need suggestion with filter

[nandax]

i'd like to amplify the signal from speaker in handset to speaker with 8 ohms impedance,
and i'd like to filter the noise in the sinal before it goes to amplifier.

i'm thinking bout low pass filter but i'm not sure about the value of the frequency of signal

is there any suggestion which filter i should use?

[sreten]

Hi,

Handset ? as in a phone ? the earpiece or handsfree driver ?

/sreten.

[nandax]

Originally posted by sreten
Hi,

Handset ? as in a phone ? the earpiece or handsfree driver ?

/sreten.

handset as in a phone..

[infinia]

Hi
Plain old telephone service landlines are designed to transfer audio with a quality that is just enough for the parties to understand each other. Typical bandwidth is 3.6-kHz; only a fraction of the frequencies that humans can hear, but enough to make the voice intelligible. No phone can improve on this quality, as it is a limitation of the phone system itself.
GSM 3G is claiming capability of at least double this BW but is depending on the service provider to deliver extra performance.


edit>How about 3 or 5 pole Caur/ellipitical low ripple at 4KHz.

[nandax]

Originally posted by infinia
Hi
Plain old telephone service landlines are designed to transfer audio with a quality that is just enough for the parties to understand each other. Typical bandwidth is 3.6-kHz; only a fraction of the frequencies that humans can hear, but enough to make the voice intelligible. No phone can improve on this quality, as it is a limitation of the phone system itself.
GSM 3G is claiming capability of at least double this BW but is depending on the service provider to deliver extra performance.


edit>How about 3 or 5 pole Caur/ellipitical low ripple at 4KHz.

hi.., thanks for the reply..
i'm using the GSM modem to make a call, and take the audio output signal and amplified it. but i'm afraid the noise also get amplified,
i'm thinking to add filter before it amplified,
but i dont know the frequency of the signal it self..
what do u mean by "Caur/ellipitical low ripple at 4KHz"

[infinia]

Originally posted by nandax

what do u mean by "Caur/ellipitical low ripple at 4KHz"

Hi
Your plan of using a lowpass filter before the audio amp is sound thinking.

see http://en.wikipedia.org/wiki/Elliptic_filter


I have used tables of normalized values for lumped element values then scaled for frequency and input/output impedance. Checked on Pspice. There maybe software that now does this.
Or there are probably designs existing on the internet, as this not a unique problem.


Specifications for a 4KHz low pass Cauer filter
1) Unequeal ripple, i.e. passband ripple = 0.25 dB and high ripple stop band.
2) Low impedance source and high impedance load. (You have to understand the driving impedance of the phone) example... add a series resistor to scale, use around Rs = 600 to 800 ohms total. The load s/b at least >20*Rs maybe higher for more rejection)
3) Decide on stop band rejection i.e use 20 dB for 3 pole and 40 to 50 dB for 5 pole.
4) 3 pole is realized with 1 inductor/ 3 caps and 5 pole " " 2 inductors/5 caps.

Tables can be found in this book.
Zverev, A. I. Handbook of Filter Synthesis. New York: John Wiley &
Sons, 1967

[sreten]

Hi,

If you are planning to amplify a voice call on an mobile handset
such that you only use the microphone it will not work at all.

You will run into intractable problems regarding echo cancellation
and the like which is internally optimised in the phone for its own
acoustic characteristics. If the phone has a handsfree mode this
possibly could work if the internal signal processing is flexible
enough, if it is not you will still get problems.

It would be a very poor phone design that has any noise issues
above 4KHz on a normal bandwidth call. The assumption is that
the noise is below 4KHz and that baby and bathwater applies.

FWIW GSM coding reduces a landline bitrate call of 64Kbits/second
(8 bits at 8kHz, the 8 bits companded to 12 by A or u law coding)
to between 4.75Kbits/s and 13kbits/s, the latter is GSM full rate.
AMR (adaptive Multirate) changes modes with radio conditions.
(resolution / output is nominally 13 bits linear at 8KHz)

Other than the signal to the earpiece being pre EQ'd, i.e. not flat
and therefore amenable to EQ correction, additional filtering will
not improve matters, if it did, simply they would already do it.

/sreten.

(Telecoms Technical Consultant)

[Pano]

Originally posted by sreten
(Telecoms Technical Consultant)

Ah ha! So that's what you do. =)

It will be interesting to see what happens in this project.

[sreten]

Hi,

I should also mention that if A is talking to B, any problems
with A's phone will largely be perceived by B and not A.

That is generally A does not know B's perception of the speech
quality from A, and vice versa, B does not know A's perception.

/sreten.