wideband FM demodulator

That'd be your delay line, used with a mixer to recover the modulated signal. It'll work but I'd be surprised if it's very stable.

For the case you're describing, straight FM isn't all that common. I used an MC13155D for something similar at one point but it's gone the way of the 12AX7 (oh, wait, you can still get those). I wonder if there's a QAM chipset available with similar bandwidth? Something comparable to a cable or DSL modem, but without all the DOCSIS baggage.

-- john, KE5FX

Reply to
John Miles, KE5FX
Loading thread data ...

He's got big FPGAs and even bigger RAMs. Pipeline as many as needed, or just shove everything into a lookup table.

:^)

Tim

--
Seven Transistor Labs, LLC 
Electrical Engineering Consultation and Design 
 Click to see the full signature
Reply to
Tim Williams

Quadrature signals are mixed, and then lowpass of product is the modulating signal. The linearity is a function of ATN(k) being approximately equal to k, so the smaller the relatively deviation the better. May be better to us e 1GHz carrier for 5% deviation.

Reply to
bloggs.fredbloggs.fred

It's not characterized as a delay line. It's apparently intended to provide a 90 degree phase shift that's independent of frequency. Not the basis of an FM discriminator.

We've been playing with pushing maybe DC-100 MHz of analog signal over a telecom-type fiber link. I was thinking of using FM, but that might become a nightmare. Brute force (merely apply money) is an ADC at one end, 8B10B coding, a 6 Gbps fiber link, and a DAC at the far end. Sampling at 300 MHz, 1.5x Nyquist, makes the filtering reasonable.

I can buy pretty good VCOs for the transmit end, but their typical modulation bandwidths are low. Then there is the receiver problem.

Oh well, back to connecting bricks.

--
John Larkin         Highland Technology, Inc 

lunatic fringe electronics
 Click to see the full signature
Reply to
John Larkin

take a look at these

formatting link

The 933/934 will do 100Mhz 12 bit parallel IO. They are limited to 15M cable length though, coax or UTP. Dont know if that's a show stopper.

--
Chisolm 
Republic of Texas
Reply to
Joe Chisolm

Mixing fixed 90 degree sine waves makes zero. The Minicircuits part is designed for minimum phase shift vs frequency.

--
John Larkin         Highland Technology, Inc 
picosecond timing   precision measurement  
 Click to see the full signature
Reply to
John Larkin

Especially in the presence of facet reflections, which cause phase nonlinearities just like multipath distortion in radio. If you use FC-APC (angled physical contact) connectors, you can get the reflection down to ~1% or a bit less, which helps, but if the fibre delay is up near your deviation bandwidth there'll be nonlinearity.

One of our main customers just blew up three lasers that we sent them.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
 Click to see the full signature
Reply to
Phil Hobbs

Yeah, digital either works or it doesn't. The plug-in SFP things do work and are shockingly cheap.

Maybe they did something wrong?

After blowing up the first two, they needed a time-out.

--
John Larkin         Highland Technology, Inc 
picosecond timing   precision measurement  
 Click to see the full signature
Reply to
John Larkin

At frequencies from 500 MHz to 1 GHz the delay is easiest with just a piece of transmission line. A full wave of 1 GHz will be about 8 inches of cable. A quarter wave is short enough to be made of stripline on the PC board.

--

-TV
Reply to
Tauno Voipio

Three!? that takes a special talent. Usually when you fry one you stop and think about what you are doing.

George

Reply to
George Herold

We sent one of them all connected and everything, and they disconnected it and blew it up. Or maybe they just put the FFCs in backwards, in which case it would probably have survived.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
 Click to see the full signature
Reply to
Phil Hobbs

I did an analysis of quadrature detectors and uploaded the results:

formatting link
usp=sharing

Here is the readme file:

Out of curiosity, I decided to investigate the delay discriminator. Here are the results.

4D620E45 ASC DBM Double Balanced Mixer Max Output

The first task is to analyze the Double Balanced Mixer (DBM). Some believe if you drive it harder, you can increase the output. This is true, to an extent. Soon, you reach the point where it starts to saturate.

The reason is the DBM acts like two diodes back-to-back. If you drive it harder, you are just pushing against the forward bias curve of a diode.

4D620F40 ASC DBM Delay Discriminator

Here is the delay discriminator using a DBM showing the classic S-curve of a phase detector. The signal swings between 400 and 600 kHz.

The reason the curve flattens is because the DBM saturates and cannot increase the output.

Note the large filter capacitor, C1, needed to smooth the output ripple. This shows the DBM may be useful in a lab setting in a low bandwidth application, but not in an instrument application requiring wide bandwidth.

4D622643 ASC XOR Delay Discriminator

An XOR gate can be used as a discriminator, but the sensitivity is much lower. Here the frequency swing is from 300 to 700 kHz, with about 2 V p-p output.

4D627281 ASC MC4044 Delay Discriminator

Phase detector #2 in the MC4044 is also a quadrature detector. Here the frequency swing is 400 to 600 kHz, with an output of about 500 mV p-p.

Conclusion

The delay discriminator can be useful in a lab setting, but the filtering required to reduce the output ripple restricts the application to low bandwidth applications.

Reply to
Steve Wilson

Heh, good point. If you want a voltage proportional to frequency deviation you probably shouldn't start with an allpass network.

-- john, KE5FX

Reply to
John Miles, KE5FX

Reply to
bitrex

Reply to
bitrex

te:

ng

se

XOR

ing

ating signal. The linearity is a function of ATN(k) being approximately equ al to k, so the smaller the relatively deviation the better. May be better to use 1GHz carrier for 5% deviation.

Buried stripline in the PC board is non-dispersive and thus preferred over microstrip on the surface of the PC board.

Semi-rigid coaxial cable is nice, but not cheap, and the connectors to get the signal out of the PC board and back in again are even more expensive.

The electric constant of the insulator - and thus the delay through the del ay line - is temperature dependent and thus variable.

Semi-rigid coax uses Teflon/PTFE as it's dielectric, where the effect is sm all and predictable.

formatting link

I've used alumina-loaded Teflon as a printed circuit board substrate

formatting link
O4350BData-Sheet.pdf

It looks as if you could buy stuff that would go the other way.

--
Bill Sloman, Sydney
Reply to
bill.sloman

No special talent. Just put a senior manager in the room and everything blows up. In physics it is called the Pauli Effect.

formatting link

Apparently his mere presence in the same town (at the railway station in Gottingen) was sufficient.

In practice, the pressure to make something happen does seem to be the mechanism.

--
Bill Sloman, Sydney
Reply to
bill.sloman

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.