OPA462: Input Bias and Powerpad query

Hi Robert,

The datasheets for OPA445 states on Pg. 12, no. 1, "The PowerPAD must be connected to the most negative supply voltage on the device, V−" and for OPA462 it states on Pg. 28, no. 1, "Connect the PowerPAD to the most negative supply voltage on the device, V–." Bonding the PowerPAD to the most negative supply voltage assures that the internal junction isolation between devices functions as designed. If the PowerPAD is allowed to float the op amp may still function, but unanticipated leakage currents can flow and the dc and ac performances can differ from that published in their Electrical Characteristics tables.

Additionally, the PowerPAD to PC board copper serves as the primary thermal path for conducting heat out of these HV power op amps. If there is significant heat being generated internal to the op amp and that heat isn't being removed the die becomes hot. That affects where the electrical parameters operate; especially, DC precision parameters such as voltage offset and input bias currents. The OPA445 is a JFET input op amp and the OPA462 is a CMOS input op amp. Both technologies exhibit significantly higher input bias current at high temperatures compared to low temperatures. The datasheet Typical Characteristics include some curves that provide an indication of a parameter's performance over a wide temperature range.

Regards, Thomas

Precision Amplifiers Applications Engineering

Hi Thomas

Thanks for that advice. It took me a while to verify that there was no existing link to VEE. Then wire a link there with a new device (the existing part was removed in one piece but just to be sure re any damage whilst desoldering) and refit - a lot easier said than done with some 30AWG wire under the chip.

So after all that the 'problem' persists. Basically the output sticks at / near the positive rail with signal source connected via two cascaded High Impedance RC filter sections. tbh I had missed that it was saturating to the positive rail as the reading to chassis ground / earth is approx -635V. But the op amp is running on rails of around +/-66V referenced to -700V. This -700V rail is the input to the filter stage such that the input to the opamp is a filtered version of it. Note that the capacitor legs of the RC filter stages are returned to Chassis Ground / Earth.

When the Filter stage is bypassed with a resistor then it works as expected. The resistor value is similar to the DC path value of the filter stage. And even higher values result in the correct output although there is a delay associated with it eg 500M climbs to the value over a (guessing) 5 to 10 seconds.

Clearly (?) this points to the filter and associated time constants and an issue with input bias current ?

I'll raise a support ticket on this as there are some details that I can't post publicly and the situation might not be clear from the text here.

Thanks

Robert

Hi Robert,

It sure sounds like the issue is the result of the OPA462 input bias current flowing through a very high resistance path in that input filter circuit.

Although adding the #30 wire from the power pad, to the V- supply level didn't affect the output behavior I wonder if the problem is being enhanced by the lack of a solid PowerPAD to PCB copper thermal connection? A high thermal resistance environment for the OPA462 will force it to operate at a much higher junction temperature. If you have some freeze spray available try giving the OPA462 a shot and see if its output moves off the positive rail. Cooling the device should momentarily reduce the op amp's die temperature and cause the bias currents to be significantly reduced.

Let's see if that makes a difference. If it doesn't then we can set up a secure way via email to get your schematic to our App's team to review.

Regards, Thomas

Precision Amplifiers Applications Engineering

Hi Thomas. Apologies for leaving this 'hanging'. After a bit more poking around it became clear that I didn't have a proper dc path at all ! Some connection is broken in the input filter. Can't see where as it's potted on this unit. So adding any sort of dc path restores operation to a degree even if it is a very high resistance. I guess previously the BAS16 diodes had enough leakage for the input bias current but that's not the case with the low leakage (but slower) BAS116. So I've replicated in values the filter externally and it 'works' with that. Still some possible issues with the circuit operation. So I'll get back on this forum if I have further questions. Many thanks for your input on this.