AD7546: Glitch perfomance

Document created by analog-archivist Employee on Feb 23, 2016
Version 1Show Document
  • View in full screen mode

We have been using your AD7546KN D-A convertor on our circuit
boards. As this has been obsolete for some considerable time,
and as a change in cicumstances has dictated that we now need
to support this product again, we have decided to redesign the
board with a later D-A IC. The original IC was a transparent (no
microprocessor interface) 16 bit monotonic D-A. I have looked
at two likely candidates from your range of 16 bit D-As. One is
the DAC16, which has no interface, and the other is one that we
already use, which is the AD569. Although the latter device does
have a microprocessor interface, my understanding from the data
sheet is that it can be made transparent by holding inputs. Can
you comment on glitch problems with either solution as compared
to the the AD7546? (bearing in mind the way the AD569 switches
its ladder resistor chain). Could you comment on the support for
both devices? Are there any other devices suitable?


A couple of general points on glitch performance.

? Using a DAC with a Latch will generally produce better glitch performance
that driving a DC in transparent mode. When you use a DAC in transparent mode
you rely on all the parallel input signals arriving at the DAC input at the
same time, differences in PCB trace and capacitance can increase the settling
time as the DAC first slews to one code level and then another.

? When using a current output DAC, the settling time and glitch performance is
usually dependant purely on the amplifier used for current to voltage

? A DAC which uses a segmented voltage divider will generally have low glitch
as only two switches are required to operate.  Conversely a DAC which uses an
R2R divider circuit will generally have higher glitch and code dependant glitch
as multiple switches are required to switch on and off for each transition.

? A DAC which uses one current source for each code level can be shown to have
small and code independent glitch.

? The glitch due to switching MSBs is usually far worse than the glitch due to
switching LSBs

? High resolution DAC will usually use a segmented architecture, employing a
different scheme for generating the MSB and LSB. The MSBs will be generated
using a low glitch topology such as the voltage divider or "one current source
per code" scheme. The LSBSs will be generated using a topology which requires
less resistors but may have poorer glitch performance.

DACs to consider in this application are DAC16, AD569, AD7846, AD7849. If you
want to use a single supply serial programmable DAC consider also the AD5541
and AD5542. In terms of a voltage output DAC, I would expect that the AD569
would show lowest glitch performance. For lowest glitch possible, use a current
output DAC and an amplifier with low input capacitance and low input bias/ high
input impedance.

All the above parts are available from AD and there are no plans to obsolete
them. For your reference, We cannot guarantee the lifetime of any of our
components. However, components will only be obsoleted when there is little or
no demand for the product or when the product is superceded by superior design.
When a product is about to be obsoleted, we will give one year's advance notice
and offer all existing customers the opportunity of a last time buy.