boji
Well-known member
Below is an excerpt from an email with Mr. Hardy (used with permission) on isolating outputs. I appreciated the rather simplified way he explained it, so I thought I'd share it here for sake of posterity:
"An "output isolator" (the Jensen Transformers JT-OLI-3) is a combination of a 39.2 ohm resistor and a 3.7uH inductor, in parallel with each other. The output isolator that is connected from the output of the 990 to the high side of the primary of the output transformer "isolates" the 990 from the effects of cable capacitance when the 990 has to deal with the outside world and the long cables that can be found there. An audio cable happens to also be a capacitor, since you have capacitance any time you have two conductors separated by an insulator.
From a schematic standpoint, you can simply draw a capacitor from the output of the 990 to ground to show the effects of cable capacitance. It is more complicated than that, but that is the essence of the situation. The longer the cable, the higher the cable capacitance.
The capacitance will cause some of the highest frequencies of the output signal of the 990 to be shunted to ground instead of traveling around the feedback loop to the inverting input. If the capacitance is high enough, there will be enough of the high frequencies of the audio signal shunted to ground that the 990 (or any op-amp) will oscillate because of the loss of proper feedback around the feedback loop.
You will typically see a low-value resistor (20 ohms, 50 ohms, 100 ohms, whatever) connected from the output of an op-amp to an output connector when the op-amp is used as an output stage that has to deal directly with the outside world. This is the simplest and cheapest way to isolate the output of the op-amp from cable capacitance. But the downside of the cheap method is that the output impedance of this output stage is increased at all frequencies by the value of the resistor.
It would be better if you could have an isolation circuit that would have zero ohms in the audio bandwidth, but a higher value, such as 39.2 ohms, at frequencies above the audio bandwidth. This is where the 3.7uH inductor comes in. An inductor has a lower impedance at low frequencies, a higher impedance at higher frequencies. So the "output isolator" that I use (the Jensen Transformers JT-OLI-3) looks like zero ohms from DC to around 200 kHz, and 39.2 ohms above 200 kHz. There is a broad transition area between 0 ohms and 39.2 ohms, but that is the general idea.
The second output isolator that I use in my M-1, M-2 and Jensen Twin Servo 990 mic preamps provides the same impedance characteristics for the low-side of the audio signal, so there will be a balanced impedance at the high side and the low side. That is another lengthy issue. A balanced output impedance (basically the same impedance at pin #2 and pin #3 of an output XLR) will provide the best conditions for the highest common mode rejection at the "differential input" of the next piece of equipment in the signal path."
"An "output isolator" (the Jensen Transformers JT-OLI-3) is a combination of a 39.2 ohm resistor and a 3.7uH inductor, in parallel with each other. The output isolator that is connected from the output of the 990 to the high side of the primary of the output transformer "isolates" the 990 from the effects of cable capacitance when the 990 has to deal with the outside world and the long cables that can be found there. An audio cable happens to also be a capacitor, since you have capacitance any time you have two conductors separated by an insulator.
From a schematic standpoint, you can simply draw a capacitor from the output of the 990 to ground to show the effects of cable capacitance. It is more complicated than that, but that is the essence of the situation. The longer the cable, the higher the cable capacitance.
The capacitance will cause some of the highest frequencies of the output signal of the 990 to be shunted to ground instead of traveling around the feedback loop to the inverting input. If the capacitance is high enough, there will be enough of the high frequencies of the audio signal shunted to ground that the 990 (or any op-amp) will oscillate because of the loss of proper feedback around the feedback loop.
You will typically see a low-value resistor (20 ohms, 50 ohms, 100 ohms, whatever) connected from the output of an op-amp to an output connector when the op-amp is used as an output stage that has to deal directly with the outside world. This is the simplest and cheapest way to isolate the output of the op-amp from cable capacitance. But the downside of the cheap method is that the output impedance of this output stage is increased at all frequencies by the value of the resistor.
It would be better if you could have an isolation circuit that would have zero ohms in the audio bandwidth, but a higher value, such as 39.2 ohms, at frequencies above the audio bandwidth. This is where the 3.7uH inductor comes in. An inductor has a lower impedance at low frequencies, a higher impedance at higher frequencies. So the "output isolator" that I use (the Jensen Transformers JT-OLI-3) looks like zero ohms from DC to around 200 kHz, and 39.2 ohms above 200 kHz. There is a broad transition area between 0 ohms and 39.2 ohms, but that is the general idea.
The second output isolator that I use in my M-1, M-2 and Jensen Twin Servo 990 mic preamps provides the same impedance characteristics for the low-side of the audio signal, so there will be a balanced impedance at the high side and the low side. That is another lengthy issue. A balanced output impedance (basically the same impedance at pin #2 and pin #3 of an output XLR) will provide the best conditions for the highest common mode rejection at the "differential input" of the next piece of equipment in the signal path."
