AD4630-16 Channels behaviour

[IssamChek]

Hello Guys, I'm working on a project that involves an ADC from Analog Devices (ad4630-16bits). I noticed that one of the channels is not outputting zero but -2.54volts, even when I use their software (ACE). And also, the signals I'm capturing from the ADC is attenuated when I increase the frequency of my input signal (I suspect that an anti aliasing filter is there. Is there any ideas on how to disable it?). Btw, I'm using a signal generator (Plus to Minus of Channel B, Ground to Plus of Channel B -- Channel A has a weird unstable offset of -2.54V). Can you please explain these to me, probably I'm missing something here!

[iMo]

Schematics?

[IssamChek]

I couldn't understand them fully, as I'm not an electronic engineer, and my experience with ADC circuits is limited.

[Kean]

A few questions, as there isn't enough detail given to help you.

Is this an off the shelf PCB -e.g. an eval board?  Or is it a custom design?
Do you understand the concept of differential inputs and common mode input range?
Is the output of your signal generator floating w.r.t the ADC?
Are you applying negative voltages to the inputs relative to circuit GND?

If you aren't sure of the answers, then just say so as it will still give us clues.

PS. Here is a document explaining the various types of ADC input: https://www.analog.com/media/en/technical-documentation/product-selector-card/2PB_sarinputtypesfb.pdf

[IssamChek]

Thank you very much for your reply.

The ADC is an eval board (AD4630-16FMC) that I'm using right now with a Zed Board.
I don't fully understand the concepts of differential inputs, single-ended and common mode to be honest, I'm aware that the eval board supports these modes.

For the signal generator, I honestly don't know about floating, I'm using a SIGLENT SDG1032X, one BNC cable is coming out of the generator and splitted into red and black wires (+ and - I guess).

For the signal, I'm generating a sinewave with 6volts peak to peak (-3v to 3v) in Channel B. Channel A is empty, didn't connect it because of that offset I noticed before (Not sure about it).

Thanks again for your help Kean.

[iMo]

How is your board configured - the input amplifiers? - https://wiki.analog.com/resources/eval/ad4630-24-eval-board

You cannot feed more than, say, 4Vpp (+/-2Vp against V_CM) into a single channel when referenced to the common voltage (your 2.5V afaik).

With your wiring as you have described you would need the Generator be floating (or isolated grounds - doublecheck) against the board's Ground. Also the second input A (the ground of the Generator) should be kept at the "V_CM" which is 2.5V, afaik.

To make it "true differential" the simplest way is to use a signal transformer, like 1:1 ratio, primary side to the generator, the secondary wired to A and B.

Be careful what you are doing, the boards are pretty expensive, imho.

PS: you cannot proceed here without drawing schematics.. This is how we at EEvblog communicate - via schematics..

[Kean]

OK, take a look at that document I linked https://www.analog.com/media/en/technical-documentation/product-selector-card/2PB_sarinputtypesfb.pdf

Your ADC inputs are differential with wide input common mode, as per fig 4a.  That means each input can only accept voltages between 0 to 5V [1].  As it is a differential pair it can handle IN- being 5V and IN+ being 0V, which is effectively -5V when summed.  To have an ADC reading of 0, both inputs need to be sitting at Vref/2.

Your signal generator is outputting -3V to +3V bipolar single ended signal, which exceeds both the absolute and common-mode input ranges [2].  You should change the signal to have an offset, and be from 0 to 5V centered on 2.5V.  Or better, you might want to limit it to 0.5V to 4.5V, which is +/-2V with a 2.5V offset.  The unused input can be connected to 0V or 2.5V, which will just change the ADC reading offset.

I see mention of different AFE options in the eval board docs, but I didn't dig into that as I suspect you are using the default.

I have no direct experience with the particular ADC or eval board, so I may not be able to help much beyond the above.  Hopefully someone else may chime in and correct me if I got anything wrong.

[1] more precisely, the common-mode input range is −(1/128) × Vref to +(129/128) × Vref, so with Vref of 5.0V the inputs can handle from -39mV to 5.039V.
[2] hopefully you haven't damaged the inputs of the AFE or ADC.

[IssamChek]

Thank you all for your valuable help. Indeed, I'm using the default AFE, but haven't done anything regarding the common-mode input. I was feeding -3 to 3 volts, since the datasheet mentioned the full scale to be -Vref to +Vref, which is -5 to 5 volts (Page 21). What was bothering me, and forgive my lack of understanding, is when I give Channel 1 a sin wave of 0-5v with an offset of 2.5v. Then connect Channel 0 to 0volts, I get -2.5volts as measurements from it. I'm not sure I understand the concepts of SAR ADC with these modes I'm afraid.

Thank you.

[iMo]

..I'm not sure I understand the concepts of SAR ADC with these modes I'm afraid.
Thank you.

With those ADCs (SAR or SD) with 5V Vref and 5V Vcc, and 0V GND, the meaning of input voltages of the ADC chip itself in the differential mode is as follows:

1. signal at A input could be from 0V (gnd) to 5V

2. signal at B input could be from 0V to 5V

3. in diff mode input signals at A and B are (ideally) symmetrically mirrored around the "V_CM - Common Mode Voltage" which is 2.5V (the Middle of Vref)

4. therefore if talking -5V to +5V input range it means for example either A=0V and B=5V (the result is -5V), or A=5V and B=0V (the result is +5V), and that gives you "the ADC input range from -5V to +5V".

ADC_OUT = ( A - B ), full scale range ( Amax - Bmax ) = +/-Vref

The ADC will return the binary coded result which is formatted that way, thus you get the result from -5V to +5V.

So the moral is: you cannot feed into ADC's A or B input a signal larger than 5V and smaller than 0V (GND).

If there is an AFE -> then it depends on the AFE's design/wiring, what could be the input range into the AFE's inputs..

PS: there are couple of exceptions - so called "ADC chips with true bipolar inputs" where the voltage could be negative against GND and larger than Vcc, but the chip you work with (and most of the highend 24..32bitters) are limited to 0..Vcc (unipolar) physical input range as described above..

PPS: FYI - nice pictures with the A and B diff signals:
https://www.planetanalog.com/signal-chain-basics-120-designing-front-end-drivers-for-linear-performance-on-sar-adcs/

[Kean]

From my reading of the AD4630 datasheet it doesn't require symmetrical differential inputs, so it can be used in a single-ended unipolar fashion.
In any case, the AFE on the eval board should also take care of that.

[IssamChek]

Thanks a lot Kean and IMO. Now understand those modes of the SAR ADC. Since the AD4630 also supports single ended as well, then I'll just be extra careful with the min and max of my inputs.

I'm thinking of contacting the Analog devices company regarding the board, because of the other channel, from the very beginning, it has an offset of -2.54V. I tried to put 1V dc for both channels, I got 1V on Channel 1, and -1.54V on Channel 0, knowing that they're identical channels, it raised some concerns about its functionality.

[iMo]

Again, you have to start to work with your board's schematics.
Here it is:

https://www.analog.com/media/en/evaluation-documentation/evaluation-design-files/eval-ad4630-24-design-files.zip

There are 3 variants of your board (ASSY - C or F or I) - it is configurable with the resistors and jumpers, etc.
So the first step is to understand the schematics of your actual board, imho.

[IssamChek]

Thanks IMO. I've decided to spend sometime trying to understand the schematics, while continue using the other channel normally.

Thanks both for your valuable help, have a nice day.

Cheers.