I am designing a broad band 40 dBm power amplifier (PA) in the frequency range of 0.5-2.5 GHz. I would like to implement Automatic Gain Control (AGC) to precisely control the output power of the PA. After much research, I decided to use HMC1120, mainly because from the RMSOUT vs. Pin diagram (as shown below), it appeared to me that HMC1120 offers consistent RMSOUT for my frequency range of 0.5-2.5 GHz and also good sensitivity. (I understand that I need to convert the RMSOUT with something like AD8062 to provide a negative loop for AGC).
When I started looking for variable gain amplifiers (VGA), the ADL5330 appeared to be an attractive choice. However, going through the datasheet, I noticed that the example shown in Figure 37 of the ADL5330 datasheet (shown below) has used AD8318 for detecting the power in AGC.
I started exploring the AD8318 datasheet and noticed that Figures 42 and 43 of the datasheet shows a similar diagram.
AD8318 also appeared to offer consistent Vout for my frequency range but obviously the slope is negative and this will eliminate the need for converting the slope:
More research led into the Figure 44 of the ADL5513 datasheet, which also offers consistent Vout for my frequency range (as shown on page 8 of the data sheet)
So, now I am confused as to which components to use. Here are my questions:
1) Can you please explain which component I should use in my design (HMC1120, AD8318, or ADL5513) and why?
2) As to VGA, can I use ADL5330 to cover the frequency range of 0.5-2.5 GHz? Is there a better choice that I should look into?
3) I am planning to use a directional coupler at the output of the PA, and then connect the coupler (with correct attenuation) to the HMC1120 to monitor output power. In addition, I need to use a gain stage between my PA and ADL5330 to provide sufficient input for my PA. That is, I will have three "stages" or "modules" in my control loop. Would the same control loop --such as the one shown in Figure 42 of the AD8318 datasheet-- work stably in my configuration? (In all examples --such as Figure 42 of the AD8318 datasheet-- the AGC is a single stage.).
Thanks again for your help.
I should also note that I am planning to use HMC8500PM5E as my PA.
1) Is your RF signal a complex amplitude waveform (QAM or AM modulation, for example)? If so, an RMS responding detector is recommended for best power measurement accuracy. Then consider the detection power range and accuracy requirement. The detector should have at least as much range as is required, preferably more, to accommodate production variation in the detector, and variation in the power coupling factor. If you are leveling over a relatively wide power range, perhaps a linear-in-dB detector is best. For most precise power leveling up at the high end of the power scale, a linear-in-Volts/Volt rms detector is very precise at the max power, while trading off some accuracy down at lower power levels. If your RF signal is always a constant envelope signal, then a log power detector can be calibrated to give excellent accuracy over a very wide power range and across temperature. From the RF power detector homepage, you can choose rms or log, then there will be a separate online product selector guide for each.
2) ADL5330 appears to be suitable for this application. Make sure the variable gain range is sufficient. If any detailed questions about this device, its best to ask in a separate EngineerZone thread.
3) Adding gain and RF power should be fine. From systems point of view, it will behave like a closed-loop control system. A simple Bode analysis could be performed to ensure closed-loop stability and a desirable transient response. Many of the Analog Devices detectors feature dual mode: measure, and controller. In measure mode, the detector is configured in the conventional way, to detect. In controller mode, the internal detector output amp functions as a built-in error amp, saving you from needing another external amp IC.
Please let us know if any further questions. Its important to choose a good detector for the application.
Bruce_H said:For most precise power leveling up at the high end of the power scale, a linear-in-Volts/Volt rms detector is very precise at the max power, while trading off some accuracy down at lower power levels. If your RF signal is always a constant envelope signal, then a log power detector can be calibrated to give excellent accuracy over a very wide power range and across temperature. From the RF power detector homepage, you can choose rms or log, then there will be a separate online product selector guide for each.
Thanks Bruce for your help. I am designing this amplifier for RF power applications and the input will be constant amplitude sine waveform. Based on what you mentioned, I think the best shot would be to use linear-in-volts detector (something like ADL5511?) because I need precise measurement at high power (lets say 0.5-10W) and I do not care about the low end. The RF power detector search does not appear to filter based on linear-in-volt or dB. Is that right or am I missing something? Are there any component similar to ADL5511 that comes with the controller?
Also, do I need to use an anti-log for AGC if I decide to go with linear-in-volts detector, for example ADL5511?
0.5 to 10W is 13dB range, which is considered fairly wide range for a linear-in-Volts-per-Volt detector. The advantage of accuracy up at the top end of the range will become degraded over 13dB range. For this wide of a range, also consider a detector with linear-in-dB response. A linear-in-dB response simplifies the design of the leveling loop, provided your VGA is also linear in dB's per volt. And sorry, linear-in-Volts or dB's is not (yet) a selector guide field.
For constant-amplitude sine wave, accurate results are attainable with log or rms detectors. ADL5511 is what we call a combo detector, because it has both rms and envelope detectors in the same package. These are popular for certain linearized PA applications which call for both detector types. A combo detector is generally not what we would recommend for a power leveling loop.
We cannot choose the detector without knowing how much amplitude detection range is needed. Please specify if known. You may wish to consider the AD8317 Log detector/controller, if 55dB detection range is acceptable. Datasheet Figure 30 shows a simple ALC design which could be adapted to this application. The datasheet text provides a through description:
I think I am now confused! Here are more questions:
Bruce_H said:0.5 to 10W is 13dB range, which is considered fairly wide range for a linear-in-Volts-per-Volt detector.
I am not sure if I fully understand as to why 13dB is considered a "fairly wide range"? ADL5511 has Input power dynamic range of 47dB. So, 13dB would be a fraction of that. Can you please clarify?
Bruce_H said:The advantage of accuracy up at the top end of the range will become degraded over 13dB range.
Lets say my output is 0.5 to 10W (27 to 40dBm). This range can be translated with the help of a fixed -30dB attenuator to -3 to 10dB range, which falls in the linear range of ADL5511 (see Figs 4 and 7 of ADL5511 datasheet rev D). I am not sure if I fully understand as to why you say the top end will become degraded. Can you please clarify?
Bruce_H said:We cannot choose the detector without knowing how much amplitude detection range is needed. Please specify if known. You may wish to consider the AD8317 Log detector/controller, if 55dB detection range is acceptable.
Do we need power detection or amplitude detection? (This is a RF power amplifier to produce RF power (not for signal) and there are no modulations). Therefore the peak voltage and the power are correlated. Also, isn't my range 13dB?
Also, will you be able to share the sys-parameters files? I am doing some simulations in Genesys and plan on importing the parameters. Thanks.
Sorry for the confusion. Lets see if we can clarify:
The linear-in-Volts-per-Volt detectors don't implement the log function. They tend to be most accurate when driven at a high level on the response curve, but always well within the Abs. Max. rating. For power leveling application with an ADL5511, the rms output is recommended. Datasheet figures 19, 21, and 26 show the approximate temp drift and part-to-part variation.
Regarding pwr vs amplitude, devices such as ADL5511 respond to amplitude, but we often take measurements relative to a known input power, with RF input impedance-matched. This is what you see in datasheet. A true power measurement would require a directional coupler, to separate reflected wave from incident wave.
Sorry, Analog Devices does not provide a system level or behavioral model for ADL5511.