choosing a suitable ADC for the fully differential amplifier

[KlausST]

Hi,

post#15

if you remember my team leader want to avoide chaining analog amplifiers

If he accepts an ADC driver, then use it. Basically it still is an amplifier, but if a differnt name solves the problem.. ;-)

so I think I will loose the advantages of the anti-aliasing filter for filtering in my analogue readout circuitry

Do I understand this correctly? You use the cutoff of an amplifer as anti-aliasing filter.
I don´t recommend this.

But if you use a true filter then the additional ADC driver won´t harm the anti aliasing function.

An anti aliasing filter is not related to "noise" in first place.
For sure you may call all unwanted frequencies as noise.
I like the definition of noise as "random" frequencies. When we follow this definition then coupled switching "noise" from an SMPS is "not noise" but a known fixed frequency.
The same applies for overtones caused by distortion of your (wanted) signal.

The noise caused by the ADC drier should be negegible.

******

if our analog front ends with amplifier havieng wide bandwidth, what will be the advantague of haveing the filter before.

AN ADC driver has to be fast to drive the ADC input. Especially successive approximation ADCs often use a capacitor at the input. If used as HOLD capacitor, then the input is switched ON/OFF to the capacitor. This causes transients and the ADC driver needs to handle it (low ringing) and needs to set the ADC input precisely to a stable value within a short time.
So the ADC driver needs to detect (feedback) error at it´s output and needs to be fast enough to correct it.
--> for this you need a fast amplifier (much faster than the signal frequency. Indeed it does not depend on signal frequency but on the expected settling time). And it needs to provide low output impedance at high frequencies.
The requirements for the ADC driver are given in the ADC datasheet.
I recommend to read some application notes about driving ADCs.

Klaus

 

[Junus2012]

Dear Klaus,

Your explanation are very useful, and I am reading from the ADC data sheets with the their recommended ADC driver as well and now I have a better understanding about the processing line to the ADC.

That is a nice idea to use a different name and might it will work

I was meaning not to use the amplifier as a filter by utilizing its frequency low pass response, I was meaning using a dedicated anti-aliasing filter.

I found a source that show my argument, it i from Analog Devices, "Op-Amp Applications Handbook", chapter 3. Unfortunately it was not explained much in the text but have good noted on the picture as shown below:

if you see especially the pictures 3-12, it is what I was puzzling me,
The noise integration in first configuration (filter before amplifier) will be integrated to the bandwidth limit of ADC or amplifier depending on which of them is less. Looks for me here I lost the functionality of the filter because the ADC driver or the ADC has almost wider bandwidth.
But if I be back to your comment, I feel relaxed when you say that the filter is essentially are not concerned with the random noise, so if it is before or after, his jop is to filter the signal unwanted frequencies.

Thank you very much
Regards

 

[sutapanaki]

You will have to place the filter as the last stage before the ADC. The ADC doesn't integrate the noise, it folds it, together with other unwanted components like harmonics, interference, etc. that are beyond Fs/2. If you put a broad band noise at the input of the ADC, it will fold down to Fs/2 everything up to the BW limits of the internal sampling network of the ADC. So, having a filter in front, as usual, limits those external unwanted garbage to below Fs/2.

 

[KlausST]

Hi,

the application note is good for really high precision applications.

Did you do some calacualtions?
They speak about noise of 110uV RMS versus 166uV RMS.
in the same path of a 12 bit ADC with an input range of 4V.
The LSB of this ADC equals to about 1mV, or 1000uV.
Even if we assume that the 160uV RMS noise may cause about 166 uV x 6 = 1000uVpp ... we still are just in the area if a single LSB.

In really high performance applications you may see a differnce.

But what I´ve seen and calculated form your circuits .... there are problems with a factor of 100 higher.

If you want to improve your circuit you need to improve the part that causes the highest errors.

Error add using squares:
So if one error is 1% and the other is 0.1% you gaet a total error of 1.005%
So if you take huge effort and try to improve the 0.1% error you barely improve the whole system.

But if you improve the "worst" error then this has about 1:1 improvement on the whole system.

Indeed I also recommend the RC in front of the ADC, but not in first place to reduce noise, but to reduce ringing of the ADC driver. An RC is much easier for the amplifier to drive than the switched S/H capacitor directly.

**********
Don´t get me wrong: I fully agree with the application note.
--> My recommendation: don´t focus on the tiny details. Focus on the big problems.
Power supply, proper decoupling, clean REFerence, PCB layout ...

Klaus

 

[Junus2012]

Thank you Suta nad Lukas,

Your explanation is very clear and answered my question perfectly.

Seems I am going to have an ADC driver, after your explanation and after I read many of the ADC data sheets.

I will tell you additional why I decided besides the reasons you covered.

My prototype chipe is programmable gain FDA. If I am going to test the closed loop gain, this mean that output impedance is going to drop at high frequency and form a voltage with the passive RC filter in front of the ADC which also help as Lukas suggested to better drive RC than input sampling of the ADC,

Since my PGA is already fully differential, is it possible to have single-ended type buffer at each output of my FDA rather than I put fully differential driver?

 

[danadakk]

Front end design :

https://www.analog.com/en/analog-dialogue/articles/practical-filter-design-precision-adcs.html

And some interesting links and references in it.

Some excellent comments on what Nyquist did not say :

https://www.wescottdesign.com/articles/Sampling/sampling.pdf

Good ref handbook :

The Data Conversion Handbook, 2005 | Education | Analog Devices

The Data Conversion Handbook, Edited by Walt Kester, Newnes, 2005, ISBN 0-7506-7841-0. Also published as Analog-Digital Conversion, Analog Devices, Inc. 2004, ISBN 0-916550-27-3.The Data Conversion Handbook is written for design engineers who routinely use data converters and related circuitry...

www.analog.com

Regards, Dana.

 

[Junus2012]

Dear friends,

Thank you all for your contribution and help to make me understand the problem, also the link resources of Dana are really nice, specially the book, i am reading it now