Matti Otala - An Amplifier Milestone. Dead or Alive

[Edmond Stuart]

Glen,

If by "collector impedance" you mean the "small signal output impedance" then you are correct. It would certainly help if you would brush a little your listening/careful reading skills; First, Ro IS NOT the output impedance. Ro is the impedance between the collector and the emitter that models the Early effect.

You changed "a little" the problem, now you are asking for the OUTPUT IMPEDANCE of a degenerated common emitter circuit. The answer is (with a small approximation):

(1) Zo=Ro*(1+Gm*(Rb'e||Re))

Before plugging in some numbers, please note that the effective degeneration is NOT Re alone, it is Re||Rb'e. And now, some numbers:

Ic=20mA
VA=100V
Beta=100
Re=100ohm

Then:

Gm=Ic/Vt=20mA/0.026V~800mA/V
Rb'e=Beta/Gm=100/800mA/V=125ohm
Ro=100V/20mA=5kohm
Rb'e||Re~56ohm

Note that the effective degeneration is about half the 100 ohm resistor, and that's simply because of the low Beta and high collector current.

Therefore:

Zo=5kohm*(1+800mA/V*0.056kohm)=5kohm*46.8 that is about 240kohm.
[snip]

>You changed "a little" the problem
Of course he's talking about the output impedance (that really matters) instead of some irrelevant quantity like Ro.

BTW, taking into account the effect of the current mirror and the input impedance (defined by R16=R17=1k8), the output impedance is 37kOhm, not 240kOhm.

 

[GK]

>You changed "a little" the problem
Of course he's talking about the output impedance (that really matters) instead of some irrelevant quantity like Ro.

BTW, taking into account the effect of the current mirror and the input impedance (defined by R16=R17=1k8), the output impedance is 37kOhm, not 240kOhm.

OK, I just peeked at your post. For that helpfull contribution you have been removed from my ignore list. Continue to be well behaved and I'll see about keeping it that way.

 

[Terry Demol]

Yes VT is the thermal voltage .0259V or so at 300K. The point is a cascoded current source still loses current via the base current (beta modulation) even though there are effectively "infinite" VT's of degeneration.

Yes, and as such unless the cascode is a 'Hawksford' type connection the cascode never buys as much more OP Z as you
would hope.

T

 

[Lumba Ogir]

Edmond,

For that helpfull contribution you have been removed from my ignore list.

You see, Glen is a kind person, but remember he can change his mind in any minute.

 

[Edmond Stuart]

ain't no angle

OK, I just peeked at your post.

You did wot? OTOH, I knew you couldn't resist the temptation. 😀

For that helpful contribution you have been removed from my ignore list.

Now, I should be very grateful, I suppose.

Continue to be well behaved and I'll see about keeping it that way.

As former fellow sin-bin-mate you should know better. You and I will never become an

 

[Edmond Stuart]

Edmond,

You see, Glen is a kind person, but remember he can change his mind in any minute.

I know, in particular concerning the latter. Thanks for the hint anyhow.

 

[Edmond Stuart]

Yes, and as such unless the cascode is a 'Hawksford' type connection....
T

Or even better: a 'Dimitri Danyuk' cascode.

 

[Wavebourn]

This is multiplied by 2 because there are two such common emitter stages with the degeneration+emitter resistances effectively operating in parallel.

No. It is one stage loaded on current source.

Since I'm in Glenn's ignore list I can say aloud: Syn08, please stop teasing the kid. He was right about 1K, but could not explain why. But he was wrong about any role of the pole of 1K and 100 pF since 1.5 mHz is way out of band. Miller compensation on audio frequencies and no resistive load that he suggested would create phase shift in audio band screwing down the whole Otala's idea-fix. Unfortunately I could not explain that to him because he was taught to follow Rules of Thumbs, like 25 mV instead of meaningful value that is temperature - dependent, so preferred to put me into his ignore-list like an interference to everything good and right he was taught.

 

[Wavebourn]

Or even better: a 'Dimitri Danyuk' cascode.

Do you mean a good old positive feedback AKA bootstrap?

 

[Edmond Stuart]

cascode

Hi Anatoliy,

Right, (almost) 100% positive FB of the cascode base current.
Maybe Dimitri is willing to send you a copy of his AES paper.

Cheers,
Edmond.

 

[Wavebourn]

Re: cascode

Hi Anatoliy,

Right, (almost) 100% positive FB of the cascode base current.
Maybe Dimitri is willing to send you a copy of his AES paper.

I hope so, after his apologies about rude comment he made when I posted schematic of my condenser microphone.

 

[MJL21193]

Re: ain't no angle

As former fellow sin-bin-mate you should know better. You and I will never become an

🙄
You guys are pussy cats... 😀

 

[Wavebourn]

Re: Re: ain't no angle

🙄
You guys are pussy cats... 😀

I've found an illustration to your comment:

 

[stinius]

Re: Re: ain't no angle

🙄
You guys are pussy cats... 😀

WTF are you talking about "big boy"?

 

[MJL21193]

Re: Re: Re: ain't no angle

I've found an illustration to your comment:

Thinking more like...

An externally hosted image should be here but it was not working when we last tested it.

🙂

 

[john curl]

Rest in peace, Matti, you have been vindicated.

 

[GK]

No. It is one stage loaded on current source.

🙄

Since I'm in Glenn's ignore list I can say aloud: Syn08, please stop teasing the kid. He was right about 1K, but could not explain why. But he was wrong about any role of the pole of 1K and 100 pF since 1.5 mHz is way out of band. Miller compensation on audio frequencies and no resistive load that he suggested would create phase shift in audio band screwing down the whole Otala's idea-fix. Unfortunately I could not explain that to him because he was taught to follow Rules of Thumbs, like 25 mV instead of meaningful value that is temperature - dependent, so preferred to put me into his ignore-list like an interference to everything good and right he was taught.

Gobbledegook.

1.5MHz is not an insignificant pole in just about any amplifier and any design (such as the lame Otala one) that drives an output stage such as a double EF (with large, variable input capacitance) from a high VAS impedance of 1k or so will suffer from a heap of additional (and unnecessary) open-loop phase modulation. I’m not sure how that is a recipe for making a “PIM free” amplifier.
As for the role of Miller compensation in minimising this issue in designs that use Miller compensation, try to look up “pole-splitting”.

 

[Wavebourn]

As for the role of Miller compensation in minimising this issue in designs that use Miller compensation, try to look up “pole-splitting”.

Too late, Glenn:

Rest in peace, Matti, you have been vindicated.

 

[GK]

Oh, very good.

 

[Edmond Stuart]

open loop phase modulation.

🙄

Gobbledegook.

1.5MHz is not an insignificant pole in just about any amplifier and any design (such as the lame Otala one) that drives an output stage such as a double EF (with large, variable input capacitance) from a high VAS impedance of 1k or so will suffer from a heap of additional (and unnecessary) open-loop phase modulation. I’m not sure how that is a recipe for making a “PIM free” amplifier.
As for the role of Miller compensation in minimising this issue in designs that use Miller compensation, try to look up “pole-splitting”.

Maybe some numbers might shed some light on this discussion:
Simmed at 1MHz (though not sure how relevant this frequency is) OTala's VAS + OPS revealed an additional phase shift of about 22.8 degrees when the output voltage was driven from 0V to plus (or minus) 20V. Replacing those two ******* 2k2 resistors by a (balanced) Miller compensation (2x100pF) the additional phase shift was only 2.3 degrees.