About AF derived AGC
Hans Summers
I read a little about AGC and that audio-derived AGC is better than
IF-derived in so far as the gain compensation is precisely according to what
is actually heard, whereas with IF-derived signals that get through the IF
passband but not the audio filtering can cause unwanted gain reduction of
the heard signal. The disadvantage with AF-derived AGC being that low
frequencies take a relatively long time to reach their peak and during this
time the early RF stages of the receiver may already have been overloaded.
My receiver (http://www.hanssummers.com/radio/polyphase/index.htm) is direct
conversion using a Tayloe switching detector and polyphase network for
opposite sideband suppression. All the selectivity and all the gain are in
the AF section, there is no RF amplifier and no IF. Therefore I HAVE to use
audio-derived AGC. Since I have no earlier stages for strong signals to
overload before the peak is reached, is there going to be a problem?
Alternatively, I had the idea that the post-audio-filter audio could be
mixed back up to say a 50KHz "IF", band-pass filtered and rectified to
provide the AGC - a simple switching mixer could be used and double-sideband
wouldn't matter as all that is needed is the amplitude. 50KHz would be low
enough to use ordinary opamps for a simple 50KHz band pass filter and
rectifier. Does anyone think this would 1) work and 2) avoid the problems
with AF-derived AGC? If there's even going to be a problem with a direct
conversion receiver?
73 de Hans G0UPL
------------------
Hans Summers
http://www.HansSummers.Com
Bozidar Pasaric
Dear Hans,
I also have a home-made Tayloe receiver and it works fine. Now, about
the AGC. The other day I found in the Italian magazine "CQ elettronica"
for March 2002 an interesting solution which I tried and I am quite
satisfied with it. It first amplifies the AF signal with a TL072, then
rectifies it with two diodes (voltage doubler) and leads it to an
emitter follower with a BC237 (which has an S-meter in the emmiter, and
at the same time its feedback positive volatage is routed to the base
of another BC237 transistor whose emitter is grounded. The signal
shorts the collector-emitter channel when necessary. The collector is
connected to the antenna through a 10000 pF capacitor, so in fact the
antenna signal is weakened. (When working, the emitter-collector
resistance of that transistor changes between 10 kohms and 20 ohms.) It
can be added to any receiver without any changes in the receiver.
Unfortunatelly I do not have a scanner to send you the scheme, but if
you wish I can send you the article by ordinary post. It has a PCB and a
photo. If you are interested, send me an e-mail.
Best wishes, Bozidar, 9A2HL
Richard Hosking
end mixers. It had an 8 pole network which gave good SSB suppression out to
8 KHz or so. It used an audio AGC system which seemed to work quite well -
in fact Rod VK6KRG still has some boards somewhere. As a bonus the AGC volts
gives quite a good approximation to a log detector for an S meter.
Due to the poor performance of the 602's with strong signals this was a
problem on HF. I even used the polyphase network at RF to generate the
quadrature LO signals - it was quite accuarte to 15 MHz or so but is rather
lossy. However you didnt have to have the LO at 4 times the wanted freq.
Richard
"Hans Summers" <Hans.S...@Tudor.Com> wrote in message
news:amn240$71o5b$1...@ID-61331.news.dfncis.de...
Wes Stewart
<Hans.S...@Tudor.Com> wrote:
|
|I read a little about AGC and that audio-derived AGC is better than
|IF-derived in so far as the gain compensation is precisely according to what
|is actually heard, whereas with IF-derived signals that get through the IF
|passband but not the audio filtering can cause unwanted gain reduction of
|the heard signal.
I believe it is exactly the opposite. Strong signals that do not get
through AF filtering and thus do not generate audio-derived AGC
voltage can overload preceeding stages. AGC should be used to
maintain appropriate gain distribution and SNR and not necessarily to
maintain constant audio output.
|The disadvantage with AF-derived AGC being that low
|frequencies take a relatively long time to reach their peak and during this
|time the early RF stages of the receiver may already have been overloaded.
