Analog to Digital Conversion Process

Thread Starter

helmi58

Joined Feb 22, 2016
16
Hi! If I would like to discretize a noisy analog sine wave signal with Vpp = 20 mV and 28 kHz (as attached) which way should do it for better result? What are the important considerations? This signal is received from an electret hydrophone located about 10 cm from the source. This signal have to be discretized before processing in 3.3V operating FPGA.

Method 1
1. Amplify
2. Filter
3. ADC

OR

Method 2
1. Filter
2. Amplify
3. ADC

Which one better?
What type of ADC is suitable?

H1 Time 28kHz 18V.png
 

cmartinez

Joined Jan 17, 2007
7,177
Hi! If I would like to discretize a noisy analog sine wave signal with Vpp = 20 mV and 28 kHz (as attached) which way should do it for better result? What are the important considerations? This signal is received from an electret hydrophone located about 10 cm from the source. This signal have to be discretized before processing in 3.3V operating FPGA.

Method 1
1. Amplify
2. Filter
3. ADC

OR

Method 2
1. Filter
2. Amplify
3. ADC

Which one better?
What type of ADC is suitable?

View attachment 106226
How much resolution do you want? 8-bit, 16-bit?
 

bertus

Joined Apr 5, 2008
20,893
Hello,

It depends.
Amplifiers can also introduce noise.
I have attached a PDF with some filter design topics.
I also have attached a PDF on opamp basics.
And a PDF on filters in combination with ADC's.

Bertus
 

Attachments

Thread Starter

helmi58

Joined Feb 22, 2016
16
Hello,

It depends.
Amplifiers can also introduce noise.
I have attached a PDF with some filter design topics.
I also have attached a PDF on opamp basics.

Bertus
It depends on what? Any factors to be considered? I'm considering using AD743 op-amp, with noise of 2.9 nV/√Hz at 10 kHz.
 

Thread Starter

helmi58

Joined Feb 22, 2016
16
Hello,

It depends on the used components.
Resistors and opamps can introduce noise.
The choise you made for the opamp is a good one.
You might also want to have a look at the LT1028:
http://www.linear.com/product/LT1028
For the resistors use metalfilm and not carbon resistors.

Bertus
Thank you for your valuable feedback. Should i split the gain for filter and amplifier or filter with unity gain?
 

bertus

Joined Apr 5, 2008
20,893
Hello,

There are filter circuits that will allow amplification, but amplification will be limited.
Also what kind of filter you had in mind.
For an ADC most times a lowpass filter is used.
See the PDF's I posted earlier.

Bertus
 

Thread Starter

helmi58

Joined Feb 22, 2016
16
Hello,

There are filter circuits that will allow amplification, but amplification will be limited.
Also what kind of filter you had in mind.
For an ADC most times a lowpass filter is used.
See the PDF's I posted earlier.

Bertus
For now i'm thinking of using bandpass filter with a bandwidth of 1kHz centered at 28kHz. I have two option: (1) Using continuous time active filter (opamp based) and (2) Using switched-capacitor. But I'm not really sure about advantage and disadvantage of each method.
 

Thread Starter

helmi58

Joined Feb 22, 2016
16
By the way, some of the PDF you posted earlier I have read before, just I need further explanation and conformation since this my first time seriously work on analog system and ADC.
 

cmartinez

Joined Jan 17, 2007
7,177
By the way, some of the PDF you posted earlier I have read before, just I need further explanation and conformation since this my first time seriously work on analog system and ADC.
In that case, you have a whole lot of reading to do just yet. PCB design and layout is critical to any "clean" ADC system
 

Papabravo

Joined Feb 24, 2006
14,702
Thank you for your valuable feedback. Should i split the gain for filter and amplifier or filter with unity gain?
This is one of those complicated tradeoffs. I was thinking you might employ a passive filter which normally has an insertion loss. If your total gain is from 20 mVPP to say 5VPP, that is a factor of 250 or 24 dB. Rather than do this in one stage you could have 2 stages at 12 dB or even 3 stages at 8dB. Each stage would be less likely to oscillate or behave badly and you get the required gain. In the three stage arrangement I would put the filter after the first stage.
 

OBW0549

Joined Mar 2, 2015
3,566
This is one of those complicated tradeoffs. I was thinking you might employ a passive filter which normally has an insertion loss. If your total gain is from 20 mVPP to say 5VPP, that is a factor of 250 or 24 dB. Rather than do this in one stage you could have 2 stages at 12 dB or even 3 stages at 8dB. Each stage would be less likely to oscillate or behave badly and you get the required gain. In the three stage arrangement I would put the filter after the first stage.
If I'm not mistaken, a gain of 250 is 44 db, not 24 db. Aside from that quibble, I think your amplify-filter-amplify recommendation is the most practical one. Myself, I'd probably choose an amplifier with a gain of 10, followed by a 30 kHz 2-pole lowpass filter, then another X10 amplifier to feed into the ADC.
 

Papabravo

Joined Feb 24, 2006
14,702
If I'm not mistaken, a gain of 250 is 44 db, not 24 db. Aside from that quibble, I think your amplify-filter-amplify recommendation is the most practical one. Myself, I'd probably choose an amplifier with a gain of 10, followed by a 30 kHz 2-pole lowpass filter, then another X10 amplifier to feed into the ADC.
Actually for voltage gain it should be:

\(20\;\times\;log_{10}(250)\;\approx\;48 \;\text dB\)

Since I deal mostly with power gain I was using a factor of 10 -- my bad.

A voltage gain of 100 will get you from 20 mVPP to 2 VPP, which shold be adequate.
 
Top