Question
I have a problem with AD7811 getting stuck at random values. Also, the chip
gets hot in this situation. Has happened on several boards. After restart it
may work for a while. All mux inputs are connected to GND or circuitry on 5V
supply like the ADC. What else should I look at?
Answer
Here are a few possibilities to look at:
Power Supply Reset Error:
A glitch on the power supply can be long enough such that the ADC control
circuitry power up in an unknown state but short enough such that the Power On
Reset circuit does not detect the glitch and no internal RESET signal is
issued. Such a situation will cause incorrect results to be read from the
AD7811. If you detect that you are in this error state (either continuous
erroneous conversion results or you have an external RESET circuit), you can
return to normal operation by cycling the power down bits in the control
register. This ensures that a RESET signal is issued and returns the AD7811 to
normal operation. We recommend that the AD7811 power down bits are cycled when
ever you power up the system or whenever the system micro is reset. Note that
the AD7811 is not damaged in any way by being in Power Supply Reset Error
condition.
CMOS Latch-up:
CMOS latch-up can occur in just about any CMOS device when any of the analog or
digital IO lines are allowed to exceed the voltage on the power supplies. An
SCR action takes place and high current will flow between the power supply and
ground. The device will generally get hot and will be irreversibly damaged if
the situation is allowed to continue for any length of time. The only way to
get out of this error condition is to remove power from the device. Most ICs
have a certain amount of protection built in to prevent CMOS latch-up and to
protect from ESD events.
AN202 and AN397 give more information on CMOS latch-up and ESD protection of
CMOS ICs (available from www.analog.com). There is a possibility of cumulative
damage to the AD7811 whenever the IO pins are allowed to exceed Vdd+0.3V or
allowed to drop below GND - 0.3V. As a general rule of thumb, provided the
current into the IO pin is limited to less than 5mA, a latch-up will not occur
and the part will not be damaged.
If the problem were due to CMOS latch up, I would not expect that toggling the
PD bits would return the part to normal operation. You would also expect to see
high current and excessive temperature and the device would eventually die
completely.
Power supply glitch:
If Vdd drops into the region of 0.8V to 1.2V and is then powered back up the
AD7811 internal reset circuit will not be triggered and the ADC will output
either it’s last converted value or it’s default value of 1023. Cycling the
power bits guarantees a valid reset.