Bipolar Photoconductive Isolation Amplifier using Vishay IL300 Strange Issue

[aarpcard]

Hello All,

This is my first post here.  Not sure if this should be in the projects forum or the "beginners" forum.  This may end up being a really stupid, simple question.  Digital circuit design is my background, so analog stuff is not my forte.

A bit of background:
I have an audio mixer connected to a USB soundcard's line out jacks via 25 Foot RCA cables.  As such, I have a pretty serious ground loop problem - fairly loud static and hum.  Not wanting to use audio isolation transformers due to their expense and potential distortion/non-ideal frequency response I decided to use Vishay IL300 Optocouplers to build an isolation amplifiers for each channel.  The goal was to have unity gain, as flat as possible frequency response across the audio range, as minimal as possible distortion, and break the ground loop, so I splurged and got binned IL300's with K3=1.0.

I built the below circuits modified from the examples in the IL300 datasheet (only showing 1 channel):



Configuration 2 is the design I came up with first.  The input and output stages of the amplifier use separate grounds and +/-15V rails generated by two NMH0515SC isolated DC-DC converters.  5VDC and DGND for the DC-DC converters is supplied from a USB port.  The line-in signal and ground is completely isolated from the line-out signal and ground.  This circuit functions, i.e. passes through the audio signal, but does absolutely nothing to eliminate the noise from the ground loop.

Configuration 1 uses a single isolated DC-DC converter to generate +/-15V and GND.  The input and output stages of the amplifier share the same power rails and share the same ground.  Thus line-in and line-out share the same ground.  This configuration completely eliminates the ground loop noise eventhough the line-in and line-out grounds are not isolated from each other.

This is exactly opposite how I expected it to work.  Can anyone explain where I'm mistaken?

[moffy]

It might not be a ground loop, but induced mains noise  in your 25' cables. Try using a cheap car audio isolator to see if its induced noise or a ground loop. Your first configuration doesn't break a ground loop, but might lower the output impedance driving the cable due to the opamp and hence reduce induced noise. It's only a guess.

[Ian.M]

Even if it is a ground loop, you don't need full galvanic isolation to break it.   A differential amplifier with high impedance inputs that have a common mode range extending well outside the supply rails, connected to the signal source and the signal source's ground will be good enough as there is unlikely to be more than a few volts between the signal source ground and your local ground.