Good shout! The lower 1/f noise corners is a great plus. And the midband voltage noise is a useful 4dB lower. And much lower distortion.
It seems that finally there is a device to take on, and beat, the mighty but ancient NE5534 in MM phono stages.
OPA2210 is surface mount only, but that is where the industry is going.
OPA210 looks better for low noise than the old obsolete (but good performing) AD743 & AD745 BiFETs.
Its a super-beta bipolar input, lower voltage noise than any FET opamp but of course much higher current noise.
There's something fishy going on with the OPAx210 datasheet, they quote the current noise as 0.4pA/√Hz yet the noise graph in fig 7.2 shows a current noise slope consistent with 1.6pA/√Hz, 4 times worse than the NE5534A... The lowest noise seems to be at roughly 1.2k source impedance, consistent with the current noise being 2.2nV/1200 = 1.8pA.
This suggests the NE5534A is actually quieter than the OPAx210 for a MM phono preamp.
This is unsurprising really if you realize they probably have some complex bias current cancelling scheme on the OPAx210 inputs, which will not help with current noise unless both inputs see the same impedance (not the case in most phono preamp designs). Perhaps 0.4pA is the matched-impedance noise, 1.6pA the unmatched? The circuit shown for fig 7.2 is completely unbalanced impedance.
This suggests the NE5534A is actually quieter than the OPAx210 for a MM phono preamp.
This is unsurprising really if you realize they probably have some complex bias current cancelling scheme on the OPAx210 inputs, which will not help with current noise unless both inputs see the same impedance (not the case in most phono preamp designs). Perhaps 0.4pA is the matched-impedance noise, 1.6pA the unmatched? The circuit shown for fig 7.2 is completely unbalanced impedance.
Last edited:
The datasheet says 400fA/rootHz. And the graph line says 0.4pA/rootHz - which is identical. And identical to that for the NE5534.
From the (correct) equation for E0 in figure 7.2 I calculate that the point of minimum distance between between E0 and the resistor line to be very close to 1k-ohm. That suggests that the voltage noise under those circumstances is 4.64nV/rootHz. Which as close as you can get from the graph line is the correct value.
There is no subterfuge or inaccuracy.
From the (correct) equation for E0 in figure 7.2 I calculate that the point of minimum distance between between E0 and the resistor line to be very close to 1k-ohm. That suggests that the voltage noise under those circumstances is 4.64nV/rootHz. Which as close as you can get from the graph line is the correct value.
There is no subterfuge or inaccuracy.
The datasheet I found shows 180nV/√Hz at 100k source impedace, 1.6µV/√Hz at 1M source impedance, in fig 7.2 - perhaps they copied the wrong graph?
By the way perhaps I should explain - the point on the total noise v. source impedance graph that is closest to the resistance line is where the noise impedance of the device is, ie the ratio of v_n / i_n. For a device with 2.2nV and 0.4pA noise densities, the noise impedance is 5.5k, and the closest approach of that graph to the resistance-only noise line would be at about 5.5k. Figure 7.2 doesn't agree with that.
Note that this is the only place in the datasheet where the current noise is given for an explicit circuit, rather than under unspecified conditions. Its a known problem that bias current cancellation circuitry has its lowest current noise when both inputs see a matched source impedance. The NE5534 has no bias cancellation, hence the 500nA typical bias current (which matches the 0.4pA noise density as the shot-noise for 500nA is 0.4pA/√Hz)
The +/- 0.3nA bias current specification for the OPAx210 clearly indicates cancellation circuitry, as simple bipolar inputs only have one polarity of bias.
Note that this is the only place in the datasheet where the current noise is given for an explicit circuit, rather than under unspecified conditions. Its a known problem that bias current cancellation circuitry has its lowest current noise when both inputs see a matched source impedance. The NE5534 has no bias cancellation, hence the 500nA typical bias current (which matches the 0.4pA noise density as the shot-noise for 500nA is 0.4pA/√Hz)
The +/- 0.3nA bias current specification for the OPAx210 clearly indicates cancellation circuitry, as simple bipolar inputs only have one polarity of bias.
So with an input current noise spec of 1.6pA/rtHz the OPAx210 would have an input bias current of 2uA, absent the bias correction circuitry. Definitely not the current gain for an input pair I'd expect if it truly were a super-beta IPS.
Huh? We're trying to characterize it from the data available, which seems contradictory or incomplete/open to several interpretations.
Opa2210 / 11 look to be an excellent device.
Should work well in a 5534 or 5532 circuit with adapter board. Keeping in mind that 5532 is an excellent part and the difference is inaudible.
Should work well in a 5534 or 5532 circuit with adapter board. Keeping in mind that 5532 is an excellent part and the difference is inaudible.
You'll find potential for inconsistency in most opamps or active device data if you dig hard enough.