On Sat, 26 Aug 2006 21:30:47 -0500, tim gorman
wrote:

Andrea Baldoni wrote:

....................................


The AGC line is derived from a fixed voltage using a 9V zener, then the RF
GAIN pot permit to divide this voltage from 100% to ground and feed it
(trough a resistor) to the first RF amplifier of the HF receiver (a DG
FET) as well as the first RF amplifier of the 2m converter, and the same
for 6m converter. It is also fed to the last but one CA3053. Other
amplifiers are fixed gain I suppose.
Everything in the receiver needs to reduce gain, lower this voltage by
more or less shorting it to ground.
For instance, the standby button shorts it to ground, silencing the
receiver completely. The RF level at the last IF instead reduce it by
means of common emitter transistor: the AGC voltage from zener at the
collector and the rectified and filtered IF at the base.
When you disable AGC, you disconnect the collector of this transistor,
thus the signal is let alone to the level adjusted with RF gain pot
(normally at maximum, so it is 9V).

Ciao,
AB

Ok, have you checked the Dual Gate FET to insure that the bias supplied by
the RF gain control puts the device at maximum gain when the AGC is off?

Dual Gate FET's have transducer gain curves that are peaked curves.
Depending upon the Gate 2 voltage, the transducer gain can actually go down
as the Gate1 to Source voltage goes up.

I would still be interested in knowing *exactly* what the AGC voltage on the
gate of the DGFET is for 1) AGC on, no signal, RF gain wide open, and 2)
AGC off, no signal, RF gain wide open. It would also be interesting to know
what the Gate 1 bias voltage is for each state as well.

Tim,

Your hitting the specifics of what I refered to earlier. I'll repeat
it for emphasis.

What I have seen in some cases is where the no
signal resting point for gain control bias voltage is not correct and
the gain can go up a bit before going down. Often seen on oder
recievers where the large part of the radio is discrete devices
and the various setpoints have drifted from age or componenet
changes.

I have seen this on older radios where parts have been replaced or
the original parts used were at opposing ends of the allowable
tolerence. Occasionally a part like a zener doide can drift form
heating. The end result is the full gain voltage can be off or full
gain for a single stage can be off (high or low from optimum).
In one case it was a mechanical switch (agc/manual) causing
difficulty (leakage path). Other suspect components seen in
Japanese built radios are those commonly used ceramic disk
caps for bypasses, they can and do go leaky(high resistance),
or short and I've even seen microphonic. I have a reciever I
repaired where the DGfet developed a substrate to gate2 short
which casued all manner of unusual problems.

The worst case by far was one that the agc bias point had drifted
a bit high. When on manual agc the RX was hot. When agc was
enabled the RX sensitvity would drop noticeably. The problem
was the higher agc bias point had the IF and RF running harder and
producing more noise and when agc was turned on it would see the
noise and pull the agc voltage. That sounds ok save for the front
end was more agc sensitivve than the overbiased IF and the front
end would loose gain faster (it was 2 jfets cascode) rendering the
reciever less sensitive. The fix was repairing the internal voltage
regulator that fed 9V to most of the circuits (it was running at 11v
due to open zener).

Just a few examples of what can occur. I havent even gone into the
golden screwdriver problems when pots are tweeked for "more".


I'll bet you'll find an interesting interaction between the bias voltages
and the actual stage gain as the controls are manipulated.

I'd be inclined to agree.

Allison

Andrea Baldoni wrote:
Roy Lewallen wrote:

: The bottom line is that I'd be hesitant to trust just about any number
: for a "worst case" maximum signal strength. Be sure to test any proposed
: design on 40 meters for a while from your location in Europe.

Uh. Very interesting, Roy.
Even a receiver with AGC has his own limits and probably what you experienced
would have surely overload most commercial ones...
Some numbers must be fixed, even if very high ones. So, how one could
proceed?

If you really want to be rigorous about it, you could set up some kind
of logging system, perhaps with an A/D converter and computer connected
to a reference antenna and simple detector, to measure and log signal
strengths over a long period of time. The tough part would probably be
deciding what kind of filter to precede it with; maybe something typical
of what you expect to use in a real receiver. Then you could do a
statistical analysis on the logged signal strengths. Whether or not
that's worth while would be up to you -- it would at least certainly
make an interesting article. Or, you could build something and put a
coarse step attenuator at the front end, noting how much attenuation you
have to apply when operating in order to keep the spurs down.

