Solid-state audio amplifier complexity vs valve/tube

[Elerion]

No. It uses the speaker as a resistor for "bootstrapping" then the current of driver transistor TR3 goes into the speaker which offsets its cone.

Could you explain a little bit how this bootstrap works?
The followng circuit shows a bootstrap using the load and the output capacitor, which is clear enough for me to understand.
How does it compare to this simple circuit?

 

[Nigel Goodwin]

Could you explain a little bit how this bootstrap works?
The followng circuit shows a bootstrap using the load and the output capacitor, which is clear enough for me to understand.

The bootstrapping consists of two resistors, with a capacitor feeding the middle of them - in this case the top resistor is replaced by the speaker, and the capacitor replaced by the speaker coupling capacitor. And contrary to Audioguru's concerns the DC current through the speaker is absolutely tiny, and of no concern whatsoever - it's a perfectly acceptable circuit configuration, and saves a capacitor and a resistor (however, I'm with AG in that I wouldn't design an amplifier in that way).

The way bootstrapping works is quite simple - in the case of the design above the top output transistor is limited to how far positive it can go, due to the resistor feeding the base.

For an example, assume the resistor is 1000 ohms, and the supply voltage is 40V, with no signal there's roughly 20V across the resistor, giving 20mA drive capability to the output transistor.

If the output swings positive by 10V, this means only 10mA drive, and for 15V (leaving 5V across the resistor), only 5mA drive - even worse, as the output goes higher it draws more current, while the drive capability is falling. This causes considerable distortion, and greatly reduces possible output power.

By connecting a capacitor from the output to the top of the resistor, as the output swings positive the top of the resistor swings positive as well, maintaining the voltage across the resistor, and thus maintaining the current drive capability.

A more modern way is to replace the bootstrapping with a constant current source, which gives a similar result.

 

[Elerion]

Thank you Nigel.
I understand the top circuit, but what about the one that Audiogurus posted?
Is R13 the other bootstrap resistor?
Tr6/Tr8 act as an emitter follower, so when output voltage raises, both sides of R13 raise too.
Am I right?

 

[audioguru]

Here is a comparison of an ordinary amplifier and one with bootstrapping. I also show an added resistor and capacitor for proper bootstrapping.

 

[Nigel Goodwin]

Thank you Nigel.
I understand the top circuit, but what about the one that Audiogurus posted?
Is R13 the other bootstrap resistor?
Tr6/Tr8 act as an emitter follower, so when output voltage raises, both sides of R13 raise too.
Am I right?

Apart from in this circuit it's 'upside down', so in this case the voltages are falling rather than rising. This is done to make the speaker connect to chassis, rather than to the main HT rail.

I don't know which circuit of AG's you're referring to?. But on the Leak circuit, C32 is the bootstrap capacitor, in conjunction with R44 and R45.

 

[audioguru]

On the amplifier with the PNP driver transistor, the bootstrapping allows the base voltage of the PNP output transistor Sziklai pair to go much lower than without bootstrapping.

 

[Elerion]

I don't know which circuit of AG's you're referring to?

Sorry!! I meant crutschow. The one I attached in my previous post.
Was I right about my assumptions?

Thanks about the info about Leak. I could recognize that one

 

[audioguru]

Here are my simulations showing the magic of bootstrapping:

 

[Nigel Goodwin]

Sorry!! I meant crutschow. The one I attached in my previous post.
Was I right about my assumptions?

If you mean the original Crutschaw one?, then C3, R6 and R7.

My previous answer already referred to the image you posted.

 

[Elerion]

Thanks Nigel and Audioguru, very helpful info.

I'm having some trouble understanding T6 on Leak amplifier.

It seems to me a common emitter, driving T7, which drives the diodes, and the remaining four transistors are the complementary output push-pull transistors.
But I never saw a bias circuit like the one used for T6 (basically the emitter side). I don't know how to analyze it. Any idea where can I study or read about this?

 

[Nigel Goodwin]

I don't think it helps with T6 been 'upside down', as it's a PNP device, also the fold in the scanned image doesn't help, nor do the unusual transistor symbols

Here's a similar sort of diagram, that is clearer to look at:

It's all to do with feedback, the junction of the emitter resistors (R10/11) needs to be half the supply rail, so this provides DC negative feedback via R5 to Q1, the base of which is at half supply (from R1/2).

Now 100% negative feedback would make for a very quiet amplifier, so while you want 100% for DC (to keep the circuit stable and safe) you don't want it for AC - and that is the function ofR5, R12 (which set the AC gain) and C4 which blocks the DC, and provides a good low frequency response.

The Leak example above actually has two negative feedback circuits, the first as in this one, the second comes after the speaker coupling capacitor and is via R48 - T6 is also bootstrapped, via C30.

 

[Elerion]

Here's a similar sort of diagram, that is clearer to look at

Much better now! I did know about that kind of feedback, but I just couldn't recognize it on the Leak amp.
I like that circuit. Very neat. I will consider building it
Maybe substituting one diode by a potentiometer. Seems like a fun and simple first amplifier to build.

 

[Nigel Goodwin]

Maybe substituting one diode by a potentiometer. Seems like a fun and simple first amplifier to build.

Personally I'd replace the diodes with a Vbe multiplier (with a pot and two resistors), something I've always preferred over the diode solutions.

 

[audioguru]

Q2 on the "similar" diagram is missing a low value frequency compensation capacitor from C to B to avoid instability and oscillation when negative feedback is used, caused by phase shift.

 

[Nigel Goodwin]

Q2 on the "similar" diagram is missing a low value frequency compensation capacitor from C to B to avoid instability and oscillation when negative feedback is used, caused by phase shift.

Interestingly the Leak design is also missing that, however there is a capacitor on T6.

 

[AnalogKid]

I found the Leak 70 Stereo amplifier quite complex as a first project of this kind.
What do you think about these simple ones?

I vote for the 15 W. It has all of the classic circuit features, plus an "active zener" for biasing the output stage, a hot new idea at the time.

ak

 

[AnalogKid]

The LM3886 is 14 years old, is inexpensive and has excellent performance:

True, but there isn't as much fun or learning in hooking up a black box as there is in an all discrete design.

ak

 

[Elerion]

True, but there isn't as much fun or learning in hooking up a black box as there is in an all discrete design.

Yes. I'm looking towards building my own amplifier, and understand as mush as possible about it.

Q2 on the "similar" diagram is missing a low value frequency compensation capacitor from C to B to avoid instability and oscillation when negative feedback is used, caused by phase shift.

I'll simulate the circuit and see how a miller capcitor affects its frequency response.

 

[audioguru]

I'll simulate the circuit and see how a miller capcitor affects its frequency response.

The TIP3055 output transistors are old and slow. I think their phase shift causes oscillation when negative feedback is added. Of course adding a compensation capacitor will cut high frequency response like in all old solid state amplifiers.

 

[Elerion]

Before going on, I'd like to share my first design, to seek for tips and advices.
Simple long-tailed pair input, with current source and active load, simple VAS (active loaded too), and simple drivers and push-pull output. No bootstraping, no fancy things. Quite straight-forward. Just +15/-15V. Around 6 watts into an 8 ohm load for a 1V input.



I already want to replace the two diodes with a Vbe multiplier (which, when built, should be thermally coupled to the output transistors), and add small emitter resistor to the output stage.

The open loop simulation shows plenty of gain/phase margins, and bandwith is almost flat between 20Hz and hundreds of kilohertzs.