Post Go back to editing

Is isolation my solution? Advice requested

I am designing and testing a measurement circuit that includes two Analog Devices ICs that fail over time.  What I mean by this is that the circuit works as intended, but after a few days of use some of the ICs fail.  Simply replacing these failed components solves the issue, but only for a day or two.  I suspect there may be current or voltage spikes causing the failure (simply because I can't think of anything else that would lead to failure), so I am seeking advice on how to isolate these components. Perhaps the spikes occur when I connect/disconnect power, but I am not sure.  The ICs that are failing are AD5504 DACs and ADG725 multiplexers.  All other components and ICs in the circuit work as intended without any issue.

The entire circuit is powered by USB and is being tested on a breadboard.  All components share ground and a 5V (measured at 4.6-4.7 V) supply from the USB.  A boost converter supplies a 60 V reference to the AD5504s (there are 4 in the circuit). Each output of each DAC is connected to shut resistors.  Either side of each shunt resistor is connected to an ADG725 multiplexer.  A microcontroller selects the output state (ground-60 V, or float) for each DAC output.  The microcontroller also signals the multiplexer to select a shunt resistor for a current measurement.  The voltage drop across the selected shunt is amplified by an instrumentation amplifier that feeds into the ADC of the microcontroller.  The voltage drop and known shunt resistance is used to calculate current which is the purpose of this measurement circuit.  These current measurements are then passed from the microcontroller to a computer via the USB connection.

Outside of testing, this measurement circuit will be used to measure current passing through sections of resistor networks with known layouts but with unknown resistor values.  The total resistance of any particular section of the resistor network is expected to be in the mega ohm range so micro amps of current are expected to pass through the shunt resistors resulting in mV drops for amplification (the ADC is looking for 0-5V).

An example of how I am testing the circuit is provided.  In the simplified diagram the measurement circuit is connected to a resistor network that has a known layout and known resistor values (only 1 DAC is shown).  As an example I may have DAC output A at ground, output B floating, output C at say 30 V, and output D unconnected.  In this case I would measure the voltage drop across shunt A. I may then switch output A to float and output B to ground and then measure the voltage drop across shunt B.  For a more complicated resistor network, all 16 DAC outputs would be set to float except two. I would then sequentially switch each output between float and a biased state to analyze the entire network.

A second problem which may be related to isolation involves the DACs.  For example (looking at the diagram with known network resistor values) if I have DAC output A at ground, outputs B and D unconnected and output C at say 30 V, I will get accurate and consistent current measurements from shunt A. However, if I have DAC output B connected and floating, the current measurement on shunt A will not be accurate or consistent.

Any insight that can be provided in regards to why the DACs and multiplexer fail or what can be tested to discover the cause would be helpful.  If creating some kind of isolation for these components is the best solution, any advice on how to proceed in this direction is appreciated.