Yes, you can use the QA401 to drive single-ended or differentially into a transformer, you just need to limit the levels depending on your load impedance. This is true on all test equipment.
Remember the output opamp max output is around 65 mA (round number) and you can find that on the TI data sheet for the OPA1612. TI doesn’t go into great detail on the specs versus the output current, but a reasonable assumption is that the opamp isn’t meeting normal specs when you are asking it to drive that hard. You can get a bit of an idea on the degradation from the data sheet. The plot below is from the OPA1612 data sheet. Look how much the performance of the opamp degrades between a 2K load and 600 ohm load. These are for 3Vrms signals–about 5 mA of drive current. It’s pretty substantial, and the OPA1612 is a stellar modern opamp.
But let’s say we want to limit the opamp currents to +/- 20 mA peak to ensure it’s performance will be solid. And let’s say we want to drive differentially into a 10 ohm load (a near short) across the + and - outputs. And we know the QA401 output impedance is 50 ohms on the + and 50 ohms on the minus. That means a total load of 110 ohms (50 + 50 + 10). If we want to limit peak output to 20 mA, then that means the max voltage (the tip of the sine) will be 0.02 A * 110 = 2.2V peak. Now, since this is differential, that means the high side is driving at 1.1V, and the low side is driving at -1.1V.
And these are peaks, so the RMS is 1.1/1.41 = 780 mV = -2 dBV.
OK, so we know if we drive the 10 ohms load differentially the current the output opamp is dealing with with will be in the +/- 20 mA peak = 14.1mA RMS. The output 50 ohm resistors will be seeing 0.02^2 * 100 = 20 mW, well below the 100 mW rating.
So, if you set the QA401 to drive at -2 dBV, and then connect a 10 ohm resistor across the + and - outputs, the QA401 will happily drive it all day. In short, you need to do a bit of math to make sure you aren’t asking the OPA1612 to drive too hard. Because if you are, you don’t have any way to know if the measurements you are seeing is due to the DUT or due to the output opamp being asked to work too hard.
For measuring high voltage supplies, the reason just a cap won’t work is as follows. Let’s say you insert a 1uF cap in series with the QA401 input, and you pick a cap rated for 500V. With the 100K input impedance of the QA401, you get a high-pass corner of 1.5 Hz. Your circuit looks as follows, where C1 is your added cap and R1 is the QA401 input impedance. The VF1 measurement is the voltage into the QA401.
And then you connect to the 300V power supply at t = 1 second. Below in the transient simulation you see the output presented to the QA401 nearly approaches 300V for 100 mS. And that 300V for 100 mS will kill the QA401. And you’ll get a similar response when you remove the high voltage supply.
What you really need to measure high voltage supplies is a circuit as follows. C1 is your DC blocking cap. The 1K R2 and R3 are current limiting caps. And Z1 is a very, very fast TVS device that will clamp at 3-4 volts (ignore the part number on the schematic).
The thermal requirements on R2 are pretty tough. R2 will need to cope with momentary currents of 350/1k = 350 mA, which means momentary power of 122 watts. So these are big resistors. You can use higher value resistors to help with power, but they will add noise. So, you’d need to consider your supply noise and potentially bump up the 1K to help with dissipation. The output, as you can see below, is clamped at +6V when connected to HV supply, and minus ~1V when removed form HV supply OR when supply is connected backwards on accident.
Again, the above is just a concept and a lot of thought would be needed into part selection to ensure it was safe. I’m not saying the above is safe–it’s just to show that a simple cap won’t accomplish the job.