I am planning to use a ADL5205 to provide variable gain in a DC coupled pulse processing system. I plan to use the ADL5205 to drive a AD9694 500 MSa/s 14 bit ADC.
It appears I should set the ADC for 2.16 V input range. I hope this is OK when DC coupled?
I intend to put an anti-aliasing filter between the ADL5205 and the ADC, so should not drive the ADC past its maximum differential input voltage.
However next I need to protect the input to the ADL5205.
The input signal will be single ended, so I will use a diff amp (to be chosen) to convert from single ended to differential, to provide the additional gain so that gain at the highest gain the input signal range is 50 mV and to offset the input to support uni-polar negative pulses.
This should be OK with the ADL5205 set to its highest gain, but as I lower the gain for larger input signals, my understanding is that I will get to a point where I exceed the maximum differential input to the ADL5205. Due to anti-aliasing filter (with a DC gain of 0.8) my full range signal at the output of the ADL5205 is 2.16/0.8 = 2.7 Vp-p
Hence the minimum gain I can use the ADL5205 at is 2.7/2 = 2.6 dB.
Have I understood this correctly? Does this mean that the lower gain settings of the ADL5205 can only be used I system using a lower output signal?
I also need to protect against atypical events producing large pulses and against user error.
I am going to look at putting an attenuator between the input diff amp and limiting the supply range of the input amplifier.
> It appears I should set the ADC for 2.16 V input range. I hope this is OK when DC coupled?
That matches the max Differential Input Voltage Range on Page 5 of the AD9694 datasheet. This should be independent of AC/DC coupling.
> The input signal will be single ended, so I will use a diff amp (to be chosen) to convert from single ended to differential, to provide the additional gain
The ADL5205 can be driven single-ended.
Furthermore, the ADL5205 already provides 26 dB of gain. Adding another stage may be counterproductive, given pulses require wide bands, so you would end up with a system of high gain-bandwidth product. In such systems noise can be a potential issue, because the wide band allows significant noise energy on to the input, which is then gained substantially. In the worst case, noise alone can drive the system output rail-to-rail. That's why multi-stage architectures, for example, usually limit the bandwidth.
Secondly, the input can be shifted and attenuated with a resistive pi attenuator, at the cost of two additional voltage sources, here's an example:
This may help you with large pulses or user error.