John Curl's Blowtorch preamplifier

Status
Not open for further replies.
Jump to Latest
#18,008
Re: Re: Re: Re: Re: Re: AD797

bear said:



Sorry, I must have missed the discussion on the output transistors. What is the significance of the "emitter sizes"??
Click to expand...

IC design stuff 😀. In IC technologies it is pretty much guaranteed that transistors of identical geometries are electrically identical, something you cannot guarantee in the discrete world (unless you choose to hand sort the devices to an arbitrary precision).

Now, imagine a non-degenerated current mirror, that is, two identical transistors with the BE junctions in parallel. In the discrete world, forcing a collector current (e.g. via a CCS) through one of the transistors will have an undetermined effect on the other transistor collector current; this is because the imperfect matching of the discrete devices. However, in the IC world those currents will be almost ideally equal.

Now, what if you want a well defined ratio of currents through the current mirror branches? In the discrete world you have no other chance than degenerating each transistor emitter, and then the currents will stay in the emitter resistors ratio. However, it can be shown that in IC technologies the current also stands in the emitter geometrical area. Therefore, at least theoretically, you don't need any degeneration to provide a current ratio, just design the transistors with the desired emitter area ratio. Two less noisy and nonlinear "dirty sand" resistors 😀.
 
#18,012
I would like to be able to emulate the actual output NPN device that I am going to change the real output for the AD797, to the best of my ability. Many devices have different peak betas. It would be nice to know the 'ballpark' of what Scott used, but he doesn't seem to know it, himself. That is the problem. Please don't waste my time with criticisms.
 
#18,014
Re: AD797

Wavebourn said:
I think, closer to the topic of Scott's Wonderopamp, Rbb is different.
Click to expand...

Sure, but I already covered this aspect (i.e. low bulk resistance) by using the models of the low noise MAT02 for Q5 and Q6. These trannies have almost the same (large!) size as Q5/6.

edit: The points is that I have issues with the output stage as well. So I wonder how much larger the output trannies (Q22 & Q30) are WRT to the size of the drivers.

Regards,
Edmond.
 
#18,017
john curl said:
I would like to be able to emulate the actual output NPN device that I am going to change the real output for the AD797, to the best of my ability. Many devices have different peak betas. It would be nice to know the 'ballpark' of what Scott used, but he doesn't seem to know it, himself. That is the problem. Please don't waste my time with criticisms.
Click to expand...

Scott knows perfectly well, and he already told you point blank that the beta variation over the normal collector current range is negligible. If the IC design engineers don't care about this dependency, and are not designing based on such, you knowing about is going to help you as much as knowing the ohm/square resistivity of the epi layer.

You were also told by several people around that loading the opamp output with a current source will do jack **** in improving the performance.

I have suggested to ask about your PhD partner, ex Motorola design engineer and Jack Bybee associate. Have you?

Your persistence on this topic makes me think you are looking to back up a sales pitch (like "an AD797 opamp gain stage, with the output stage optimized for maximum gain linearity") rather than a real piece of information.
 
Status
Not open for further replies.
Top Bottom