Exactly.
ye...@removecet.com
<Hans.S...@Tudor.Com> wrote:
>
>I read a little about AGC and that audio-derived AGC is better than
>IF-derived in so far as the gain compensation is precisely according to what
>is actually heard, whereas with IF-derived signals that get through the IF
>passband but not the audio filtering can cause unwanted gain reduction of
>the heard signal. The disadvantage with AF-derived AGC being that low
>frequencies take a relatively long time to reach their peak and during this
>time the early RF stages of the receiver may already have been overloaded.
>
>My receiver (http://www.hanssummers.com/radio/polyphase/index.htm) is direct
>conversion using a Tayloe switching detector and polyphase network for
>opposite sideband suppression. All the selectivity and all the gain are in
>the AF section, there is no RF amplifier and no IF. Therefore I HAVE to use
>audio-derived AGC. Since I have no earlier stages for strong signals to
>overload before the peak is reached, is there going to be a problem?
I'm not sure what you are asking. Are you debating between pre-filter
and post-filter AGC? I would think that you would want to feed your
AGC with whatever you are hearing (post-filter) because that's why you
have AGC in the first place. If the previous stages overload, add an
attenuator or use a smaller antenna, don't try to control the gain one
RF cycle at a time. You could also add an FET on the RF input as an
AGC-controlled attenuator, and in combination with the post-audio AGC
should solve the problem before it's a problem.
But the best idea would be to just build it and find out what really
happens -- no point in solving problems that may not exist.
>Alternatively, I had the idea that the post-audio-filter audio could be
>mixed back up to say a 50KHz "IF", band-pass filtered and rectified to
>provide the AGC - a simple switching mixer could be used and double-sideband
>wouldn't matter as all that is needed is the amplitude. 50KHz would be low
>enough to use ordinary opamps for a simple 50KHz band pass filter and
>rectifier. Does anyone think this would 1) work and 2) avoid the problems
>with AF-derived AGC? If there's even going to be a problem with a direct
>conversion receiver?
It would probably work. I may be missing something here, but it sure
seems like a mighty extensive kludge when you can get the same results
with a much simpler design: If you are trying to prevent RF overload,
add an RF limiter on the front end.
Bob G. Mahrenholz
signal will represent the amplitude of the carrier (which is constant) and not
that
of the modulation.
K4QQK
Hans Summers
"Wes Stewart" <n7...@nowhere.com> wrote in message
news:3d8fae1c...@news.triconet.org...
> On Mon, 23 Sep 2002 13:40:32 +0100, "Hans Summers"
> <Hans.S...@Tudor.Com> wrote:
>
> |
> |I read a little about AGC and that audio-derived AGC is better than
> |IF-derived in so far as the gain compensation is precisely according to
what
> |is actually heard, whereas with IF-derived signals that get through the
IF
> |passband but not the audio filtering can cause unwanted gain reduction of
> |the heard signal.
>
> I believe it is exactly the opposite. Strong signals that do not get
> through AF filtering and thus do not generate audio-derived AGC
> voltage can overload preceeding stages. AGC should be used to
> maintain appropriate gain distribution and SNR and not necessarily to
> maintain constant audio output.
>
overloading the mixer etc since it is a FST3253 switch. My main reason for
AGC is just to maintain constant audio output.
Hans Summers
"Richard Hosking" <zapri...@iinet.net.au(remove the zap)> wrote in
message news:3d8f3517$0$41...@echo-01.iinet.net.au...
> I have built a polyphase DC RX some years ago which used NE602s in the
front
> end mixers. It had an 8 pole network which gave good SSB suppression out
to
> 8 KHz or so. It used an audio AGC system which seemed to work quite well -
> in fact Rod VK6KRG still has some boards somewhere. As a bonus the AGC
volts
> gives quite a good approximation to a log detector for an S meter.
> Due to the poor performance of the 602's with strong signals this was a
> problem on HF. I even used the polyphase network at RF to generate the
> quadrature LO signals - it was quite accuarte to 15 MHz or so but is
rather
> lossy. However you didnt have to have the LO at 4 times the wanted freq.
I think the Tayloe mixer is great, it doesn't seem to have any strong signal
problems. I built my 8-column polyphase network according to design
guidelines in a 1995 QEX article. Unlike most networks which either use a
constant resistance value or constant capacitance value, in this case both
vary in such a way that the network can be made lossless. I measured the
loss of the network and it was essentially 0dB within a couple of dB, see my
graph at the bottom of http://www.hanssummers.com/radio/polyphase/index.htm.
