[Premium-Rx] Reaction Instruments 685 AM-FM demodulator

[Peter Gottlieb]

I finished sketching out the demod schematic and had time to fire up the 8753C 
VNA.  Sure enough, the different filter boards in the demod assembly have a 
definite (sort of) linear phase response centered about the 21.4 MHz center 
frequency.  The different filters have a different slope to cover the bandwidth 
of the corresponding IF filter.

To properly demod my new 10 kHz BW I need to make a filter which has a -180 to 
+180 degree phase response over 10 kHz.  That's my next little research project.

***

I am ordering a bunch of J106 J-FETs and will play around to get the circuit to 
work as intended.  These don't look like they need to match.

Peter


On 11/1/2015 3:14 PM, Charles Steinmetz wrote:
> Peter wrote:
>
>> [FM] Detection is done with a MC1496 balanced modulator/demodulator chip.  
>> One input is fed raw IF and the second is after going through one of the four 
>> selected filter sets.    *   *   *    The 1496 detects the phase difference 
>> and produces a voltage proportional to the deviation of frequency from 
>> center.    *   *   *    Has anyone else ever seen such a scheme?
>
> You used to see this circuit in IC cookbooks.  Balanced modulators can be used 
> as phase detectors, and of course phase and frequency are kissing cousins 
> (frequency being the time derivative of phase, and phase the time integral of 
> frequency).  That said, using phase detectors to demodulate FM is an 
> inherently noisy way to go about it and the 1496 is a particularly noisy 
> balanced modulator, so the circuit was never popular.
>
>> one set consists of 2N4447 FETs and the second set are 2N2222A bipolars.  You 
>> cannot parallel the inputs to these two as the bipolar B-E junction will 
>> saturate at about 0.65 volts which is well below the 3-4 volt threshold for 
>> the 2N4447, so the 2N4447's will never turn on.
>
> The 2N4447s are depletion-mode JFETs, so they are fully on when the 
> gate-source voltage is zero.  You need to pull the gates 2-10v below the 
> source to turn them fully off.  Note that you can only reverse-bias the 
> 2N2222A's B-E junction by 5 or 6v before you degrade or damage the device, so 
> it's possible you wouldn't be able to turn the FETs *off* without breaking the 
> bipolars.  (You also don't want to forward-bias the FET gate junction, but 
> that is a function issue, not a damage issue.)
>
>> In any case I'm going to try and find some 2N4447's and restore it to a 
>> design which will work as intended.  Again, comments?
>
> I believe only Siliconix and Crystalonics made 2N4447s.  I don't believe 
> Siliconix still does.  My most recent Crystalonics catalog is from 2013, and 
> they still listed them at that time.  Expect them to be expensive.  2N4447s 
> are similar to 2N5432s (Crystalonics listed those, too, and I believe InterFET 
> makes them, as well).  Alternatively, the J106 should work (available for less 
> than a dollar from Fairchild and others), although you may need to replace 
> both sets if they need to match.
>
> Best regards,
>
> Charles
>
>
> ______________________________________________________________
> Premium-Rx mailing list
> Home: http://mailman.qth.net/mailman/listinfo/premium-rx
> Help Page: http://mailman.qth.net/mmfaq.htm
> Post: mailto:Premium-Rx at mailman.qth.net
> Help Contact eMail:  paul at 8zo.com
> Home Page:  http://www.premium-rx.org/
>