Pro Audio Design Forum

FWIW I didn't see a strong 1/f component in the current noise with the ZTX851 when I went from 10R R source to "open circuit" which was the 499R X2 bias resistors.

The increase was as expected broadband and didn't show a strong corner.

I was more interested in the difference between 10r and 0r between old and new low rbb devices. Unfortunately there are too many moving parts in the dual balanced input to easily isolate just the noise current contribution from a single device..

Probably academic since I suspect vinyl surface noise will swamp out these noise floors, and pico amps times 10 ohms are still picovolts (rt Hz).

JR

Once I'm able to get back to it it will be easy enough to find out.

mediatechnology wrote:Once I'm able to get back to it it will be easy enough to find out.

I guess you could short both inputs to ground...If you just short the bases together the input current noise x500 ohm will be CM. If you look at one op amp at a time there is still some interaction.

Don't waste too much time on this, I'm just curious to see how much of a difference device beta makes.

JR

I had a chance to get back to this project and wanted to share some results.

I'll post a schematic soon but its similar to the OP.

The inputs are a single pair of ZTX851 running at 7.5 mA.

The op amp iseither an NJM2043, NJM2068, NJM2114, NJM5532 and OPA1612.
All op amps were within a dB or two on noise floor; the NJM2043 and NJM2068 were the quietest, the OPA1612 had higher distortion owing to wimpy outputs.

Rg is 2R2, Rfa and Rfb 332 Ohm.
The gain is 50 dB. (For flat applications I'm figuring 30, 40, 50 dB gain in the INA plus 6 dB in the cross-coupled for an overall gain range of 36, 46 and 56 dB will be just about right.)

Rsource is 3R3 with a thermal noise of -147.5 dBu.
Ein on the protoboard is -140 dBu for a NF of 7.5 dB.
There is a little bit of hum contamination due to lack of shielding.

When Rg is decreased to 1R (and the resulting gain to +56 dB) the NF drops to about 5 dB.
This is for a balanced input/balanced out preamp having a 3 dB noise penalty...
Self claims 6.4 dB for his super-duper 3X 2N4403 MC single-ended pre-preamp.

The ZTX851 have such a low rb that ultra-low Rgain resistors are justified.
At 1 Ohm however 50 dB gain requires a very difficult-to-drive 162 Ohm resistor. (Lower gains will use higher Rg; Ra and Rb are fixed.)
Thus Rg was set to 2R2 so that Rfb could be increased to 332R.
With the 6 dB gain in the output the signal swing is reduced and 332R is driveable for all devices I tested except the OPA1612.

If the NF were to include the unavoidable noise contribution from the 2R2 gain resistor, which is -149 dBu, the 50 dB gain example would have a NF of about 5 dB.

So it looks like a couple of ZTX851 paired to an NJM2043, NJM2068 or 5532 will be just about as quiet as Self's Small Signal Design example.

Yup, when you try to keep Rs small and voltages no so small, current ensues... I have recently been learning about SMD resistor problems from high voltage but I am not aware of any high current issues other than the obvious power dissipation. I recall voltage coefficient issues with thru hole resistors too, in high power audio amps with lots of voltage swing.

At some point after signals are up to line level the resistors don't have to be kept very low resistance. Several of the new uber op amps can drive pretty low Z, perhaps for exactly that reason (low noise).

Keep up the good work...

JR

Thanks John.

The only op amps that would drive the 162R feedback resistors well were the NJM2114 (60mA) and the NJM4556 (70mA).

The OPA1612 should be able to drive 45 mA but struggles even at lower peak currents.
This is also true with 332 Ohm Rfb.
The OPA1612 THD measures much lower with 332R and would be competitive but the signature becomes very high order with strong >5th harmonics.

With 332R Rfb none of the NJM op amps, including the 5532, break a sweat and rise significantly above the generator THD.
What they do produce is added 2nd and third.

I'll try to get a clean schematic posted.
What makes the 47 year-old circuit interesting is the ZTX851.
In a post 2SB737/2SD786 era the ZTX851/951 are probably about as good as we can hope for.

I haven't kept up with modern devices but IIRC Brad was messing around with a scary low noise JFET. He ended up making a pretty exotic circuit to wrap around it (apparently his topology had poor PSRR so he designed an uber quiet PS for it.).

I did a quiet PS for my last MM pre, but can't say it was necessary, I did it because I could, and to amuse myself.

JR

Schematic of the Balanced Input Moving Coil Preamp using the ZTX851.


Balanced Input Moving Coil Preamp using the ZTX851

The 50 Ohm balance trim adjusts the differential offset measured from TP1 to TP2.
It is a coarse adjustment.
The remaining DC offset drift is servo-adjusted at the outputs using a TL072 or dual BIFET.
The preamp input IC was tested using the NJM2043, NJM2068, NJM2214, NJM5532 and OPA1612.

With a 56 dB gain and 3R3 source the input voltage noise density is approximately 0.42 nV/√Hz owing to the low rb of the ZTX851 and low Rgain value.

Would using a 1246 (+6dB) allow you to drop the resistances even lower...?

How linear is hop off resistance of the gain trimpot? Probably OK is you don't measure anything.

JR