Thank you for your response. I send you an e-mail instead of using the
EngineerZone of the webpage because I have attached some documents.
The response to your inquiries is described below.
1) Description of the problem:
As I commented in the previous post, we are having exactly the same
problem reported in the following
What happens to us is that from time to time we obtain strange data in
all channels from the ADC converter. It appears to work well during a
certain time (it can be 5 seconds, 10 seconds, 20 seconds,...), but
suddenly all the channels show up very strange and repeatable values.
The two most repeated combinations of values are the following ones:
a) All channels take the same value: 0011 1111 1111.
b) Some channels: 0011 0000 0000 and the remaining ones: 0000 1111 1111.
It seems that as it is a successive approximation ADC, for any reason
one of the most significant bits is set wrongly to 1 or 0, and then the
remaining bits "try to compensate the error".
We detect this malfunctioning because some of the obtained values are
outside of the allowed range, then the system stops (it is programmed in
this way), and we can consult the last values obtained from the ADC
before stopping. Probably, the problem also happens with less
significant bits and we don't realize because the obtained values are
within the allowed range.
2) Details of the set-up:
First of all, I would like to remark that in a first prototype we built
1 year ago, we already had the same problem that we are having right
know in the current second prototype.
In the first prototype, we used a FPGA-based DE0 Altera educational
board to control the AD7658 with serial interface. That prototype was
built upon an bakelite hole board, so we didn't have digital and analog
ground planes. We experienced the mentioned problem, but we thought that
with a new better designed PCB-based prototype, this problem will
disappear. However, we are still having the same problem with the new
In the current second prototype there are some differences compared to
the first one: 1) It is built in a PCB board with analog and digital
ground planes. 2) We are using the new version of the AD converter
(AD7658-1). 3) We control the AD7658-1 with parallel interface
(previously was serial).
In both designs:1) the reference voltage is 5 V. 2) we are using the
recommended decoupling capacitors (in the new design with AD7658-1 is
actually oversized). 3) Five channels are used (the remaining one is
connected to analog ground).
In the second prototype, we control the AD7658 using the signals
/start/, /busy/, /ncs /and /nrd /as shown in the figure below. Signal
/cad_dades/ represents the output data of the ADC converter.
Please see the documents attached for further details about the set-up.
3) Documents attached:
Please find attached the following documents (all the documents
correspond to the second prototype):
1) Schematic: The whole schematic of the board. ADC circuit is on page
2. Power supplies are on page 11.
2) Layout 1: Overview.
3) Layout 2: A little bit zoomed in (ADC + FPGA + Power Supply of the ADC)
4) Layout 3: More zoomed in (ADC).
5) Layout 4: Same zoom as Layout 3, but better to distinguish analog and
digital ground planes. The planes are in an inner layer and are
separated by the thick green line. They are shorted through the pads
that can be seen below the words "digital ground" and "analog ground"
using small wire (see picture 2).
6) Picture 1: Overview.
7) Picture 2: FPGA.
If you have any question, please do not hesitate to contact me.
We would be very grateful if you can help us.
On 17/02/2017 3:05, jcolao wrote:
Re: AD7656 occasional bad data 2
reply from jcolao
/Precision ADCs/ - View the full discussion
There are times when the problem manifestation are the same but the
source of the issues are different. As you mentioned you have shorted
both analog and digital ground but did not resolved the issue. The
AD7656 has supply and reference pins. Please make sure that all the
recommended capacitors in the supply, reference and refcap pins are
present. May I know some details of your set up. What is your
reference voltage? If it is okay for you to share your schematic or
layout so we can have a better understanding of the problem. How does
the bad data look like? if there is a plot of the output code of good
and bad data then it will be helpful.
Reply to this message by replying to this email, or go to the message
Start a new discussion in Precision ADCs by email
or at EngineerZone
Following Re: AD7656 occasional bad data 2
in these streams: Inbox
This email was sent by EngineerZone because you are a registered user.
You may unsubscribe
instantly from EngineerZone, or adjust email frequency in your email
Dr. Joan Nicolás Apruzzese
Dept. d'Enginyeria Electrònica
Escola Tècnica Superior d'Enginyeria Industrial de Barcelona
Universitat Politècnica de Catalunya
Av. Diagonal, 647, planta 9
Telf.: (+34) 934017152 /605936199