This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

OPA2365: DC bias for ADC sampling of an RF tuner (IF 800KHz)

Part Number: OPA365
Other Parts Discussed in Thread: THS4531A, THS4561, OPA2365

Hello

I have an RF tuner providing differential 2.0V PP output, which i need to sample with the ADC  of an STM32L4, this ADC can only accept positive voltage and the package i used (QFN48) does not give access to VREF+ and VREF-, they are internally tied to VCCA and AGND, and AVCC has to be 3.3V. Thus the ADC REF can only be 3.3V. The ADC can be used as single-end or differential input, depending on the configuration, but in any case the input cannot be negative with respect to STM32L4 GND. There is 3 problems here, balanced to unbalanced conversion, amplitude scaling and DC biasing. 

The first issue is sorted by using an impedance transfomer 4:1 which has balanced input and unbalanced output.

The second issue could be sorted with the OPA365 which leads to the third issue : DC bias.

Of course the signal coming out of the transformer is swinging over and below 0V, 2.0V PP max. If i directly connect one transformer output to GND and the other to OPA365 i end up clipping the whole bottom half of the wave, which is a problem since the system does demodulation of an RF signal. So the only option here seems to be : DC biasing the OPA365 at 1.65V so that it provide the full sine wave swinging between 0V and 3.3V (ideally), however i fail to even simulate that in TINA. Is there some advice on how i should proceed with this? Of course if another approach is better i am open to suggestion, especially if i can get rid of the transformer, keeping in mind that the design has some contraints, mainly : available space and BOM cost. Basically the less parts we have between the tuner and the ADC the better.

Kepping the signal differential all along including to ADC would be ideal, but i can get to that in a second time, for now i am mainly trying to achieve the DC biasing with single end on the current test board, which only has one OPA365. 

Thanks.

  • Why don't you use an AC coupled input Diff to diff with the output Vcm for the ADC set using the Vocm control pin for the THS4531A FDA. 

    This will take care of all your issues in one stage - the newer THS4561 might be a little lower noise if that matters. You might need a post RC to the ADC, but only one active device - leave the output DC coupled so the FDA output CM biases the ADC where it needs to be, You can also add an active LP into the FDA stage at the cost of RC elements if you wanted to

  • Thanks for the suggestion, i'll look into it but at first glance the cost of THS4531A is 3 times the cost of a dual OPA365.

    Meanwhile i have managed to make a working model with TINA :

    and even tested it, it turned out it works perfectly.

    Right now i use the balun single end output to feed the OPA365 which in turn feeds the ADC in single end mode.

    So i wonder if i could get rid of the balun and use the same circuit on the other side of the tuner output? It would allow me to feed the ADC in differential mode.

    About using THS4531A, what would be the advantage compared to a dual OPA365? The really higher cost is a problem, but if it can bring some improvement i can consider it.

  • Well Bidault,

    If your ADC is diff in, driving single ended loses a bit of resolution I think. 

    For this lower F diff I/O the FDA might be overkill, any dual low voltage op amp can do the same thing with diff input inverting mode and the ADC bias as a DC bias on the V+ inputs. 

    The OPA2365 brings a nice feature set that you don't use here. You don't need RRInput if you are running inverting. - yoju don't need good DC precision if you are AC coupled signal path. You probably do need some speed for some dynamic range margin. 

    Your ckt above is using very high R values, that will introduce noise and phase margin issues. Here is a concept - substitute what dual you want, Adjust RC's to design needs, 

    Here is the cost page

    Here is the file, 

    OPA2834 diff IO.TSC