Lowest noise BJT transistor?

[mchambin]

The Zetex ZTX851 was designed as a high-current low saturation voltage switch. As a consequence of its device geometry, the ZTX851 also has a low rbb resistance making it suitable for low-noise, low source impedance audio amplifiers.

There are plenty of such transistors.
Presumably they have low rbb, suitable for low noise applications.
What about searching for matched pairs ?

 

[EUVL]

Many of these low rbb transistors have rather low hfe.
I have a hard time finding any with hfe > 350, except for the Toshiba's in SOT23.


Patrick

 

[billshurv]

There are plenty of such transistors.
Presumably they have low rbb, suitable for low noise applications.
What about searching for matched pairs ?

Not sure what the avialability NOS of the LM394 is these days, but there is the THAT300 series if you want matched (quads not pairs). But depending what you want to match on SMALL numbers variability within a batch of the ZTX parts is low and its easy to match up and then thermally couple them.

Many of these low rbb transistors have rather low hfe.
I have a hard time finding any with hfe > 350, except for the Toshiba's in SOT23.

This is indeed a problem as I noted. No free lunch sadly.

 

[mchambin]

Not sure what the avialability NOS of the LM394 is these days, but there is the THAT300 series if you want matched (quads not pairs). But depending what you want to match on SMALL numbers variability within a batch of the ZTX parts is low and its easy to match up and then thermally couple them.

There is AS394 available, a remake of LM394.

About matching and fitting ZTX851 parts together.
I rather have a pair in the same case with good matching.

 

[billshurv]

That's fine, but I want lower noise and this thread is 'lowest noise BJT'. You want 'most matched lowish noise BJT'

 

[mchambin]

I wonder about the discovery of the ZTX851, not designed for low noise.
How come there is no hunt for other ones ?

 

[mchambin]

That's fine, but I want lower noise and this thread is 'lowest noise BJT'. You want 'most matched lowish noise BJT'

How many ZTX851 to get a pair within some quoted Vbe and hfe match ?

 

[billshurv]

Depends how close you want the match to be. For normal MC duties and within the same batch seems to be around 80%.


As for the hunt, please take a look at Art of Electronics V3. H&H did try a lot of suspects and significant sample sizes. I would be suprised if there is anything lower than the 851/951 without some serious compromises. I would post the diagram from the book, but don't want to out of respect for their hard work. But it has been put on this forum recently in the richard lee MC thread.

 

[jan.didden]

Those AS394 are ideal matched pairs! Can't believe I missed them. I remember the original LM394, years ago.

Physically, these are about 100+ transistors on a single die, cross-connected, giving very tight thermal coupling and incredible matching. Vbe drift 0.1uV/C, Vbe offset 100uV. Hfe > 500, matched to about 0.5%.
Only $ 4 too.

Jan

 

[billshurv]

Am I reading the spec right though, 40Ohm Rbb?

 

[mchambin]

I do not find where this 40 Ohm Rbb is specified.

 

[billshurv]

AN-222

 

[mchambin]

Got it, thanks.

 

[29285]

Noise factor/figure is ratio of amp noise to (amplified) source noise, 1dB is 1.26 ratio, so the amp acts like 260 extra ohms on top of the source 1k ohms, Johnson noise of 260 ohms is about 2nV/sqrt(Hz)

Actually Johnson noise is a very British phenomenon

 

[dotneck335]

So, in my low-noise-required, high gain circuit (reverb recovery) with my ZTX-851 transistors, how much NOISE difference do metal-film resistors make compared to carbon comp ones? I read somewhere that they were 10x quieter, but 10x WHAT? Do they make an audible (≥ 1db) improvement?

 

[MarcelvdG]

Depends on where and how they are used in the circuit. The thermal noise is exactly the same as that of any other type of (passive) resistor, but they have less 1/f noise, meaning smaller random fluctuations of resistance. By the way, wirewound resistors are even better in this respect.

 

[MarcelvdG]

As far as I know, the 1/f noise is usually specified in microvolt RMS in one decade per volt of bias voltage, so actually in ppm RMS resistance variations over a 1:10 frequency range. The greater the bias voltage, the more noise you get; without DC bias, you only get sidebands around the signal.

 

[johnego]

Groner's Noise Threshold for Opamps

A bit off topic, I'm asking about the noise threshold that Samuel Groner mentioned in his opamp measurement publication. He mentioned 5nV/rtHz as his minimum standard. Anybody know the background of this number? Is it an audibility threshold? I'm aware that noise is a big issue in line level where opamps are usually used. In my current project I'm trying to find answer to this 'noise threshold' using LN opamps like AD797. The weakest link in my calculation is the resistor noise that I assume around 15nV/rtHz at HF. The LF noise is even worse. So isn't this 5nV/rtHz of voltage noise seems to be too small to be audible?

 

[DPH]

Citation? I'm trying to read between the lines and it almost sounds more like an arbitrary distinction/cut off. Wouldn't some FET input opamps just above this cutoff be quieter than some bjt inputs, depending on the network impedances surrounding the opamp? (Don't forget i_n)

If you need help getting the noise down in your application, and 15 nV is pretty high for low noise, have a look at some of the high gain measurement preamps here on diyaudio.

 

[johnego]

and 15 nV is pretty high for low noise,

This is actually what I don't understand yet. Why do you think that 15nV is pretty high. Is it based on audibility? This is for line level audio application.