Hello,
I am trying to determine the cause of an ADM660ARZ failure on one of our boards. I am reaching out because I’ve received two of the same failures over a small number of test units. To my knowledge, I am using the chip in a typical application and am well within its specifications. However, due to my limited analog design experience and some data I recorded I’m hesitant to write it off as damaged or defective.
Application
The ADM660 is configured as a voltage inverter, switching +6.0V to -6.0V. It functions exclusively as the negative supply for a total of 18 op-amps (4 Dual OPA2134, 1 Quad TL074, 3 Dual TL072). The op-amps amplify various small ac signals, including audio in typical inverting configuration. An LT3509 buck regulator supplies the ADM660 and the op amps’ positive supply. This regulator is dual output and additionally provides a 4.0V line to other components.
As yet, the load on the ADM660 never appears to draw more than a consistent 35mA.
Failure Mode
The ADM660 failure was discovered when the op-amps failed to output properly. Sampling the inverting output with a DC voltmeter yielded -0.07V (I’m pretty sure it was nonzero, not just calibration error). There was no visible damage to the package. Replacing only the faulty ADM660 with a new one resolved any issues. In at least 3 weeks of continuous operation, e.g. driving a signal through the op amps, neither of the suspect boards have failed again.
Some of the ADM660s on our factory-produced boards are showing wildly different output voltage ripple, current ripple, and switching frequency. However, I haven’t proven conclusively that they are defective – it could be degradation of the chips during use, a manufacturing process issue on our end, or an operating condition I haven't thought to reproduce. I’ve ordered a sample of pristine chips from the factory and will test them in isolation soon. I also plan to recall our test units from the field and look for signs of wear.
Upon discovering the first failure, we increased C6 and C7 ('C1' and 'C2' in the application note on the datasheet) from 2.2 uF to 10 uF in case the ADM660 isn’t switching as fast as it should. I’m not convinced this is the root cause of the issue.
Schematics
ADM660 In Circuit
LT3509 Dual 700mA switching power supply
Large isolating caps for the entire op-amp power rail (we are not sure these do anything)
Array of 18 op amps in total, fed by the ADM660 output and LT3509 +6V output
Troubleshooting Data
I’ve probed several working ADM660 chips in circuit, to try to find evidence of strain/misuse. I also constructed a test board to isolate it.
In all the data below, the FC pin is shorted to Vin (through a 0 ohm resistor) and C6 = C7 = 10 uF. This was verified with a voltmeter in each instance.
I1 – Current across a 1 Ohm resistor on the supply line to the op-amps. This is not Iout – it sums the current coming from the ADM660 and from C7.
Vin – ADM660 input voltage.
Vout – ADM660 output voltage.
Vout Data
Vout on a factory-supplied ADM660 (Malaysia plant), note the apparent 44kHz switching. FC shorted to Vin, verified with voltmeter.
Vout on ADM660 from Digikey (China plant), switching is 152kHz with better ripple.
Vout of isolated ADM660 (Malaysia) with 180 Ohm resistive load and 33 uF isolator cap (electrolytic). Now the individual switching states are visible. Oscilloscope reads ~280 kHz (2x operating frequency).
Vin Data
Vin on a factory-supplied ADM660 (Malaysia plant)
Vin on ADM660 from Digikey (China plant)
Vin of isolated ADM660 (Malaysia) with 180 Ohm resistive load. Oscilloscope reads ~280 kHz (2x operating frequency).
I1 Data (1 mV = 1 mA)
I1 on a factory-supplied ADM660 (Malaysia plant)
I1 on ADM660 from Digikey (China plant). Note the differing time scale.
I1 of isolated ADM660 (Malaysia) with 180 Ohm resistive load. Oscilloscope reads ~280 kHz (2x operating frequency).
Questions
- Does the circuit I’ve presented appear to strain or misuse the ADM660? Is there an issue with supplying the ADM660 with a buck regulator?
- Aside from probing Vin, Vout, and I1, and isolating the ADM660, do you have any suggestions for troubleshooting?
- Do my findings (failure mode, high output current ripple, inconsistent operating frequency) warrant concern over defective units in my supply of ADM660ARZ?
