We use the AD8279 to measure battery cell voltages in a pack with 4 cells in series. We use the part in gain = 1/2 to get the four cell voltages down to a range we can put into our micro. Each cell is about 3.3V, so Cell4+ which goes to the AD8279 VS+ is about 13.2V.
So the input signals are DC: the inputs to the AD8279 are tied to the cells directly (traces on the board, then leads to the cells). The output goes to a micro:
- we have a clamp so we don't blow up the micro
- we have a cap on that circuit to hold the signal while the micro samples.
Circuit below:
- "Cell1-" (INA- pin2 on U7) is tied to GND
- "Cell2- same as Cell1+", etc
- "Cell4+" is VBAT
- VC1 - VC4 go directly to inputs on the micro for A/D.
We have a problem where randomly U7 (never U8) gets hot (50 degC on the board), and draws lots of current:
- for eg: 150mA into VS+
- for eg: 75mA into (or out of not sure) pin 3.
During the "failure" the output voltages remains correct, and we do not see oscillation on the outputs:
- the scope is 100MHz (Tektronix TDS2012B)
- bandwidth limit is off
- (But - maybe it's really hard to see?)
If we disconnect the leads and reconnect them, the part starts working again. Looking online, we see a couple of possible failure modes:
- latch up: don't really understand this or how it would happen
- oscillations: maybe our effective capacitive load is too high?
We did not include the .1uF decoupling caps suggested in the datasheet near the part. But, we have since added them to our test boards, and the failure still happens the same.
