Hello Matt
I have made some measurements with the QA472 and I have noticed that the PRE1 output has a DC offset of roughly 130mV at +20dB input gain setting when the XLR input with active phantom power is selected. The offset scales with the gain setting. The offset is negligible if the phantom power is switched off or if the balanced BNC input is selected. It seems that this is due to leakage of the phantom power input coupling capacitors. I was wondering if the input circuit is the same or similar as shown in the INA849 datasheet (section 9.2.2)?
Thank you and best regards
Andy
With a 48V bias you’d definitely expect leakage on that sort of scale for electrolytics, a few microamps. Audio signals are expected to be AC coupled at some point, so 0.13V isn’t a normally an issue.
Hi Mark
Thank you for the feedback. I’m evaluating a similar circuit for a design where a DC coupled second stage adds another +30 or +40dB gain. I use a servo to keep offset under control so the amount of offset from the first stage matter somewhat. But I agree this is nothing which can’t be managed.
Beat regards
Andy
Hi @Avo, yes, I think the 9.2.2 schematic is pretty typical. The phantom power, of course, needs to be applied on the left side of the caps, and after the caps you have the R4 and R3 so the source has something to drive into, and R6 is there to help with source impedance imbalance. And of course the steering diodes.
the QA472 has much higher input Z than typical mic-preamps. That is largely because I don’t like the idea of paying $500 for a reference mic with factory supplied calibration, and then losing that calibration in the interaction between higher mic output Z and low pre input Z.
Now, the INA849 has 1/500th of the bias current of older gen mic pre’s like the SSM2019. So, that points to a higher input Z being supported too. But, the higher input Z at high gains will have other tradeoffs. Even though the INA849 bias is super low compared to the SSM2019, at very high gains and higher input Z it matters. And the phantom leakage is another place the interaction with the higher input Z matters.
the QA472 input caps are 10u/100V. The input Z is about 25k/leg (50k diff). 200mV of output offset at G=100X suggests 2mV input referenced, thus about 10uA current imbalance leaking through the caps. I think leakage current roughly scales linearly with cap voltage. So a 200V cap would leak about twice as much as a 100V cap. And a 50V cap leak half as much.