LAUNCHXL-F28379D: Confusion over capacitive load drive of OPA350 and others.

JM Holland

Part Number: LAUNCHXL-F28379D
Other Parts Discussed in Thread: OPA350

This may be a more general question for the amplifiers team, but it concerns something I have always been confused about regarding the ADC voltage buffers on the Launchpad.

They use a capacitive load as high as 22uF. Yet, the data sheet states that the maximum capacitive load is much less than this, and requires the use of a resistor of X value to isolate the opamp from the effects of the capacitive load.

How can the OPA350 be used to drive such a high capacitive load? How come a small isolation resistor is not included in the reference buffer output?

My suspicion would just be because the input is a D.C. voltage, but what about start-up? Would we expect that the reference voltage would massively overshoot at start-up with such a high capacitive load? Would a compensatory capacitor in the feedback loop with a resistor help at all? If the OPA350 can be used with such a high capacitive load, could a much higher bandwidth opamp with capacitive drive as low as 10pF be used with a 22uF capacitive load? It not, why? Thanks!

best regards,

Joel

over 2 years ago

0 Raymond Zhang1 over 2 years ago

TI__Guru* 77501 points

Hi Joel,

JM Holland said:
If the OPA350 can be used with such a high capacitive load, could a much higher bandwidth opamp with capacitive drive as low as 10pF be used with a 22uF capacitive load? It not, why?

With 10pF capacitive load, OPA350 should have BW at approx. 38MHz as shown in the datasheet. With 22uF capacitive load, the OPA350's BW will be significantly reduced, depending on the input waveform.

Please take a look at the op amp stability video series, and there are 7 video series on the op amp stability topics, and I only listed 3 links below.

When additional pole is introduced within an op amp's closed feedback loop, the compensated op amp circuit within the loop has to create zero to cancel out the pole, and maintain -20dB/decade of rate of closure pass through the unity gain bandwidth of the loop gain.

https://www.ti.com/video/4080320453001?keyMatch=OP%20AMP%20STABILITY

https://www.ti.com/video/4080336905001?keyMatch=OP%20AMP%20STABILITY

https://www.ti.com/video/4080254925001?keyMatch=OP%20AMP%20STABILITY

If you have additional questions, please let me know.

Best,

Raymond

0 JM Holland over 2 years ago in reply to Raymond Zhang1

Prodigy 240 points

Hi Raymond

Thanks for your reply. Just to note, this circuit diagram I have presented is taken straight from the Launchpad user guide - I did not design this myself. If you are right and the opamp configuration will be unstable, is the schematic for the launchpad wrong? Our requirements are no different from the Launchpad, this is for the ADC VREFHI and VREFLO, but we will also use it to drive the ADC channels with lower capacitances of about 2nF. We may be able to use the simple isolating R method to make this stable with low overshoot, but with 22uF it is a different story. Our input waveforms will be mainly D.C. signals, with one 175kHz signal that is filtered before being passed to the comparator subsystem. The ADC On that channel will be deactivated.

If you could please clarify why the Launchpad is able to have the circuit I have sent, or whether it is missing something important, please do let me know.

Best regards,

Joel

+1 Raymond Zhang1 over 2 years ago in reply to JM Holland

TI__Guru* 77501 points

Hi Joel,

I made a mistake, and the buffer circuit is stable upon further investigation. Sorry for the confusion.

I forgot to place 100mΩ resistor in series with 22uf ceramic capacitor initially, and this results no adequate phase margin in the circuit. The 100mΩ + 22uF behave like a snubber circuit and create a zero at approx. 72kHz, and lift the phase of the circuit with phase margin of 58 degrees at the unity gain bandwidth.