It seems doing this overcomes a common objection to polyphase networks i.e.
that they are lossy.
>
> Richard
>
Hans Summers
<ye...@REMOVEcet.com> wrote in message news:3d8fb2cb...@news.cet.com...
I don't think previous stages are going to cause overload in my case as it
is direct conversion with a switching mixer. What I'm interested in is
keeping the perceived audio more constant, in which case it sounds like I
will be fine with audio derived AGC and a gain-controllable amplifier in the
audio path, I have some CA3086 transconductance opamps for this purpose.
>
> But the best idea would be to just build it and find out what really
> happens -- no point in solving problems that may not exist.
>
> >Alternatively, I had the idea that the post-audio-filter audio could be
> >mixed back up to say a 50KHz "IF", band-pass filtered and rectified to
> >provide the AGC - a simple switching mixer could be used and
double-sideband
> >wouldn't matter as all that is needed is the amplitude. 50KHz would be
low
> >enough to use ordinary opamps for a simple 50KHz band pass filter and
> >rectifier. Does anyone think this would 1) work and 2) avoid the problems
> >with AF-derived AGC? If there's even going to be a problem with a direct
> >conversion receiver?
>
> It would probably work. I may be missing something here, but it sure
> seems like a mighty extensive kludge when you can get the same results
> with a much simpler design: If you are trying to prevent RF overload,
> add an RF limiter on the front end.
>
True, true. Just a crazy idea I guess. I will stick to simple AF-derived
AGC and a CA3086 gain-controlled amplifier.
Hans Summers
"Bob G. Mahrenholz" <bo...@edge.net> wrote in message
news:3D8FDCD6...@edge.net...
> I believe that the AGC voltage you get when you rectify the 50kHz
modulated
> signal will represent the amplitude of the carrier (which is constant) and
not
> that
> of the modulation.
>
> K4QQK
True, though I suppose if one used a balanced mixer the carrier would be
mostly suppressed. You could also subtract the constant voltage representing
the carrier amplitude. But I think it is all unnecessary complication and I
am best off sticking with simple audio AGC.
--
Bob Lewis (AA4PB)
the attact times are not fast enought to prevent a "pop" in the audio
when receiving strong signals.
Richard Hosking
The gain control element was a T attenuator with a darlington pair to gnd as
the control element (ie the vertical part of the T). This was followed by a
standard opamp (TL081) for gain in the loop and a transistor/diodes for the
DC rectifier at the output of this. It has a control range of better than 60
dB
Richard
"Hans Summers" <Hans.S...@Tudor.Com> wrote in message
news:amp9e5$7tv1a$1...@ID-61331.news.dfncis.de...
Wes Stewart
<Hans.S...@Tudor.Com> wrote:
|
|"Wes Stewart" <n7...@nowhere.com> wrote in message
|news:3d8fae1c...@news.triconet.org...
|> On Mon, 23 Sep 2002 13:40:32 +0100, "Hans Summers"
|> <Hans.S...@Tudor.Com> wrote:
|>
|> |
|> |I read a little about AGC and that audio-derived AGC is better than
|> |IF-derived in so far as the gain compensation is precisely according to
|what
|> |is actually heard, whereas with IF-derived signals that get through the
|IF
|> |passband but not the audio filtering can cause unwanted gain reduction of
|> |the heard signal.
|>
|> I believe it is exactly the opposite. Strong signals that do not get
|> through AF filtering and thus do not generate audio-derived AGC
|> voltage can overload preceeding stages. AGC should be used to
|> maintain appropriate gain distribution and SNR and not necessarily to
|> maintain constant audio output.
|>
|In my case all the gain is at audio and I do not think there is much risk of
|overloading the mixer etc since it is a FST3253 switch. My main reason for
|AGC is just to maintain constant audio output.
I understand your situation, however, you are the one that brought up
the comparison between i-f and audio-derived AGC. I was responding to
your comparison, in which I believe that you came to a wrong
conclusion.
In your specific case, audio is the result of downconversion and can
be considered the "i-f." Even here, my comments about appropriate
gain distribution apply. You should have some delay (amplitude, not
time) in the application of gain reduction to allow the SNR to
increase. This means that you cannot maintain a completely flat audio
output. Nor should you want to. Stronger signals should sound
stronger than weak ones.
