AGC control related issues

呜呼哀哉

AGC control related issues [Copy link]

 

邀请：@maychang   @chunyang   @gmchen   参与回复

There is an idea, but it is not easy to realize. As shown below:

ACL and ACN are waveforms of arbitrary shapes (frequency is not limited to 50Hz and will not exceed 1k). The amplitude and frequency change. How can the amplitude of the output signal be fixed at a certain value? The frequency and phase remain basically unchanged. When looking up AGC-related information, I found that this AGC is mostly used in communication receivers. The signal is actually very small and the frequency is very high. So what are the solutions to realize the idea of the signal in the above figure? If it is a DC fixed amplitude signal, it is easy to deal with. You can give a benchmark and control the amplifier to a certain value. For AC uncertain waveform signals, there seems to be no benchmark to refer to.

maychang

[How can I fix the amplitude of this output signal at a certain value? ]

It is impossible to fix it at a certain value. The only way is to transform a signal with a large amplitude change into a signal with a smaller amplitude change.

maychang

As for the signal you mentioned, you can refer to the circuit of a wired telephone, which usually uses an ordinary diode to limit the amplitude.

呜呼哀哉

maychang posted on 2024-4-15 16:06 [How can I fix the amplitude of this output signal at a certain value? ] It is impossible to fix it at a certain value. I can only fix the amplitude of the signal to a certain value.

Moderator, what I hope is to proportionally enlarge or reduce the ACLN to a certain value without distortion. Once it is limited, the waveform will change immediately and the goal cannot be achieved.

maychang

Alas, published on 2024-4-15 16:39 Moderator, what I hope is to proportionally enlarge or reduce the ACLN to a certain value without distortion. Once the amplitude is limited, the waveform will change immediately, and the goal cannot be achieved...

[What I hope is to proportionally enlarge or reduce the ACLN to a certain value without distortion]

This is even more impossible.

You can refer to the amplitude stabilization circuit of the Wien bridge oscillator circuit. Common ones include diodes, junction field effect tubes, and thermistors.

maychang

Alas, published on 2024-4-15 16:39 Moderator, what I hope is to proportionally enlarge or reduce the ACLN to a certain value without distortion. Once the amplitude is limited, the waveform will change immediately, and the goal cannot be achieved...

[What I hope is to proportionally enlarge or reduce the ACLN to a certain value without distortion]

You can use AD conversion to get the amplitude of your AC signal, and then control the programmable amplifier according to the amplitude of the AC signal. However, the amplitude cannot be determined, and the time lag is quite large.

呜呼哀哉

maychang posted on 2024-4-15 16:56 [What I hope is to proportionally enlarge or reduce the ACLN to a certain value without distortion] You can use AD conversion to obtain the amplitude of your AC signal, and then...

There are too many problems with using ADC. First, the input signal frequency is uncertain and the ADC sampling rate is difficult to control. Second, sampling is performed at an extremely high speed, and then the DCA output is used to perform variable gain amplification on the output, which is indeed slow.

I thought of the sampling peak hold method. After all, there is a maximum amplitude limit for the input signal. No matter what the waveform is, the peak value is taken out, compared with the DC reference, and the overall amplification is proportional.

The only problem is that the signal is periodic and the amplitude will change. The peak hold circuit needs to have a discharge process. During the discharge, the difference between the peak value and the DC reference will fail to control the gain. I haven't thought of a solution yet.

maychang

Alas, published on 2024-4-15 17:36 There are too many problems with using ADC. First, the input signal frequency is uncertain, and the ADC sampling rate is difficult to control. Second, the sampling is performed at an extremely high speed, and then...

[The amplitude will change, and the peak holding circuit needs to have a discharge process]

This problem existed before transistors were invented. The solution is to use a compromise value for the time constant of the peak detection (the discharge process of the peak hold circuit you mentioned), which is neither too large nor too small.

呜呼哀哉

maychang posted on 2024-4-15 18:45 [The amplitude will change, and the peak holding circuit needs a discharge process] This problem has existed since the days before transistors. The solution is the peak...

I don't quite understand. The electricity is discharged instantly through the ohm-level resistor in series with the MOS tube. It cannot be too fast or too slow. How can we control it?

gmchen

Isn't the answer already given in your diagram! AGC circuit!

AGC is a circuit created to achieve your purpose.

