qsc topology, but without opamp?

[gruni]

Hi everybody!
I was wondering if its possible to build a power amp based on the typical qsc topology with floating power supply, but using a discrete circuit, eg. LTP instead of the opamp driving the output transistors.
has anybody done this so far?

if you dont know what i mean, you can look at the qsc 1080 schematic. its so far the simplest amp i found using this topology.

thanks in advance

[StevenOH]

Yes, of course. You don't need an opamp, just an amp with inv and non-inv inputs for feedback arrangements. And that amp two inputs need not to be of the LTP stage (balanced). Opamp is used for more gain in open loop, which is not a requirement for the circuit to perform well.
QSC use drains/collectors connected to ground, so this is suitable for many typical transistors/V-MOS, since it will have heatsinks at zero potential.
For L-MOS (where the sources are on the case), it makes sense to ground the sources.
Plenty of ideas for inquiring mind can be found here:
DNFB 2SJ200/2SK1529 overturned power amplifier
and in Hafler pat. US4467288

[KAZVARGA]

You will need a seperate power supply for the input stage, one which does not swing with the audio. The qcs type kept charge on a pair capacitors by using the output swing as bootstrap, this is why shorting the amp made the signal go down.

[djk]

http://www.hafler.com/techsupport/pdf/MAN1461C_P1500_P3000_man.pdf

[tiefbassuebertr]

Originally Posted by djk

http://www.hafler.com/techsupport/pdf/MAN1461C_P1500_P3000_man.pdf

interesting topology.
here two URL's regarded this amp:
Hafler 9505
Hafler - Home
also Altec and Harrison works with grounded collector resp. grounded source in their output stage, but with integrated OP-AMP driver solution:
What was the first Complementary AB bipolar amplifiers? (post #102)
Harrison type "K4000" (K 4000) MKII 2x1000W/4ohm

[jcx]

Dr Ed Cherry advocated for CE output stages, pointing out that taking all of the V gain in the output Q lets the input circuitry run at low Vsupply allowing faster Q

but the question today would be why avoid the op amp input and driver when you could use some of the past decade's better op amps?

except for lowest noise with exotic source impedance today's best op amps built on fast complementary isolated processes have better internal Q parameters than you see in discrete diy audio amp circuits here

also skilled design teams at Analog, BB/TI, Linear, National, backed with deep institutional knowledge have considerably upped op amp circuit topology for low distortion at high frequency performance over the past decade due to A/DSL performance demands

a composite multiloop op amp circuit with a modern low noise precision op amp input and CFA A/DSL driver op amp output should easily outperform any discrete diy hobbyist (and at this point any audio pro discrete design) circuit with consumer line level input

[StevenOH]

Hafler patent for LTP interface (9505 topology) is US5936467.

[balaboo]

Er, you do NOT need a separate power supply for the input stage - I have a standard LTP and VAS running off 15V zener rails from a QSC style design. The "bootstrap" capacitor is to provide drive limiting during overload conditions.

[tiefbassuebertr]

Originally Posted by StevenOH

Hafler patent for LTP interface (9505 topology) is US5936467.

Amazing! Only eleven years old:
Dynamically invariant AB linear operation amplifier

Originally Posted by balaboo

Er, you do NOT need a separate power supply for the input stage

Yes, but for another problem I don't find a solution: the oscillation both in real live and by simulation (see attachement). Who have an idea, where I must introduce the compensation caps?

[KAZVARGA]

What makes these circuits a little tricky is the op amp in a global feedback loop. The op amp used must allow the loop to be closed in a short time. A high slew rate op amp may still have a long propagation delay. A transistor input stage is often easier for diy for this reason. A 2n5550 ltp has a very short delay, and you have access to everything for compensation. Qsc modified thier version over the years, so check a newer ('90s or later) schemo. Some compensation points : 10pf across feedback resistor, or 47 to 100 pf across c-b of drivers. It all depends on the op amps ,drivers and outputs used. In general, the sooner global feedback closes the loop, the better. So compensation caps should only reduce gain spikes, but not slow down the amp much.