9fa56f6e
From: apogosso@tpg.com.au
"Patrick Turner" wrote in message
news:f9f2af98-6417-4abb-88c8-c61cb6b6ed03@v20g2000prl.googlegroups.com...
> I'm having fun mucking around with 6 x PN100 garden variety
> transistors used to make a discrete Gilbert Cell, like the bjts inside
> an MC1496.
> Instead of a rail of +12Vdc and two CCS bjts for the "bottom pair"
> LTP, I have a +41Vdc rail and -30Vdc rail because the bjts are quite
> happily biased with a max Vce = 15Vdc approx.
>
> Anyway, I'm trying to produce an RF AM signal generator better than my
> present one which uses a couple of triodes for an oscilator and CF
> buffer, followed by a pentode RF amp which is grid modulated and with
> Rk bypassed for RF only. The oscilator and pentode tank coils are
> tuned with a two gang tuning cap from an old AM radio. Tracking is
> good enough. At 90% audio modulation the envelope THD is about 5% 2H
> and I want 97% mod with 1% max Dn, and without using loop FB from a
> detector in the unit, or resorting to PP operation.
>
> So I thought a Gilbert Cell might work better but with balanced or
> unbalanced AF and RF inputs all I get is a DSB signal, ie, the two
> side bands and no RF carrier.
>
> I have not got as far as producing an AM detector using the Gilbert
> cell yet.
>
> Anyone else have any experience worth relating to the group?
>
> Patrick Turner.
Alex:
The most linear modulator is a CMOS changeover (SPDT) switch, e.g., DG445,
DG442 or others with lower resistance. Get the switch fly with the carrier
frequency from some voltage to ground. Then the output will be perfectly
chopped between this voltage and zero. Thus the voltage becomes the envelope
of the output modulated signal. If this voltage is audio voltage -- here you
have a modulator. If the voltage has no DC, the output will be DSB, if you
offset it by DC component, you will have AM. The voltage (signal) source
shall have low impedance at RF. Output needs to go through LPF since it is
basically rectangular and "dirty" for "broadcasting. Note that LPF shall
start with an inductor (not capacitor) for the same reason -- the signal is
rectangular.
You might be put off by the digital flavour of the thing. If this is the
case, here is another more analogue solution. Use not a double-balanced
modulator (which you call Gilbert cell), but simply a differential pair (2
bjts emitter coupled). Apply about 100...150mV of carrier between their
bases. This is easily done by putting one or two turns of tickle winding on
top of your tuce oscullator (which I guess has about 10...20V on the grid).
Now between the collectors the modulated signal will be almost exactly
proportional to the current in the long tail. E.g., if you make the current
swing from 1 to 20mA you will have 95% AM. Note that impedance of the long
tail shall be large, and capacitance as small as possible to maintain
linearity up to very small current. Thus it is the best to have a simple
resistor, say 1K directly going to the emitters. Another resistor in series
with the former will probably go to negative supply rail. Apply audio to the
junction of these two resistors.
Pick up the AM from the collectors via centre tapped winding. RF load
resistance should be low enough to avoid clipping. Do not forget
anti-parasitic 20...50R stoppers in the bases or collectors.
I would recommend to build cascodes on top of such modulator. This will
nullify base-collector capacitance and eliminate interplay between the
oscillator and antenna tanks. Thus your AM will be perfect -- without mix of
parasitic FM.
Regards,
Alex
--- SoupGate-Win32 v1.05
* Origin: you cannot sedate... all the things you hate (1:229/2)
|
