Symphony Amplifier

[OnAudio]

Here's the symphony amplifier
https://www.pcbway.com/project/shareproject/SYMPHONY_audio_power_amplifier_f05686aa.html

 

[Mark Tillotson]

Isn't the Zobel network the wrong side of the inductor? Many of the zeners have the wrong circuit symbol which is confusing. The lack of current mirror load on the input stage is noteworthy, usually that's an easy performance win.

 

[OnAudio]

Thanks Mark. There's a 10ohm resistor across the inductor so it wont make much difference. Yes the Zener symbols need editing. Yes there are many input stage loading methods.

 

[PMA]

Why is the Zobel behind L1? (R33, C7)

 

[EdGr]

Zener diodes are not usually used in the signal path because they are noisy. Regular forward-biased diodes can be used as level shifters.
Ed

 

[Mooly]

The QSpice files below

Nice One of the first Qorvo sims to appear on the forum.

(.qsch files can now be attached directly to the forum, no need to zip)

I was puzzling over the 15v references as I only see 11 volts on the opamp... there are some small 15v Zener's as default models if those might be suitable.

 

[OnAudio]

Why is the Zobel behind L1? (R33, C7)

Its fine as is due to the resistor across inductor. If moving it does make a difference and fixes oscillation here then there will be bound to be ghost oscillations later

 

[OnAudio]

Zener diodes are not usually used in the signal path because they are noisy. Regular forward-biased diodes can be used as level shifters.
Ed

Depending on where used feedback will generally try best it can to fix that

 

[OnAudio]

Thanks Mooly, theres a label looseness here left the label as such but ended up picking 12volt zeners

 

[HAYK]

Why is the Zobel behind L1? (R33, C7)

Quasi all Japanese amps and McIntosh use this configuration.

 

[PMA]

Its fine as is due to the resistor across inductor. If moving it does make a difference and fixes oscillation here then there will be bound to be ghost oscillations later

Parallel resistor across the inductor is not only OK, it is a must to fix ringing. But the Zobel should be placed in front of the inductor, or two Zobels can be used, one in front.of the inductor and another one behind the inductor.

 

[Mark Tillotson]

Well that's if you feed square waves through the amp, with audio signals ringing usually isn't an issue. You can run without the damping resistor quite safely - given the inductor usually has less than 1 ohm reactance across the audio band 10 ohms in parallel isn't really visible at audio frequencies, and the damping resistor isn't needed for stability.

 

[EdGr]

Depending on where used feedback will generally try best it can to fix that

I have never subscribed to the "feedback-fixes-all" school of design.
Ed

 

[Mark Tillotson]

Its a (very powerful) tool in the box, not a school of design.

 

[EdGr]

The "feedback-fixes-all" school of design is the use of feedback to (partially) correct a design deficiency, rather than avoiding the deficiency in the first place. I can point to numerous examples. The results are never as good as not having the deficiency.
Ed

 

[OnAudio]

Have updated schematic

 

[Mark Tillotson]

The "feedback-fixes-all" school of design is the use of feedback to (partially) correct a design deficiency, rather than avoiding the deficiency in the first place.

Since all active devices are non-linear, there's no design deficiency involved, just clever design that improves performance of said devices.

I can point to numerous examples

Go on then.

 

[Bonsai]

Nice result 👍. I’d put the Zobel on the amp side of the output coupling inductor though. Placing it after the inductor acts as a termination for the cabling<>speaker cable load but does not present the amplifier output with a fiat load impedance over frequency which is what it does if placed before the L. See Self and Cordell for more detail

 

[PMA]

Since all active devices are non-linear, there's no design deficiency involved, just clever design that improves performance of said devices.

This must be verified by practical tests into both R+L and R//C loads, of all possible values, including worst case electrostatic speakers, as well as with long cables. Simulation and/or testing with resistors is absolutely insufficient. I know very well what I am speaking about. Very high loopgain in combination with something like 3-pole compensation or 3EF output stages are always suspicious, regardless the perfect loopgain simulation with resistors.

 

[Mark Tillotson]

So please clarify "avoiding the deficiency in the first place" in this context? Here the deficiency would be failure to have very low distortion, how would you avoid that without feedback? You can't... Its the only tool in the box that can achieve this.