OPA4350: TMS320F28379D VREF-HI

Hi Dany,

The main things to consider here are:

• Op-amp offset error will affect the total gain error specification (along with whatever IC or other source you choose to provide the reference voltage, and the ADC itself).

OPA320 => 150uV = 0.2LSBs (12-bit mode, 3.0V reference) = 3.3LSBs (16-bit mode, 3.0V reference)

OPA350 => 500uV = 0.7LSBs (12-bit mode, 3.0V reference) = 10.9 LSBs (16-bit mode, 3.0V reference)

OPA378 => 50uV = 0.1LSBs (12-bit mode, 3.0V reference) = 1.1 LSBs (16-bit mode, 3.0V reference)

OPA625 => 3mV = 4.1LSBs (12-bit mode, 3.0V reference) = 65.5LSBs (16-bit mode, 3.0V reference)

(I just pulled the first max Vos number I saw out of the datasheets, you would also want to add in or ensure this comprehends the operating temperature range of your application)

Note: A chopper-corrected op-amp, while having very low offset error, may not be a good choice for the ADC reference driver because you may see some tones centered at the chopping frequency (and they are usually lower BW).    

• Op-amp bandwidth.  In addition to the large capacitor on the pin, you are going to want a high bandwidth op-amp to help settle out the high current transients on the VREFHI pin.  There is somewhat of a trade-off you can make between the VREFHI capacitor size and the op-amp BW.  We haven't evaluated this thoroughly, but I would recommend the BW be at least 10MHz. 

OPA320 => 20MHz

OPA350 => 38MHz

OPA378 => 900kHz (Note: probably not appropriate for driving the ADC reference)

OPA625 =>120MHz

• Op-amp stability when driving a large capacitor.  You may have noticed in our recommended VREFHI driving circuits we have OPA350 + 22uF capacitor + 100mOhm resistor and OPA320 + 2.2uF capacitor + 560mOhm resistor.  The resistor is to ensure the op-amp remains stable when driving the capacitive load.  A better op-amp for ADC reference driving will be able to handle a larger capacitor while still using a small damping resistor.  The only way I know how to check this is to simulate the phase margin using TINA-TI (I can explain how to do this in more detail if needed).
• Op-amp output noise.  Note: the capacitor on the VREFHI pin will also help attenuate this noise a little bit.   

OPA320 => 7nV / sqrt(Hz) = roughly 0.1 LSBs of peak-to-peak noise (12-bit, 3.0V range, 1MHz BW) or 1.2 LSBs of peak-to-peak noise  (16-bit, 3.0V range, 1MHz BW)

OPA350 => 7nv / sqrt(Hz) 

OPA378 => 20nv / sqrt(Hz) = roughly 0.2 LSBs of peak-to-peak noise (12-bit, 3.0V range, 1MHz BW) or 3.5 LSBs of peak-to-peak noise  (16-bit, 3.0V range, 1MHz BW)

OPA625 => 2.5nV / sqrt(Hz) = roughly 0.0 LSBs of peak-to-peak noise (12-bit, 3.0V range, 1MHz BW) or 0.4 LSBs of peak-to-peak noise  (16-bit, 3.0V range, 1MHz BW)

(These are all more than capable for 12-bit operation and the OPA378 is maybe borderline for 16-bit operation, but would probably be OK).  

• And you obviously want to consider other factors like package size, cost, current consumption, supply range, temperature rating, automotive qualification, etc.

Hi Devin,

Thank you very much for your detailed explanation. I will use the OPA4350 for my design.

It will be great if you could explain how to simulate the phase margin with Tina.

Best Regards