Clifton T. Sharp Jr.
> I believe that the AGC voltage you get when you rectify the 50kHz modulated
> signal will represent the amplitude of the carrier (which is constant) and not
> that
> of the modulation.
Depends on your time constant, no? I can picture great fun following the
BBC's six beeps on the hour. Any pause in speech/modulation will let the
smoothing capacitor discharge. Kinda like the breathing in some noise-
reduction hi-fi methods.
--
It is better to light just one little candle than to curse the darkness.
If everyone lit just one little candle, the EPA would outlaw candles.
Roy Lewallen
>
> In my case all the gain is at audio and I do not think there is much risk of
> overloading the mixer etc since it is a FST3253 switch. My main reason for
> AGC is just to maintain constant audio output.
That's an important distinction. My experience is that if you try to
control the RF or IF gain with an audio-derived signal, it's just about
impossible to get decent operation. If the AF pickoff point is earlier
than the audio filter, then the gain can be reduced by signals that are
outside the AF filter bandpass. If you pick off after the AF filter, the
delay in the filter makes it impossible to get fast AGC operation, and
loop stability becomes a problem.
The only good solution I've seen is to design the receiver so that all
stages before audio are strong enough to handle whatever signal gets in,
and don't try to control the gain of these stages with the AGC. Then use
an AGC controlled audio amplifier to maintain the audio output level. I
know there are a number of satisfactory circuits around -- probably some
in the form of a single IC -- hopefully someone can point you to one.
Roy Lewallen, W7EL
Hans Summers
> On Tue, 24 Sep 2002 09:48:29 +0100, "Hans Summers"
> <Hans.S...@Tudor.Com> wrote:
>
> |In my case all the gain is at audio and I do not think there is much risk
of
> |overloading the mixer etc since it is a FST3253 switch. My main reason
for
> |AGC is just to maintain constant audio output.
>
> I understand your situation, however, you are the one that brought up
> the comparison between i-f and audio-derived AGC. I was responding to
> your comparison, in which I believe that you came to a wrong
> conclusion.
Sure. I don't know enough about AGC which is why I am asking the question
and trying to learn more. Whilst in my case I have to use some form of audio
derived AGC I am still interested in both methods and the comparisons
between them.
>
> In your specific case, audio is the result of downconversion and can
> be considered the "i-f." Even here, my comments about appropriate
> gain distribution apply. You should have some delay (amplitude, not
> time) in the application of gain reduction to allow the SNR to
> increase. This means that you cannot maintain a completely flat audio
> output. Nor should you want to. Stronger signals should sound
> stronger than weak ones.
I am interested in why you say I shouldn't want to maintain a flat audio
output. If I'm sitting there and there's QSB on the band surely it's nice
for me if the receiver compensates for it as far as possible by keeping the
audio relatively constant?
Hans Summers
"Roy Lewallen" <w7...@eznec.com> wrote in message
news:3D90D826...@eznec.com...
Hopefully this is the situation I'm in - I think the earler stages in the
receiver are strong enough. The Tayloe detector + polyphase seems a very
clean design.
>
> Roy Lewallen, W7EL
>
ye...@removecet.com
<Hans.S...@Tudor.Com> wrote:
<snip>
>I am interested in why you say I shouldn't want to maintain a flat audio
>output. If I'm sitting there and there's QSB on the band surely it's nice
>for me if the receiver compensates for it as far as possible by keeping the
>audio relatively constant?
What you could do is make an AGC circuit with variable attack and
release. Use a faster response for audio, and a slower response (or
switch it out) for CW. This should satisfy all your requirements.
Instead of an AGC circuit, you could probably just plug in a cheap
compressor, like an Alesis or something else along those lines.
Jim Pennell
>
>
> I am interested in why you say I shouldn't want
> to maintain a flat audio output. If I'm sitting
> there and there's QSB on the band surely it's nice
> for me if the receiver compensates for it as far as
> possible by keeping the audio relatively constant?
>
One reason to NOT maintain a flat audio output is what happens to general
band noise...
