TDA4VM: Recommendation for SAMPLEDELAY to charge the AFE capacitance during the acquisition period

Jelena

Part Number: TDA4VM

Hi Kevin,

In the referenced E2E post https://e2e.ti.com/support/processors-group/processors/f/processors-forum/1096609/tda4vm-what-s-the-input-impedance-of-adc-port-of-tda4vm, you state the following:

But here is the key point: the 1K resistance and the 5.5pF (+ parasitic capacitances) creates an RC time constant of 5.5ns. You always need 9 or 10 time constants to charge to a 12-bit accuracy level.

Is this relationship between RC time constant and the time it takes to charge to a particular accuracy level (9-10 time constants for 12 bit accuracy) specific to the TI ADC design or is this ADC design in general? Can you point me to any further collateral on this relationship?

Thanks,

Jelena

over 2 years ago

0 Jelena over 2 years ago

TI__Expert 7225 points

Rephrasing the earlier question and adding a second one:

The referred post mentions 9-10 RC constants needed to charge to 12-bit accuracy level (the specific example had 5.5ns RC constant and the recommendation was to setup SAMPLEDELAY to give 55ns before sampling).

1. Is the need for "9-10 time constants to charge to 12-bit accuracy" documented somewhere other than the E2E as a recommendation for the choice of SAMPLEDELAY?

2. Does this recommendation for SAMPLEDELAY change if sample averaging is used (i.e. is 55ns still needed if the input is sampled and averaged 16 times)?

Thanks,

Jelena

0 Kevin Lavery over 2 years ago in reply to Jelena

TI__Mastermind 19970 points

Jelena,

9 or 10 time constants is generic.

You want the voltage to be sampled to 1/2 LSB which is 1/8192 for a 12-bit ADC.

Since the RC charging is exponential, you need exp(-9.01) to get to a 1/2 LSB level.

Kevin

0 Jeff Panos over 2 years ago

Intellectual 465 points

Hi Kevin,

Thank you for your response.

Can you also address question #2 from Jelena?

Thanks,

Jeff

+1 Kevin Lavery over 2 years ago in reply to Jeff Panos

TI__Mastermind 19970 points

Jeff, Jelena,

It is absolutely true that averaging will keep re-sampling the same voltage many times and therefore create a long sampling time; however, the first (or first few) conversions are not improved by this method.

Kevin

Processors

Processors forum

TDA4VM: Recommendation for SAMPLEDELAY to charge the AFE capacitance during the acquisition period