I have a problem with the VGA AD8368. It shows a strange behavior when using it with higher gains.
I attached a couple of measurements which hopefully helps to find out what I'm doing wrong.
The output centering loop, which the high-pass corner is set by either the HPFL cap or the DECLs, should only null out signals that are below the high-pass corner, so I wouldn't worry about the loop being fast enough, because it's only supposed to work at frequencies relatively close to DC. I think the highest frequency corner you can set the loop to is 1 or 2 MHz. There could be something wrong with the output centering loop, but that's not the first thing I would investigate.
What I'm guessing is going on, is that the part is probably oscillating and/or on the verge of oscillation; hence the overshoot in your oscilloscope screen captures. I noticed you do not have the RL snubber network (R3 and L1) on the input that is recommended in the datasheet (see pages 16 and 17) for stability reasons. I would try the RL snubber network first and see if that has an affect on your on the behavior. I would also try looking at the output on a spectrum analyzer to see if there is an oscillation. Usually, the oscillation will happen at frequencies between 1.5 GHz and 2.5 GHz, so you will know it when you see. Running the same gain step response test at high temperature (85 °C) might give a clue as to if the AD8368 is oscillating because the oscillations usually get worse at high temperature.
Let me know if any of this helps and/or if you have any questions.
I moved this question about the AD8368 to the RF Components community. Please continue the discussion here.
EngineerZone Community Manager
you were right. I checked with a spectrum analyzer today and found out, that there is an oscillation of 1.4...GHz and 2.8...GHz. Inserting the RL snubber network fixed the problem, thanks a lot.
Now I can also play a little with the HPFL and DECL values to optimize the signal quality.
I have also to replace the AD8009 because it is not performing well in inverting mode (large signal BW is too low). In our application we need a noninverted signal and regarding to my measurements a saw the AD8368 is an inverting amplifier (which I think was not mentioned in the datasheet). That's why I'm thinking about stacking 2 AD8368 which will increase the gain range and will give me a noninverted output signal.
I'm glad your problem is fixed with the RL snubber network. Yeah, there should be some text in the datasheet about the AD8368 inverting the signal. In a lot of the AD3868 applications either phase isn't really important (hence no phase delay plots) or there is some sort of PLL downstream of the AD8368.
If you want 2 cascaded VGAs, we have the ADL5336:
It's a little different than the AD8368 (200 Ohm input Z, differential input/output, programmable gains via a SPI port, etc. ) but if real estate on your PC board is tight, it might be a better option depending on your application.
Retrieving data ...