In the LT1028 datasheet there are figures with spectral densities of input voltage noise and input current noise.Unfortunately it is not mentioned if the current noise spectral density graph shows the differential noise current or the common mode noise current.
From the graphs showing "Total Noise vs Matched Source Resistance" and "Total Noise vs Unmatched Source Resistance" I estimate that the current noise spectrum shown is the differential current noise. I have tried to derive a fe values from these graphs and my results are: At 10Hz: 4.5pA differential and 12.2pA common mode. At 1kHz: 1.5pA differential and 3.25pA common mode. As 10Hz seems well below the 1/f corner frequency with both noise currents, I estimate for 1Hz: 14.2pA differential and 38pA common mode.
Can you confirm these values or do you have graphs of measured performance for the differential current noise down to 1Hz (or even 0.1Hz) and for the common mode noise current also?
With balanced source impedance, the differential (incoherent) current noise is measured, as the common mode (coherent) is cancelled. At 14kHz I found 1pA/rtHz incoherent noise (balanced source) doing the Erdi calculation of dividing by 2Rs. Common mode noise was measured at 2.5x that.
At 10Hz, I have measured 4.3pA/rtHz incoherent and 15.2pA/rtHz common mode. (Note that the unbalanced circuit mesures both noises, so the incoherent must be subtracted to arrive at the pure common mode component.)
So our numbers are pretty close.
Hope this helps. Please also have a look at LT6018. It has practically identical current noise as LT1028, but slightly better voltage noise in the 1/f region as well as no noise bump at 300kHz.
Thank you very much for your answer and for the measured values.
While the LT6018 seems to have almost the same unbalanced current noise, it has a significantly better balanced current noise. Unfortunately there seems to be no decompensated version of the 6018.
I wonder why the unbalanced noise is so much higher than the balanced one. Is the tail current of the difference amplifier that noisy? Or is it the input bias current compensation circuitry?
Having to add another noisy resistor in order to get rid of the unbalanced noise is not what you would wish for. Especially if adding a balancing resistor would get rid of the bias current related offset, which is targeted by the noisy cancellation circuitry, anyway. So in my opinion a noisy bias current cancellation circuit does not really make sense (at least if the resistor noise is not significantly lower than the balanced current noise).