Prev: Problems with Zadoff-Chu auto correlation
Next: Smartbook for DSP developer ?
From: Green Xenon on 16 Dec 2009 20:49
Hi:

What is the maximum amount of bits-per-symbol of FSK possible using a
telephone system [including the phone lines and any devices from start to
finish of the phone's signal chain]?


Thanks,

Green Xenon
From: glen herrmannsfeldt on 16 Dec 2009 20:57
Green Xenon <green_xenon1(a)yahoo.com> wrote:

> What is the maximum amount of bits-per-symbol of FSK possible using a
> telephone system [including the phone lines and any devices from start to
> finish of the phone's signal chain]?

All the FSK I know of are one bit/symbol. That may not be a
requirement, but it seems that QAM, even at two bits/symbol,
(The old 212, for example) is easier than trying to get more
from FSK.

Stories are that 103 (FSK) can be used up to 600 baud, but
I only ever knew it up to 300. Unlike most later standards,
such as 212 or V.29, 103 is not synchronous. The modem does
not care about the bit rate, but transmits and receives a
signal that is zero or one. In comp.dsp terms, it is quantized,
but continuous time. In that sense, FSK does not send symbols.

-- glen
From: Green Xenon on 16 Dec 2009 21:21
>Green Xenon <green_xenon1(a)yahoo.com> wrote:
>
>> What is the maximum amount of bits-per-symbol of FSK possible using a
>> telephone system [including the phone lines and any devices from start
to
>> finish of the phone's signal chain]?
>
>All the FSK I know of are one bit/symbol. That may not be a
>requirement, but it seems that QAM, even at two bits/symbol,
>(The old 212, for example) is easier than trying to get more
>from FSK.
>
>Stories are that 103 (FSK) can be used up to 600 baud, but
>I only ever knew it up to 300. Unlike most later standards,
>such as 212 or V.29, 103 is not synchronous. The modem does
>not care about the bit rate, but transmits and receives a
>signal that is zero or one. In comp.dsp terms, it is quantized,
>but continuous time. In that sense, FSK does not send symbols.
>
>-- glen
>


Isn't MFSK [Multiple Frequency Shift Keying] a form of FSK that uses more
than 1-bit-per-symbol?
From: Steve Pope on 16 Dec 2009 22:54
Green Xenon <green_xenon1(a)yahoo.com> wrote:

> [glen writes]

>>All the FSK I know of are one bit/symbol.

>Isn't MFSK [Multiple Frequency Shift Keying] a form of FSK that uses more
>than 1-bit-per-symbol?

Does the M stand for Multiple or does it mean Minimal?

You can definitely have 4-FSK, 8-FSK, etc but you are somewhat
driven to using a minimally-shifted orthogonal set when using these
higher order FSK's, otherwise you have a pretty high modulation
index and large sidelobes. So it is usually MSK. Viewed at baseband
n-MSK can be thought of as n/2-FSK concatenated with 2-ASK. (I have no idea
if it's common to look at it that way.)

Steve
From: steveu on 17 Dec 2009 01:12
>On Wed, 16 Dec 2009 19:49:39 -0600, Green Xenon wrote:
>
>> Hi:
>>
>> What is the maximum amount of bits-per-symbol of FSK possible using a
>> telephone system [including the phone lines and any devices from start
>> to finish of the phone's signal chain]?
>
>Infinite, but you have to be patient.

Since telephone lines are heavily quantised by the PCM network, that's
only going to be true for very funky definitions of "symbol" or "FSK"

>Go away and don't come back until you can ask a question that makes
sense.
>
>--
>www.wescottdesign.com
>

Regards,
Steve

 |  Next  |  Last
Pages: 1 2 3 4 5 6 7
Prev: Problems with Zadoff-Chu auto correlation
Next: Smartbook for DSP developer ?