A Low Noise Balanced Input Moving Coil Preamp Using the ZTX851

[Gold]

The new KEF LS50 at $1500 is likely to be good. A bit larger & more $$ than you specified.
http://www.stereophile.com/content/kef- ... KUbUC1O.97

I heard it at the same Hi-Fi show and was very impressed. I loved it. If I was to get something for home that would be it. It's a little larger box than I want for the cabin and it's rear ported. I'd like something I could put against a wall. I'll check out the eggs.

[Hans]

Hi Wayne,

Getting back to the subject of your design, I made a simulation model, with and without A-Weighting, showing the generated noise referred to the input.
I used 2 noiseless transistors and inserted in both a 2 Ohm resistor to simulate the Rbb of the ZTX851
1.
Total input noise is as shown is 57nV at the input, and 34nV when A-weighted.
If you like me to do some simulation with other component values, just let me know.

Hans

Wayne Circuit.jpg (213.65 KiB) Viewed 11815 times

[JR.]

I'm not sure what Ricardo meant by small but I mean a 3" or 4" woofer. Smaller than a studio monitor. I'd be interested to check out his speaker design. The Soundfield is no slouch.

3"-4" seems small for a woofer. I have an old pair of small 2 way Hifi speakers (Fourier), in need of being reconed that use 6" LF drivers. My old (AMR) studio monitors are 3-way with 8" LF drivers (The amr 308 was the speaker that took 1st and 3rd in the blind listening test performed by the recording magazine back in the 90s). I've had to replace the woofers in my 308 once so far. I wouldn't expect to find old soldiers in pristine shape.

There are a lot of choices. Including the ADS speakers I've had for 25 years and like. I've heard a ton of speakers but mostly studio varieties. The PSM M10 looks interesting to me because I liked the designer's larger speaker so much.

NHT was an early adopter of the satellite and sub arrangement which in general I don't like.

I don't have a complaint about subs and satellites.

If your full range speakers only use a 3"-4" LF driver you may need to consider a sub. While not exactly hifi the Auratones were small and popular (no crossover inside is good for the vocal region too, but not much bandwidth).

Sorry about the veer Wayne...

JR

[JR.]

Hi Wayne,

Getting back to the subject of your design, I made a simulation model, with and without A-Weighting, showing the generated noise referred to the input.
I used 2 noiseless transistors and inserted in both a 2 Ohm resistor to simulate the Rbb of the ZTX81.
Total input noise is as shown is 57nV at the input, and 34nV when A-weighted.
If you like me to do some simulation with other component values, just let me know.

Hans

Wayne Circuit.jpg

Real devices have noise current and 1/F noise.

Can you simulate for process related noise? I wouldn't know how much to add.

JR

[Gold]

John,

I'm looking for speaker for my cabin. Not a studio monitor. I don't expect full range out of a 4" woofer. It's how the stuff the speaker can't reproduce is dealt with that makes it a design challenge. I'm set for studio monitors. I use ATC SCM 100's I'm not looking to change those.

I have found it difficult to impossible to place a single sub in a place that produces a convincing presentation. I have a bit of practical experience with this.

Riccardo got very close to the mark on his first guess.

Edit:

Those KEF 301's look perfect. They even have a built in wall mount bracket. I'll certainly check those out.

[Hans]

Real devices have noise current and 1/F noise.
JR

After some searching I could find a complete spice model of the ZTX851
Here it is including the ZTX851.
Noise is even lower as with the simulated Rbb of 2 Ohm, 44nV Flat and 34nV A-Weighted.

Hans

Wayne Circuit.jpg (212.96 KiB) Viewed 11799 times

[JR.]

Less noise than a 2 ohm resistor is impressive.

Do real world results match that?

JR

[Hans]

Less noise than a 2 ohm resistor is impressive.

Do real world results match that?

JR

It matches perfectly well.
On page 3 of this thread, Wayne has published 2 graphs.
Looking at the second graph, a total noise of -143.5dBu is measured, with a 3.3 Ohm resistor at the input, simulating the Cart.

In the simulation below with the same 3.3 Ohm, total noise is 55.22nV.
20*log(55.22e-9)+2.218 = -142.9dBu.
The added 2.218 is of course for converting dBv to dBu.
The simulation and the real world measurement are thus only 0.6dB apart !!

Hans

P.S. When replacing the LT1056 for a very low noise LT1028, I even get -143.2dBu !!

Wayne Circuit.jpg (221.24 KiB) Viewed 11791 times

[Hans]

I found it hard, having this LTSpice simulation model, to resist not replacing the Bipolars by Fets at the input. The BF 862 was used for that purpose, with 0.8nV/rtHz at 1kHz at Vgs = 0 Volt
I inserted not only a single pair, but also a quadruple pair.

One pair plus a 2 Ohm resistor in between will produce a theoretical 1.15 nV/rtHz and a quadruple pair 0.60 nV/rtHz both figures at 1kHz.
LTspice reported resp. 1.25 nV/rtHz and 0.72 nV/rtHz, because Vgs was at ca. 280mV instead of a very unpractical 0 Volt, in that case giving a lower gm and a higher noise figure because of that.
The IF 3601 or the dual IF 3602 specified for 0.3 nV/rtHz at 30mA, produces some 0.4 nV/rtHz at the 14mA as used for the (combined) BF 862.
So the quadruple pair of BF 862’s will have the same noise figure as an IF 3602 dual.

Having two ZTX851’s plus 2 Ohm resistor in between, producing 0.31nV/rtHz, boils down per transistor to the same 0.17nV/rtHz noise as a 1.4 Ohm resistor.
You would need at least 12x IF 3602 duals to get the same low noise!!
So if you are looking for the lowest possible noise, the ZTX851 can’t be beaten.

Here are the comparable results:

Hans

Input Noise.jpg (48.32 KiB) Viewed 11743 times

[JR.]

Just to be practical 8x BF862 is still lower Ciss than 1x IF3601.

The 862 appears to be a HF device.

JR