The Cyber-Spy.Com Usenet Archive Feeds Directly
From The Open And Publicly Available Newsgroup
This Group And Thousands Of Others Are Available
On Most IS NNTP News Servers On Port 119.
Cyber-Spy.Com Is NOT Responsible For Any Topic,
Opinions Or Content Posted To This Or Any Other
Newsgroup. This Web Archive Of The Newsgroup And
Posts Are For Informational Purposes Only.
From: Chuck Simmons
Organization: You jest.
X-Mailer: Mozilla 4.61 [en] (X11; U; Linux 2.0.33 i586)
Subject: Re: 5V Reference Voltage for AD7707 AD-Converter??
References: <3D7CB0A6.9C1F804E@informatik.tu-chemnitz.de> <3D7D77DE.1C671621@webaccess.net> <3D7E143F.4EBBAD5E@informatik.tu-chemnitz.de>
Date: Wed, 11 Sep 2002 01:23:27 GMT
NNTP-Posting-Date: Tue, 10 Sep 2002 18:23:27 PDT
Mario Trams wrote:
> thank you very much for your advices. Further comments below.
> Chuck Simmons wrote:
> > Mario Trams wrote:
> > > ...
> > > Application notes for the sensor state that the sensor supply voltage
> > > should be used directly as reference voltage for the AD-converter.
> > > This eliminates some noise in the measurement caused by minor changes
> > > in the supply voltage (because the sensor is ratiometric).
> > >
> > > Anyways, the problem is that the data sheet of the AD7707 states
> > > that a reference input of 2.5V shall be used "for specified
> > > performance".
> > > It is left unclear how much we are "off the specified performance"
> > > with a 5V reference (at least, 5V are permitted by the datasheet).
> > >
> > > ...
> > The idea of using the sensor supply as the reference works well with SAR
> > ADCs as long as they are implemented with multiplying DACs. For example,
> > if you need to convert (A-B)/(A+B) with a more conventional ADC, A-B
> > goes to the input and A+B goes to the reference. This is essentially the
> > arithmetic you need with your sensor. You need output/supply which is
> > equivalent.
> > The AD7707 appears to be able to do this but it is not perfectly
> > straight forward. The AD7707 has a single ended full swing of 0 to +2.5
> > on the low level inputs when operated on 5 volts. The limitation on this
> > will be the dynamic range of the internal buffer and the PGA. It would
> > appear that your best approach would be to use a by 2 voltage divider on
> > the sensor output to get within the AD7707 input range. This is at most
> > the resistive divider and two buffers depending on input and output
> > impedances. Similarly, you divide the sensor supply by 2 probably using
> > a single buffer to obtain the reference. All of this assumes that the
> > conditions I'm recommending will end up with the PGA set at a gain of
> > one.
> That's what I thought as well. Regarding the _two_ buffers you
> are they really required? (I'm sorry for this probably stupid question,
> but I'm not _that_ expert in analog issues...)
I don't know the exact specifications of the sensor or the AD7707. For
the AD7707 I went by the block diagram. If your sensor has low output
impedance and the AD7707 has high input inpedance on the direct (low
level) inputs, no buffers are needed there.
> > Another suggestion is to treat the reference as I suggested above and
> > use one of the high level inputs for the sensor and set the PGA so that
> > the attenuation times the gain is 1/2.
> > The idea is that division by 2 is optimum for getting the reference from
> > the supply so you need to use the same division on the sensor output to
> > perform the desired normalization in the ADC itself.
> From an economical point of view using the high-level input of the
> seems to be the best way as the external buffer (or two...) could be
> saved. Only the reference voltage had to be halved from the sensor
> However, the noise of the high-level input appears to be somewhat higher
> than for the low-level inputs.
Since the noise is refered to the inputs, this will be the case. When
you refer the noise back through a voltage divider (the high level
inputs use voltage dividers) the noise will be a larger number. For
example, a divide by two set for a high level input should about double
the noise as quoted in the specification. This is not something you can
do much about.
Going back to the reference. The device may have a specification on
current in the reference input or an impedance. If the reference input
is specified as high impedance, it is internally buffered so you can use
a voltage divider directly. If not, a single opamp with low offset drift
can be used as a non-inverting buffer for the reference. There are
several good ones available from AD, Burr-Brown (TI) and some others.
Since the ADC is up to a 16 bit converter, your LSB is about 76
microvolts with a divide by 2. Certain care should be taken in wiring
for such voltage levels otherwise some of the resolution will be lost.
Thermal voltages at interconnects are a particular problem. Use gold
contacts or like metal contacts in any connectors (gold is to be
preferred). Device sockets must be of the highest quality if used.
Solder joints should be inspected carefully to make sure there are no
cold joints or other partial contact problems. Run a separate ground
wire for the ADC from the sensor so that power supply current is not in
the measurement ground. That should get you close to best performance
from the device.
... The times have been,
That, when the brains were out,
the man would die. ... Macbeth
Chuck Simmons email@example.com
Go Back To The Cyber-Spy.Com
Usenet Web Archive Index Of
The sci.electronics.design Newsgroup