low noise bjt pair recommendations

[Colon]

Hi,

I'm having trouble finding a suitable pair of transistors for my amplifier push pull output. They are being driven by an op amp output as part of a feedback loop so low source resistance making bjts best. They also need to be medium power as they supply 0.5W max each. My application operates from 10kHz to 1.5MHz, from a 12V rail.

The problem I am finding is that datasheets and manufacturers rarely give noise data unless it's an rf bjt. Then, they do specify noise figure, but they don't specify the 1/f corner frequency, so I am not 100% sure the noise will be good at the bottom end (or is 10k well above most 1/f, even for rf?).

Is there something I am missing or am I best just asking manufacturers for a recommendation based on my spec? If not, does anyone have any to suggest?

Thanks
James

 

[crutschow]

The noise of the output transistors is unlikely to be a problem since their noise contribution is reduced by the closed loop feedback. Generally the only noise you have to be concerned about is the first gain stage (or op amp if that's the first stage), since the, noise relative to the output, of all the following stages, is reduced by a factor equal to the gain of the stages preceding them.

 

[FvM]

A push-pull output stage doesn't need transistors with particularly low-noise figure if it's inside the feedback loop. Overall amplifier noise will be dominated by the input stage.

Silicon BJT have generally a low noise corner frequency (in a 10 or maximal 100 Hz range). Most modern small and medium power transistor have little excess noise above it's "natural" shot noise. High B is preferable if noise current density matters.

 

[Colon]

Ok, see what you mean.

Would this still be true if my amp is unity gain though?

The reason I am looking at the transistors is because my measured noise figure is 2dB higher than my model suggests. I was going to wire the amp alone(without push pull on the output) and measure the noise figure again.

Thanks
James

 

[FvM]

Would this still be true if my amp is unity gain though?

Yes, as long as the amplifier has sufficient loop gain.

 

[Colon]

I see, because although close loop gain is unity there is still high open loop gain before the output stage

Not sure where my extra 2dB is coming from! I'll investigate tomorrow.

Thanks
James

 

[Colon]

Ok, just got to work. I'm about to do a few tests.

I've been thinking though, I'm not sure (I might find out I'm wrong in a short while...) I agree with the idea that the amps open loop gain will reduce the noise contribution of my push pull output in this configuration. By my logic, if the effect of negative feedback reduces the noise contributions of the parts in the amplifier loop (such as the feedback and gain setting resistors) then the op amp configuration with the most negative feedback (i.e. the unity closed loop gain case) would have the lowest noise. However the opposite is true. This also concurs with my model, based on this TI document originally. Here, the voltage noise from Rf appears directly at the amp output (similar to my push pull), and then it is divided by the signal gain (unity in my case) to refer it back to the input.

Any thoughts?

James

 

[Colon]

well, I hold my hand up, you were right. Measuring the noise without push pull output made no difference to the noise figure. I learn something new everyday!

Off to try and work out why my model is wrong...

Thanks for your input.
James

 

[crutschow]

What is your model for the op amp noise?

Are you aware than an inverting op amp circuit with a gain of 1 has a noise gain of 2 for the internal op amp noise?