Hey all,

I have some serious questions regarding the following circuit:



This is the internal schematic for the LM384 5W power amplifier. I've been studying analog design and I analyzed what I could from this circuit, which is all the DC operating points, and the function of each 'piece' of the circuit. I was interested in some parts of the circuit, specifically those which gave me a real headache when trying to do some AC analysis. I took to SPICE to try and simulate the circuit, and it worked as described, with about 50 gain (33dB) at audio frequencies and above, but I actually couldn't use the transient analysis to figure out what was going on in some places. See below.

Confusing points: The two transistors on the bottom right, a PNP from the 10pf capacitor, and an NPN connected to its collector and emitter in some sort of PNP sziklai pair setup. By my calculations, and then SPICE, these are not biased on during normal operation. But they're surprisingly close. Is this some protection in case of a larger amplitude signal or a fault on my part?

Next: the 25k resistor in the middle of the schematic. At first I thought it was a feedback resistor, since it connects from the output back to near the input. But what bugs me is the resistor seems to be integrated into the amplifier- at DC it sets (or helps set) the output operating point, and at AC I don't even know what it does. It's connected to the emitter output of a differential amplifier, so it sees the input signal on that end, and it sees the output signal on the other, giving some difference voltage drop. I don't get it- what is going on with that resistor.

Finally: I believe the ultimate gain of this device comes from the differential amplifier and the NPN transistor in the bottom right with a 10pf capacitor connected from base to collector. This might seem to make sense, but honestly I can't come up with why that capacitor is there for a start, and second I have my doubts about whether those two stages supply all the gain. It comes back, I guess, to the 25k resistor, and in general whether or not the circuit uses negative feedback, which seeing how flat the response is and controlled the gain is I'd say it is more than likely it does. What do you all think about this circuit? Can some brilliant soul make heads or tails out of this and clear up some of these points? Thanks in advance!

Sam Gallagher

P.S. Attached is my SPICE sim in LTSpice

 

Sam, did you mean to short both inputs to ground, one explicitly, the other implicitly via the input voltage source. That is not the way these types of amps are operated. They need an input coupling capacitor, and the other input is left open. That might change the bias conditions a bit...

Also, do you have any idea that the default transistors NPN, PNP without qualification, are anywhere close to those inside the chip.

 

In no particular order...
The transistor with the 10 pF cap across it is the main voltage amplifier.
The 10 pF cap is called a Miller capacitor. This is a form of frequency compensation to prevent the amplifier from breaking into oscillation at low forward gain values.
The 25 K resistor sets both the gain and the output DC operating point.

AND - This is not a complete schematic of the part. This is more of a concept schematic.

ak

 

Mike, great points that I did spend a fair amount of time thinking about. There's no way the transistors are properly modelled in my simulation as compared to whats actually inside the chip. I was scratching my head a while trying to figure out how the inputs and outputs are actually used, and in every instance they simple couple the signal directly in. I could not find much in terms of circuits that used the chip, most simply showed the application notes.Shown here:

The whole "Vin" thing did nothing to help me! I wonder if it's simply an ac coupled signal or if it has biasing but I assumed it was AC coupled. But obviously one of the inputs is grounded, the other is connected to some AC input. What do you think? Would it make much of a difference? After all, the circuit does function in the SPICE simulation. But I am not the wisest of them all when it comes to analog!

AnalogKid thanks so much for your help! The Miller cap is really interesting thing, it's like it's there to use the Miller effect as opposed to fighting it? I'm interested to know more- I'm guessing it fights some sort of positive feedback that happens when the phase shift changes. Could you explain more?
Now, what other details would be included if this is simply a concept schematic? How different is the actual interior of the chip?
Thanks again, sorry to be a bother with all the questions.
Sam Gallagher

 

If you have time try read this
http://www.electrosmash.com/lm386-analysis
https://en.wikipedia.org/wiki/Miller_effect

 

Jony,
I'm a student on break, I have nothing but time! Thanks for the great read. The electrosmash article is great, stirs my mind well. I feel silly I did not find that in my research, I had no idea this was a thing people did, analyzing internal schematics and all!
Thanks again.
Sam Gallagher

 

If you have time try read this
http://www.electrosmash.com/lm386-analysis
https://en.wikipedia.org/wiki/Miller_effect

I was going to point him to Wiki for the Miller effect. That 386 article is outstanding.

ak

 

Hey all
I've done some thinking and I'm fairly sure now the capacitor is there to increase the effective feedback capacitance, which lowers the high pass response. I have also come across "pole splitting" where the second pole of the filter is moved higher to increase stability but I can't say whether this effect happens for any and all amplifiers using a Miller cap.
Finally, (hooray, only one question left!) I'm not understanding how both that voltage gain transistor (with the Miller capacitor) and the long tailed pair both supply voltage gain. The voltage gain from the diff amp is 50, the specified gain of the amplifier. So what is the purpose of the "voltage gain" transistor? It amplifies the original signal, and so does the diff amp. What's going on here? What is the net result? The output of the voltage gain transistor is coupled to the output through the transistors, emitter followers. I can't say I understand!
Thanks.
Sam Gallagher

 

The voltage gain from the diff amp is 50, the specified gain of the amplifier.

Or, the specified gain of the amplifier is 50, and there are two gain stages in series and fixed negative feedback to accomplish this. The total forward gain is much greater than 50, and the 25K and 1K resistors set the negative feedback ratio to stabilize the device AC gain at 50.

ak

 

AK, you're the man! That made it much more clear. I took another look at the internal schematic and saw it right away. You guys stuck with me, so thanks a bunch. I can always count on this forum.
Sam Gallagher