If you have massive AGC, then when you are not tuned to a station, the
AGC will raise the background noise to the full audio level and you get
constant noisy hissing and crackling in the speakers.
If you apply agc slightly to the really weak signals then it won't raise
the noise so much when you are not recieving a signal.
Generally, most signals you want to hear will be loud enough to be in AGC
range, but really weak ones will be 'fainter' since they may be just barely
above the noise and that actually sounds better to the listener since the
slight differences in how loud the signal is gives a sense of how powerful
the other station is.
Jim Pennell
N6BIU
Wes Stewart
<Hans.S...@Tudor.Com> wrote:
Well, you're hedging a bit by saying, "relatively constant", where
before you said, "My main reason for AGC is just to maintain constant
audio output.."<g>
There are at least two possibilities.
One, your r-f/i-f subsystem cannot handle the full dynamic range of
all possible incoming signal levels so you need some gain control on
the front end. This could be in the form of an AGC-controlled PIN
diode attenuator, etc.
Two, (your case I believe) the r-f hardware is strong enough to
operate at full gain regardless of incoming signal strength. I'm
dubious that this is possible but that is just my opinion.
In case one, there should be some threshold, below which no AGC is
developed. This maintains the noise figure of the first r-f device,
whatever it is. (I realize that at say 160 meters the SNR is set by
external noise, but I'm speaking in the general sense.) This is what
we mean by "delayed AGC"; the application of control is delayed until
the signal level reaches some threshold. Clearly, if this is the case
then you cannot maintain a completely constant output without major
audio AGC and the resulting problems of loop stability and the effects
that I describe next.
In case two, your ACG system becomes nothing more than an audio
compressor. If your goal is constant output, then with no signal, you
get max noise out. When a signal comes along it just punches a hole
in the noise. For a good example of this listen to the guys (Italians
leap to mind) that transmit using a lot of compression and set the
audio gain wide open and sit across the room. You get to hear the
blower on the 10 KW amplifier, the wife in the next room and the dogs
barking in the street. When they finally say something the
compression kicks in and the noise subsides.
If we use your hedging position, of "relatively constant", then we
seem to be in agreement. There can be some area in the dynamic range
where the system gain is constant, thus stronger signals sound
stronger than weak ones.
Regards,
Wes
Bob Lewis (AA4PB)
> operate at full gain regardless of incoming signal strength.
This would mean that you could turn the AGC off, run the RF gain at
full tilt, use a good antenna, tune across 75 meters on a contest
weekend and not have any distortion on the strongest signals on the
band.
Roy Lewallen
Or switch in an attenuator, and switch it back out on Monday morning.
Roy Lewallen, W7EL
Bob Lewis (AA4PB)
morning.
Yes, but that represents a form of manual gain control. The
electronics is not capable of handling the full range of possible
signals. My point is that I have not experienced a receiver that can
handle the full potential range of signals without some type of gain
control, be it manual RF gain, an attenuator, or AGC.
Hans Summers
Thanks for the discussion, I have learnt a lot more about AGC.
I said "relatively constant" because I thought in practise "constant" isn't
likely to be possible. I now see why one wouldn't always want truely
constant output e.g. you don't want huge noise level when there is no
signal.
In my design I have no RF amplifier, just the FST3253 switching mixer. This
operates on a +5V supply and the incoming signal is biased to midpoint so I
think a 4V p-p signal would be switched without problems. Following the
mixer is 33dB of amplification before the phase shift network, these are
NE5534 op-amps on a 12V single-rail supply, so perhaps 10V p-p signal coming
out of those before clipping - which means 227mV p-p. So I should be able to
tolerate something like a 200mV input without overloading, right? What is
considered a maximum level that a receiver should be able to handle?
Ok, so what I will build is an AGC system having a variable "delay"
threshold below which it does not operate. Adjustable attack and hang times.
And adjustable depth of AGC to control how "relatively" constant the audio
output is (slope of the out dB vs in dB curve). I am planning on using the
CA3086 transconductance opamp similar to my use of LM13700 here
http://www.hanssummers.com/electronics/equipment/muscle/index.htm which was
very effective.
Thanks for all the answers
73 de Hans G0UPL
------------------
Hans Summers
http://www.HansSummers.Com
[...]