    279
    37
    You cannot use a 741 opamp with a single 5 volt supply. It requires aminimum of +/- 4.5V (i.e 9V)

    The purpose of connecting the non inverting input to pin 4 of the 8870 is, as I explained in my previous post, to derive a reference voltage to allow the TLO61 opamp to work from a single DC supply. This pin provide a 2.5V reference voltage for the internal 8870 opamp, and is a simple way of biasing the external one as well!

    You cannot reference the non inverting input of an opamp to the same potential as it's negative supply. You will be able to find more detail on this from the opamp data sheets posted all over the web, if you're interested.
     
Loading...