AGC is not only used in communication equipment, there are similar circuits in the audio range, but it usually has another name: ALC, automatic level control

gmchen

The main structure of the AGC circuit is a voltage-controlled gain amplifier (VGA). The signal is amplified by this amplifier, and then the amplified signal is detected to obtain the peak value (or average value, or effective value, as needed) of the output signal. The voltage obtained by this detection is fed back to the control end of the VGA. By changing the amplification factor of the VGA, the amplitude of the output voltage can be controlled within a certain agreed range of variation.

gmchen

There are many VGA chips available now, so the implementation is not complicated

gmchen

There should be clear indicator requirements, including:

1. Maximum gain. Usually when the input signal is very small, the amplifier has maximum gain.

2. Start-up level. When the input signal (or output signal) reaches the start-up level, the AGC amplifier enters the start-up state. At this time, the output amplitude no longer changes with the input amplitude, but the frequency and phase of the signal remain unchanged. In fact, only the gain of the amplifier begins to decrease.

3. Control range. The control range of AGC is not infinite, it starts from the start control level and ends at the end level. After exceeding the end level, the amplifier works at the lowest gain (the gain will not decrease any further)

maychang

Alas, published on 2024-4-15 19:41 I don't quite understand. The discharge is instantaneous through the ohm-level resistor in series with the MOS tube. How can this be controlled so that it can't be too fast or too slow?

The load of the peak detection circuit is a resistor and a capacitor in parallel, which of course has a time constant.

呜呼哀哉

gmchen posted on 2024-4-15 19:43 Didn’t your diagram already give the answer! AGC circuit! AGC is a circuit created to achieve your purpose. AGC is not only used in...

Thank you, teacher. I will search for some ALC related content.

I once used a MOS tube to make a variable resistor, divided the input voltage by a certain ratio, and then used a digital potentiometer to divide it again as a reference, which was used as a reference signal for the constant current source. I didn't think it through at the time. When the input signal was small, the current could not be very large. Now I want to control the current to be the same when the signal changes within a certain range.

呜呼哀哉

maychang posted on 2024-4-15 20:19 The load of the peak detection circuit is a resistor and capacitor in parallel. This resistor and capacitor in parallel, of course, has a time constant.

If the load is a resistor and a capacitor, the time constant is small, the gain amplitude drops rapidly, and the waveform will still change after gain control. When the time constant is large, the signal amplitude changes slightly faster and cannot keep up.

I am also thinking about adding a peak hold function, using an op amp with enable. The enable signal is obtained from the zero crossing point of the signal. The rising edge turns off the rear op amp, and the op amp holds the previous peak value. After the falling edge turns on the op amp for a certain period of time, the front peak circuit is directly discharged to wait for the next peak value. In this way, the gain can at least be kept fixed within one cycle.

maychang

Alas, published on 2024-4-16 08:38 If the load is a resistor and a capacitor, the time constant is too small, the gain amplitude drops rapidly, and the waveform will still change after gain control. If the time constant is too large...

[If the load is a resistor and a capacitor, the time constant is small, the gain amplitude drops rapidly, and the waveform will still change after gain control. When the time constant is large, the signal amplitude changes slightly faster and cannot keep up.]

That was the method before the emergence of digital technology. After the emergence of digital technology, this phenomenon has been greatly improved.

Please consider the method mentioned by teacher gmchen on floors 10 to 13.

gmchen

Alas, published on 2024-4-16 08:38 If the load is a resistor and a capacitor, the time constant is too small, the gain amplitude drops rapidly, and the waveform will still change after gain control. If the time constant is too large...

The AGC circuit has a response time problem.

If the response time is too slow, the output amplitude will still fluctuate.

If the response time is too fast, fluctuations in the input signal that contain information (such as amplitude changes in an amplitude modulated signal) will also be smeared out, a condition known as demodulation.

Therefore, the response time of AGC should be designed reasonably.

The response time is mainly determined by the RC time constant of the detection link.

gmchen

Woohoo! Posted on 2024-4-16 08:30 Thank you, teacher. I will search for ALC related content. I once used MOS tube to make a variable resistor, divided the input voltage by a certain ratio, and then used digital...

You don't have to search for ALC, it's just a different name.

I have an article "Talking about AGC Circuit" in this forum, you can refer to it.

https://bbs.eeworld.com.cn/thread-643656-1-1.html

呜呼哀哉

gmchen posted on 2024-4-16 09:02 You don't need to search ALC, it's just a different name. I have an article "Talking about AGC Circuit" in this forum, you can refer to it. ...

I have started reading this post. Someone in the forum also asked about AGC issues just now. I saw the relevant content and started learning. Thank you, teacher.