Help for 10mbit/s FSK transmitter and receiver

[Oscar (X.)]

I have to design a fast 10mbit/s FSK transmitter and receiver. The
frequency coding for mark and space can be chosen freely because the
transmission is between two very close electrodes (electromagnetic
coupling) and not across long distances in air. Actually I have some
design options:
1) ideally, a totally integrated solution (single chip which turns ones
and zeroes into the corresponding frequencies, and vice versa). But
I've been unable to find chips which were fast enough (I found
something like 300kbps at most).
2) I could use a PLL as a demodulator/modulator, but the TLC2933 is the
fastest I found (100Mhz max operating frequency) and I seriously doubt
that the loop bandwidth would be enough. Faster PLLs are usually
intended to be frequency synthesizers (for the implementation of clocks
and local oscillators) and there is no output pin for the VCO control
voltage...
3) probably I should find a PFD and a VCO separately if the integrated
solutions are not good enough
4) as a modulator theoretically I could drive a VCO directly, but it
must be suitable for that (fast frequency stabilization, decent
precision).

Do you already know about good ICs/solutions for this problem?

 

[linnix]

"Oscar (X.) дµÀ£º
"

I have to design a fast 10mbit/s FSK transmitter and receiver. The
frequency coding for mark and space can be chosen freely because the
transmission is between two very close electrodes (electromagnetic
coupling) and not across long distances in air. Actually I have some
design options:

You could use the fastest FPGA (~500 MHz) and wide spectrum
transceivers. Even so, I doubt you can go more than a few mbit/s. The
technology is just not there, or is waiting for you to invent.

 

[Mark]

4) as a modulator theoretically I could drive a VCO directly, but it
must be suitable for that (fast frequency stabilization, decent
precision).

yep a free running VCO should be stable enough compared to 10mbps

your talking about a signal with over 20MHz BW

Mark

 

[Le Chaud Lapin]

I have to design a fast 10mbit/s FSK transmitter and receiver. The
frequency coding for mark and space can be chosen freely because the
transmission is between two very close electrodes (electromagnetic
coupling) and not across long distances in air. Actually I have some
design options:

Does it have to be FSK? Theoretically not impossible, but since you
have choice of band and electromagnetic coupling, you'd be far better
off going with something like a broadband solution.

There are DSL modems that cost $25 OEM. You could cannibalize one of
these designs, as they already support up to 24 mb/s uni-directional
and do the bit-spreading for you. You'd only have to focus on
interface circuitry.

http://en.wikipedia.org/wiki/Digital_Subscriber_Line

-Le Chaud Lapin-

 

[John Larkin]

I have to design a fast 10mbit/s FSK transmitter and receiver. The
frequency coding for mark and space can be chosen freely because the
transmission is between two very close electrodes (electromagnetic
coupling) and not across long distances in air. Actually I have some
design options:
1) ideally, a totally integrated solution (single chip which turns ones
and zeroes into the corresponding frequencies, and vice versa). But
I've been unable to find chips which were fast enough (I found
something like 300kbps at most).
2) I could use a PLL as a demodulator/modulator, but the TLC2933 is the
fastest I found (100Mhz max operating frequency) and I seriously doubt
that the loop bandwidth would be enough. Faster PLLs are usually
intended to be frequency synthesizers (for the implementation of clocks
and local oscillators) and there is no output pin for the VCO control
voltage...
3) probably I should find a PFD and a VCO separately if the integrated
solutions are not good enough
4) as a modulator theoretically I could drive a VCO directly, but it
must be suitable for that (fast frequency stabilization, decent
precision).

Do you already know about good ICs/solutions for this problem?

A stock VCO should do fine for the transmitter. The receiver could be
a MMIC amplifier, a simple phase-shift network and a diode mixer,
followed by a lowpass filter and a comparator. Probably you can do the
whole receiver in a single chip, somewhere, if you shop around.

Run at maybe 100 MHz or so with a pretty wide deviation.

John

 

[Oscar (X.)]

Le Chaud Lapin ha scritto:

Does it have to be FSK? Theoretically not impossible, but since you
have choice of band and electromagnetic coupling, you'd be far better
off going with something like a broadband solution.

Technically I could choose something different from FSK... anyway the
transceiver should be "invisible", i.e. there are 2 systems
communicating via a wire, and they shouldn't notice if this couple of
transceivers are placed at the middle of the wire. With FSK first I
wouldn't have to bother about protocols etc., and then the
communication must be "real time", unlike with many protocols like
wi-fi etc.

You could cannibalize one of
these designs, as they already support up to 24 mb/s uni-directional
and do the bit-spreading for you. You'd only have to focus on
interface circuitry.

I don't think you can open a DSL modem and identify a subsystem which
transmits the data and another separate subsystem which cares about the
protocols... probably there are integrated modules which do these two
things at the same time, anyway I could try and see.

 

[Le Chaud Lapin]

Technically I could choose something different from FSK... anyway the
transceiver should be "invisible", i.e. there are 2 systems
communicating via a wire, and they shouldn't notice if this couple of
transceivers are placed at the middle of the wire. With FSK first I
wouldn't have to bother about protocols etc., and then the
communication must be "real time", unlike with many protocols like
wi-fi etc.

Since you mentioned 10 mb/s, I figured that you were making an Ethernet
link. If that is the case, then you would simply plug the Ethernet
cables into each modem, put phone wire between the two, and be done. I
do not recall if there is a "cut-through" mode for DSL, but there might
be.

I don't think you can open a DSL modem and identify a subsystem which
transmits the data and another separate subsystem which cares about the
protocols... probably there are integrated modules which do these two
things at the same time, anyway I could try and see.

It's not very elegant, but if you were to take the signal just before
the line amplifiers for each modem and up-convert them to an
appropriate frequency, you might just get away with this. The modems
already have state-of-the-art forward-error-correcting schemes in them,
so if you lose a frame or two, it would happen rarely if you have taken
care in doing coupling and shielding.

Also, a circuit based on free-space optics might be a consideration.
After all, if the couplers are that close, then you are almost
guaranteed good performance.


-Le Chaud Lapin-