Voltage-Bootstrapped Current-Boosted NE5534 Swinging Op Amp Audio Line Driver

[mediatechnology]

One of the limitations of bootstrapping common op amps is the output current limitation. You can get the high voltage swing but not enough current to deliver it into a lower impedance load.

These posts were split from the "Bootstrapped Op Amp Articles" thread found here: viewtopic.php?f=12&t=699

This circuit fragment can be used as a starting point to combine voltage-bootstrapping with current-boosting. By adding two diodes to the voltage bootstrap voltage dividers, two additional transistors and emitter resistors, a minimum component count line amp can, with ±30V supplies, produce +27 dBu into 600Ω. Due to voltage-bootstrapping the 5534 only sees +/-20V supplies.

Placing links in the "B" position connects the output collectors to the bootstrap rails to shift some of the output dissipation to the bootstrap transistors. For higher-current outputs and >60V Vcc-Vee, BD139/140s mounted on a heatsink are a better option. At ±32V, with the values shown, the 5534 sees 42V. Scaling the 4K7s higher reduce op amp supply voltage. An OPA604, with its 48V maximum operating voltage, is also worth trying in this socket.

This circuit is a starting point for someone interested in pursuing it. I tested this with ZTX851/951 pairs and the THD at full output is pretty respectable. At lower levels the THD is essentially the op amp residual. Some additional tweaking can be performed and your mileage may vary.


Voltage-bootstrapped current-boosted NE5534 swinging op amp for audio line driver providing +27 dBu into 600Ω

Overload=+27 dBu into 600Ω with ±30V supplies. 17.4Vrms; 48V P-P, 500 mW.

[JR.]

The 5534 breaks out a compensation pin, I wonder if there are any tricks available for alternate compensation schemes?

I recall the old LM301 had a feedforward compensation that only worked with inverting topology. IIRC there was a cap that bypassed around the input LTP at very HF for better slew rate and HF performance... But 5534 already uses a weird split-pole compensation so may not like trick compensation schemes.

JR

[mediatechnology]

But 5534 already uses a weird split-pole compensation so may not like trick compensation schemes.

I know I've seen external feed-forward compensation with the LM318 and I think I've seen it with a 5532. It might have been in Cyril Bateman's sub-PPM oscillator.

Weird it is. The kink in the 5534 open loop frequency and phase response curves between 100 kHz and 1 MHz are real despite being smoothed out by the artist:



I discovered 5532s/5534s required lots of over-compensation when they get nested within feedback loops with other 5532s/5534s.
When the kinks line up in two or more nested stages there can be an out-of-band snap resonance around the kink's frequency.

I think the best option for the booster circuit is the local Cfb directly around the op amp with the one across R2 open. That's the Jim Williams "fast output slow error amp" approach. In my tests circuit I didn't use any Cc or Cfb but the gain was high in order to get the output swing.

I probably should note for our readers that the C across the output bases, if used, should be an electrolytic. I used 47 µF and saw a very slight reduction in THD.

[JR.]

I thought the kink in the Bode response plot was at lower frequency (more like 2kHz) but my memory is harder to trust these days and/or process improvements could have shifted it higher.

The kink was one of the (very) few things I didn't like about the 553x but it's still a respectable op amp when used rationally, and I've used truckloads over the years.

At Peavey we had a 5532 house number in the system, graded for low noise. The selection didn't define a low noise floor, but the spectral balance of the noise, to screen out audible 1/F noise.

JR

PS: I like that topology, adding two more diodes (reverse polarity across output device base-emitters) will provide some crude current limiting, FWIW

[mediatechnology]

PS: I like that topology, adding two more diodes (reverse polarity across output device base-emitters) will provide some crude current limiting, FWIW

Thanks John. I normally put rev Vbe diodes in but left them out of the schematic fragment for clarity. On the DCAO I use them.