snip for brevity.
John Byrns
Surf my web pages at, *http://fmamradios.com/
You grasped the idea perfectly well John. This frequency compensation is
useful to prevent a lousy OPT and 6AQ5 overloading in a lousy boring audio
amp found in those boring AA5 style radios.
I agree with everything Alex says below except for a few things.
In most old radios with DC flow across the volume control pot track
the adjustment of the volume is often very noisy after the pot has
aged a few years.
So instead of the conventional arrangements put forward by RDH and
most others to avoid parts costs I will have the last IFT coil biased
at say +50V at one end, and the live end goes to a triode grid of 1/2
12AU7 CF buffer to remove any loading effect of diode detection on the
last IFT LC. Then I use Ge diode feeding RC circuit, and this can
directly feed second 1/2 12AU7 CF buffer and then usual CR coupling to
any a volume control and while employing time passive poles to give
say -3dB at 30Hz before any power amp which has NFB. I often add in
another 12AU7 gain stage for tone control to boost/cut treble; bass in
AM is usually OK.
To slightly widen AF response the Q of all IFTs may be reduced by
strapping 100k across each coil. It doesn't work in all sets, but may
be tried. There is a simple method explained in RDH4 to add a few
turns of fine wire around the primary of IFT1 and switch it to being
in series with the sec of IFT1 which will give a doble peak to IFT
response which effectively increases IF channel bandwidth from a
typical 4kHz to 8 kHz. Quad used it.
I've used it - finally there is the the treble which most sets refuse
to give me.
I like to use paralleled 12AX7 as V1 and EL34 in triode as V2 for the
audio amp and with 12dB global NFB.
This works far better than anything with 6V6 or EL84 etc.
The Ge diode in the detector is biased on with a low current in the R.
Its works best with high level signals and up to 100% mod, with 10Vrms
available if you want it.
But anyone could use the normal arrangement of IFT plus tube
rectifiers found in IF amp tubes like 6N8 and thus get say -2Vdc
generated by a carrier with 2pk volts of signal. at 100% mod the Vpk-
pk of audio = 4V, so max audio = 1.4V, no?
The audio signal with negative vdc and some 455kHz ripple could be
direct coupled to a CF triode buffer stage but you'd need a -150Vdc
rail for the cathode resistor, while the anode would go to a +100V
supply. The CF buffer and negative rail was viewed as a waste of money
in 1950, so it was never done. But there's no reason why it can't be
done NOW. And the screen voltage applied to mixer and IF amp is best
regulated at 100Vdc, so that's a good point of supply with CF anodes .
I don't much like tubes like the 6AV6 with grounded cathodes and 10M
grid leak biasing.
The very low F signals generated in AM sets don't seem to cause any
problems in anything I have made.
Patrick Turner.
Below is the sequence of mods you do to improve the radio.
1. In an AM detector you remove a cap coupling a detector load (470K||100pF)
to the volume control (1M pot). In fact you are making the volume control
the detector load. This is for the detector to be able to handle nearly 100%
modulation. But now you det DC on the input of the audio amp.
2. You decouple this DC from the pot wiper by a 0.02...0.05uF, but you need
10M input impedance (grid leak) so that the AM detector is not loaded and
100% modulation is still handled.
3. On most stations your AM detector delivers 1...4V of audio, while
sensitivity of a typical boring feedbackless two-stage audio amp is
100...200mV. The radio always works with volume control close to minimum.
The speaker is boomy (not damped), distortion is high. You want to trade the
sensitivity excess for distortion and apply NFB, reducing sensitivity to
0.5...1V. You are enjoying tight crispy sound, but THAT IS WHERE PROBLEMS
BEGIN.
4. Now the bandwidth of your amp is in theory goes to 5...10Hz, but because
of the lousy OPT, these low frequencies do not reach your ears, but only
overload the 6AQ5, since the error signal becomes too large at low
frequencies.
5. Note that neither the american aggressive LF cutoff by reducing
interstage cap to 2000pF nor the Patrick's shelving does not work, since
still yjr NFB is pushing to maintaim unmaintainable LF output, the error
signal is large, the first stage is overloading, and (IMPORTANT!) since the
shelving is virtually a differentiator, it accentuates all the harmonics
generated in the overloaded 1-st stage and feeds them to the 6AQ5. Instead
of overloading the 6AQ5 we have emphasized distortion from the first stage.
6. That is where you need this RRC compensation in the feedback to roll the
low-frequency response of the amp in line with the capability of the lousy
OPT. Typically from 80....100Hz down. Applying the RRC divider provides
undistorted output down to 10...25Hz,
because the error signal remains under control.
7. However at the frequencies lower than 10...25Hz, where the NFB RRC
network levels out at 100% beta, the error signal continues to rise. This
VLF content caused by fadings, AGC knocked by atmospheric interference, etc.
is smaller, and some might stop here, but a purist might like to prevent
overloading even at VLF.
8. To do the above one needs to degenerate the first stage gain, rather than
to use a Partick's attenuator after it. To degenerate the 1-st stage gain
you need to place say 22K resistor in series with the 1-st stage cathode,
and shunt this resistor by a 0.22...0.47uF capacitor. Thus for medium
frequencies the 1-st stage will be working as usual, but ay low frequencies
the transconductance will be degenerating with the perfect linearity of the
first stage maintained. It is the same shelving, but implemented in a wise
linear mode. (Of course the grid leak can not be taken to GND any more, it
should be connected to the cathode or a tap in this 22K resistor.) At
infinitely low frequencies the gain of the first stage is to be degenerated
to 10...20 so that the 1-st stage output is just below the negative bias of
6AQ5 and the later is never overloaded.
Usually I apply mods up to #7 and sometimes #8 too if I can find enough free
solder lugs around the 1-st audio stage tube. Of course, all of the above
implies the cathode of the 1-st audio tube has to be free of duo-diode
functionality. That forces to use separate diodes for AGC and the AM
detector. In some cases instead of 6Q7, 6B8, 6AV6, etc. I might use a 6SL7
tube with one triode as the 1-st audio and the other triode as a diode for
the AM detector, and a silicon diode as an AGC detector.
Regards,
Alex- Hide quoted text -
- Show quoted text -