JR. wrote: ↑Tue May 15, 2018 9:16 am
I apparently missed this the first time around.
One question.... at max gain, how much is the current density of the two input devices shifted apart? This might affect HF CMRR but this is picking a very tiny nit.
Nice circuit.
JR
Thanks JR. I missed your post or would have responded sooner.
The worst pair I have, at full gain, produces about 500 µA delta Ic (Or ±250 µA per device from a nominal 5.5mA.)
PSRR vs. Servo
Curiously what I found was that this pair, when pulled into correction, had reduced PSSR.
Running the preamp on batteries verified that it was PSRR, not CMRR.
The first place to look is obviously the collector loads so I built a cap multiplier for them.
That did not solve it: The ingress was actually into the reference voltage for the bias/servo bridge. (R13-R18.)
As the servo corrects the bridge gets pulled.
At this point I introduced the TL431 shunt regulator for a low noise +11V reference.
I chose the '431 over a cap multiplier because I wanted both input PSSR and low output impedance.
The '431 solved the PSSR vs. servo problem and gives me an accurate +11V reference with a low dynamic impedance.
The 11V supply also eliminates the LED string used by Cohen
et al to limit the common mode range seen by the 5532.
CMRR
I've looked at CM rejection and it seems pretty good.
It's difficult to take any measurement on a Protoboard too seriously.
In checking CMRR on a Protoboard you normally have two things working against you: Capacitance and sub-Ohm contact resistance.
Its difficult enough testing a modestly-high impedance line receiver until its on a PC board. (18KΩ for a THAT124X.)
The MC preamp has a common mode input impedance of 250Ω.
A tenth of an Ohm makes a huge difference so I'll need boards to make any meaningful CMRR measurements.
ZTX851 as a Mic Preamp
I made some quick noise measurements at 100 and 200Ω source impedances with 1K5Ω bias resistors and gains of about 20 and 56 dB.
The high gain NF were excellent.
At low gain, 20 dB, there is 6 dB gain in the cross-coupled output with high-ish value feedback resistors.
The actual preamp gain is 14 dB.
The output stage noise dominates as expected.
Using the basic circuit as a mic preamp I would consider lowering Ic to maybe 1 mA.
The object is to reduce base current noise for the higher source impedance where it matters more than for an MC cart.
There may not be much improvement but its worth trying.
The other thing I would do for a mic preamp using this front-end would be to make the common mode rejection stages use 2K resistors to lower low gain noise.
For a MC preamp enough front-end gain is required that the output stage noise doesn't dominate and the precision of line receiver common mode rejection stages are preferred.
I did have a look at the ZTX951 datasheet and the Cob is almost double the ZTX851 at 74 pF.