ALC circuits discussion

[tszaboo]

Recently I've become interested in ALC (automatic level control) or AGC (automatic gain control) circuits for RF.
They were in several RF equipment that I was repairing, and they can be quite complex, or extremely simple in some cases.
I don't see too much discussion about them online. Often times the published circuits will contain obsolete RF diodes, or circuits without any sort of explanation or specifications. What is the current method of building such a circuit?
I know that technically, we can just connect a single chip RF logarithmic sensor to a VGA, and have the circuit working. But that feels like throwing money at a problem, that was solved with simple diodes in the past. By simple I mean PIN diodes but you get my point.

[ftg]

I expect to see less and less PIN-diode attenuators or voltage variable attenuators and more and more of everything happening in the digital domain.
You already have fully digital RF generation at baseband, IF or direct to RF anyway. And you have feedback from the transmitter output for DPD.
So closing the loop in software saves parts.

It does not help that PIN-diodes, especially the nice quads meant for variable attenuators are increasingly gone from the selections of various manufacturers.
And in real world designs that I encounter the RF logamps are increasingly disappearing from boards.
In more consumer electronics oriented stuff the RF detector diodes in wifi stuff for forward power monitoring seem to be also going away and getting replaced by forward power detectors integrated directly into the RF power amplifier chips.

[tszaboo]

I expect to see less and less PIN-diode attenuators or voltage variable attenuators and more and more of everything happening in the digital domain.
You already have fully digital RF generation at baseband, IF or direct to RF anyway. And you have feedback from the transmitter output for DPD.
So closing the loop in software saves parts.

It does not help that PIN-diodes, especially the nice quads meant for variable attenuators are increasingly gone from the selections of various manufacturers.
And in real world designs that I encounter the RF logamps are increasingly disappearing from boards.
In more consumer electronics oriented stuff the RF detector diodes in wifi stuff for forward power monitoring seem to be also going away and getting replaced by forward power detectors integrated directly into the RF power amplifier chips.

So you think a modern design is supposed to use one of these digital attenuators, such as the IC below?
I guess that makes sense. Of course we don't see too many log-amps anymore, for the majority of RF transmitters. They will have a well-characterized RF front-end, which will work at a single frequency band. I also use BLE and Zigbee ICs, almost everything except the antenna and a few passives for matching is integrated, and the output power is 2dB accurate from -40C to 85C. Which is good enough.

Would you place the power sensor before or after the attenuator?

[ftg]

...
So you think a modern design is supposed to use one of these digital attenuators, such as the IC below?
...

Nope, I mean that the dynamic range in the DAC's used is so high that while there might be a step attenuator in the signal path to set the coarse RF level, the actual levelling happens in the DSP's digital domain, before it even enters the DAC to be turned into analog RF.

I would place it after the power amplifier. Because that's where the level we actually care about is.
Different temperatures, supply voltages, tolerances, PA droop and all that adds up and that's what I'd want to compensate for.
So I want to know the actual RF output power and not just hope for the best and always undershoot.

[tszaboo]

...
So you think a modern design is supposed to use one of these digital attenuators, such as the IC below?
...

Nope, I mean that the dynamic range in the DAC's used is so high that while there might be a step attenuator in the signal path to set the coarse RF level, the actual levelling happens in the DSP's digital domain, before it even enters the DAC to be turned into analog RF.

I would place it after the power amplifier. Because that's where the level we actually care about is.
Different temperatures, supply voltages, tolerances, PA droop and all that adds up and that's what I'd want to compensate for.
So I want to know the actual RF output power and not just hope for the best and always undershoot.

OK, if you want a digital radio. But for example I was looking at the possibility to do ALC on a signal, that's not generated in the device.
For example you want to make a high frequency input module for a timer-counter. I have a Phillips PM 6680, without option C, high frequency input. I've reviewed several RF frequency dividers, and they have an ALC circuit on the input, to normalize the input signal to somewhere between 0-10dBm. The input frequency is not known.
But this is just one example. ALC would be very useful after an RF filter, to compensate for the insertion loss. Where there is a lot of ripple in the passband.

[ftg]

Aaah. Now I'm starting to understand.
But does the PM6680B even care about the absolute amplitude? as long as it's within the acceptable range?
Unless you want to use the AC/DC voltage measurement range, which is definitely it's own can of worms.
If you want it to do levelling down to DC it will be a rather challenging build.
Especially when I'd expect the ALC/AGC leveller to make that function useless.

Maybe something like:
Input to leveler -> gainblock that works across the frequency range of interest -> continuously variable attenuator, for example four PIN-diodes if at RF and not 50Hz. -> resistive tap to logamp -> commercial +10dBm limiter chip -> PM6680B input.
With the variable attenuator being steered with the output from a logamp with power amplifier control function.
Maybe something like AD8317 or AD8363, for example.

[tszaboo]

Aaah. Now I'm starting to understand.
But does the PM6680B even care about the absolute amplitude? as long as it's within the acceptable range?
Unless you want to use the AC/DC voltage measurement range, which is definitely it's own can of worms.
If you want it to do levelling down to DC it will be a rather challenging build.
Especially when I'd expect the ALC/AGC leveller to make that function useless.

Maybe something like:
Input to leveler -> gainblock that works across the frequency range of interest -> continuously variable attenuator, for example four PIN-diodes if at RF and not 50Hz. -> resistive tap to logamp -> commercial +10dBm limiter chip -> PM6680B input.
With the variable attenuator being steered with the output from a logamp with power amplifier control function.
Maybe something like AD8317 or AD8363, for example.

This is just one example, but it's something that I've been working on (system level). The C, high frequency input is typically between ~100MHz to few GHZ. It has a frequency divider inside that divides the input down to about 150MHz and then converts it to ECL logic. The divider is the sensitive part, which needs typically a few dBm  (lets say +5dBm) input power, so the ALC would need to convert the incoming unknown level RF to this.
My system block diagram is similar to what you described. I was wondering if it's possible to do it without microcontroller at all. The original circuits didn't had that, they had dividers that are obsolete now, pin diode limiters that they also don't make anymore. I brought this up as an example as I want to actually build this, and learn from it.

[ftg]

If you want to do it without any sort of control loop, then one method could be driving some high gain gain block into a limiter, followed by an attenuator.
Gain block saturates, but never produces enough power to destroy the limiter.
Limiter limits the output to somewhere between +10 to +16dBm, for example and then the final attenuator drops that down to below +10dBm.

No control loops to think about and gain variation in the gain block does not matter, as it will be run into saturation and then clipped by the limiter anyway.

[BigBoss]

If you talk about to fix the RF Power Level of a synthesizer/source circuit, I can say that the most reliable method is to adjust the RF level during interested band is to use
a digital RF attenuator with smallest step (0.25dB is good enough ) then to control this attenuator by aid of a uC circuit regarding to calibrated level data.
Calibration can be done automatically by a software piece to compensate the level variation at the output.

Linear PIN Diode continuous time attenuators are not used anymore due to their poor temperature characteristics. VGAs can also be used but their frequency response is limited.
But it's always possible to find precise and repeatable digital attenuators in the market.

For instance,
https://psemi.com/products/rf-attenuators/