Roy Lewallen, W7EL

wrote:
. . .
It's relevence is I've seen this before and understood it's origin
and also elsehere. The other aspect is that if a commonly
accepted part is not fully understood and can lead to undesired
effects then, why not others. . .

Very nearly 30 years ago, I was looking into "feed forward" circuits, a
technique developed by someone at Tektronix for ultra-low distortion
amplification. It turns out that the topology of the MC1350 is similar
to what's needed, and a feed forward amplifier can be made from one plus
just a few external components. But even by then, I'd learned that it's
risky to use components for other than their intended purpose. So I
collected 8 or 10 samples from various vendors (the part was widely sold
then), and opened them up. Those in cans were easy, using a little can
opener that worked like a tubing cutter. Some of the plastic DIP ones
were more difficult, but one of the labs at Tek was able to dissolve the
plastic while leaving the chip intact. Then I examined them carefully
with an inspection microscope. Here's what I found:

1. There were at least three very different designs. The chip size of
the largest was several times that of the smallest.
2. Some designs were inherently better balanced than others. Some had
resistive "cross unders" where traces cross, which weren't the same on
both sides of the circuit.

Based on this, I decided it was too risky to make a design based on that
part number, since a vendor could change chip suppliers or designs
without notice.

Interestingly, about six months later, I got a call from the component
engineering group asking if I still had the chips. It seems that one or
more of the vendors supplying that part (which was used for other
applications at Tek) had changed their design, causing failure of some
products and the shutting down of their production lines. Tek was big
enough that vendors were often required to give advance notice before
such changes, but they hadn't given any notice in this case.

I'm bringing this up because I'm hearing the MC1350 being spoken of as
though all are the same. It wouldn't surprise me if, after all these
years, they're now all being made with one design from one foundry. But
those ones in your junk box might be way more different than you think.

This is almost certainly true of just about any IC.

Roy Lewallen, W7EL

On Sun, 27 Aug 2006 13:56:50 -0700, Roy Lewallen
wrote:

wrote:
. . .
It's relevence is I've seen this before and understood it's origin
and also elsehere. The other aspect is that if a commonly
accepted part is not fully understood and can lead to undesired
effects then, why not others. . .

Very nearly 30 years ago, I was looking into "feed forward" circuits, a
technique developed by someone at Tektronix for ultra-low distortion
amplification. It turns out that the topology of the MC1350 is similar
to what's needed, and a feed forward amplifier can be made from one plus
just a few external components. But even by then, I'd learned that it's
risky to use components for other than their intended purpose. So I
collected 8 or 10 samples from various vendors (the part was widely sold
then), and opened them up. Those in cans were easy, using a little can
opener that worked like a tubing cutter. Some of the plastic DIP ones
were more difficult, but one of the labs at Tek was able to dissolve the
plastic while leaving the chip intact. Then I examined them carefully
with an inspection microscope. Here's what I found:

1. There were at least three very different designs. The chip size of
the largest was several times that of the smallest.
2. Some designs were inherently better balanced than others. Some had
resistive "cross unders" where traces cross, which weren't the same on
both sides of the circuit.

Based on this, I decided it was too risky to make a design based on that
part number, since a vendor could change chip suppliers or designs
without notice.

Interestingly, about six months later, I got a call from the component
engineering group asking if I still had the chips. It seems that one or
more of the vendors supplying that part (which was used for other
applications at Tek) had changed their design, causing failure of some
products and the shutting down of their production lines. Tek was big
enough that vendors were often required to give advance notice before
such changes, but they hadn't given any notice in this case.

I'm bringing this up because I'm hearing the MC1350 being spoken of as
though all are the same. It wouldn't surprise me if, after all these
years, they're now all being made with one design from one foundry. But
those ones in your junk box might be way more different than you think.

This is almost certainly true of just about any IC.

Roy,

That is my engineering experience as well. At the time I did my
testing I had Motorola, National and Hitachi parts Some fairly current
date codes and a few from early 80s and and while the general
behavour was similar I noted differences in gain, overall noise
and DC balance as well. The noise increase was enough to
be noteable in a particular case but on analysis understandable
and to be expected.

Then again I date back to when the Fairchild UA703 was a
breakthrough gain block for RF.


Allison
KB!GMX