AD1583: Temperature coefficient out of the specified values

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

Ten samples were tested in following configuration:
+5V regulated supply decoupled with 100uF
output loaded with 1.5kOhm and 1uF
As clearly seen from the attached charts something really bad happens with the
part when temperature falls below zero. TC-measured is, on the average, two
times bigger then specified (50ppm/deg). The parts tested were only samples,
but the question is if the observed and measured temperature properties were
only incident or this is the real nature of AD1583BRT. Regularity of the
measurements is terrifying. Customer ordered a reel (3000pcs) after preliminary
testing in positive temperatures. However, the performance of the part in
negative temperatures seems to be the reason of the quality claim.


From the datasheet you can see that the tempco of Vout is non-linear. The
typical curves given in TPC3 and Figure 4 both indicate that the slope of the
TC is likely to be steeper for temperatures between –40degC and +20degC and
that this will vary significantly from device to device. The comments given on
page 10 under Voltage Output Non-linearity Vs. temperature confirm that the TC
will have a shape similar to that seen in Aplisens measurements.

The absolute Vout of the AD1583BRT is 3V +/-28mV over the temperature range
+125degC to –40 and typically +/-12mV over the same temperature range. The load
regulation adds a further +/-0.5mV (2mA load 0.25mV/mA load regulation) of
error and the line regulation adds a further 25uV/V.  Looking at the data, the
absolute measured value is in spec at all times.

We optimize and trim the AD1583 in production to maximise TC flatness over the
entire temperature range –40 to +125degC. If the TC is calculated over a
smaller temperature range it is possible to see a higher TC for that
temperature range.

However, even taking into account that ADI optimizes TC over a 165 degC span
and Aplisens have measured TC over a 100degC span, the TC still excessive.

The excess drift could be due to the AD1583 not meeting spec, the test circuit
or the measurement equipment.

To proceed, I suggest the following:
1) Isolate and quarantine all tested AD1583BRT devices in case failure analysis
is required.
2) Review the test circuit.
· Check the TC of the 1.5k load resistor.
· The 1uF load capacitor is required and check it is rated for operation below
· Check ripple on the power supply.
· The 100uF decoupling will give good low frequency decoupling, but almost no
high frequency decoupling. It should be paralleled with a 1uF or 0.1uF chip
· Consider any possibility of EMI/RFI coupling into the VIN or VOUT pins,
heaters and coolers can radiate significant EMI.
3) Review measurement setup accuracy and ensure the measurement instrument is
not affected by temperature.