Walt Jung biasing opamps in class A

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
There is information about doing this in Douglas Self's book "The Design Of Active Crossovers".
For reference:
"13.10 Reducing 5532 Distortion by Output Biasing" page 389 in Design of Active Crossovers.

"For the 5532, the current must be injected from the positive rail; currents from the negative rail make the distortion emphatically worse... there is a general reduction in distortion across the whole of the 5532 output range, which seems to indicate that the 5532 output stage is better at sinking current than sourcing it, and therefore injecting a positive current is effective at helping out."

Self wrote about this after he had already written the first edition of Small Signal Audio Design, so I guess he just forgot to add it to later editions.
 
In this case it is, because pin 5 comes out after the VAS and before the AB output buffer. FWIW they use this technique in a lot of the new RND products, which sound absolutely fantastic.
But you are not using the output buffer, you are replacing it so it is not the same. :)

Cheers

Ian
 
For reference:
"13.10 Reducing 5532 Distortion by Output Biasing" page 389 in Design of Active Crossovers.

"For the 5532, the current must be injected from the positive rail; currents from the negative rail make the distortion emphatically worse... there is a general reduction in distortion across the whole of the 5532 output range, which seems to indicate that the 5532 output stage is better at sinking current than sourcing it, and therefore injecting a positive current is effective at helping out."

Self wrote about this after he had already written the first edition of Small Signal Audio Design, so I guess he just forgot to add it to later editions.
It's definitely not in the edition I have and I could find no reference to it on his web site.

Presumably he measured how much the distortion was reduced?

Cheers

Ian
 
I reckon adding two transistors, a couple of resistors and diodes to a 5534 is a very neat way of making a high performance true class A opamp with the freedom to easily set the idle current to suit expected loads.
 
I reckon adding two transistors, a couple of resistors and diodes to a 5534 is a very neat way of making a high performance true class A opamp with the freedom to easily set the idle current to suit expected loads.
Back in the very early days of op amps like the 709 and the ubiquitous 741, the most pressing problem for audio circuits was their poor noise performance and gain bandwidth product. Both of these were significantly improved by adding a discrete transistor (or two) at the input (one of the better designs being the Tweed Audio mic pre designed by David Rees). Later op amp deveelopments improved the noise performance to the point where an external transistor was no longer necessary. I wonder why this was never extended to the provision of class A output stages?

Cheers

Ian
 
I reckon adding two transistors, a couple of resistors and diodes to a 5534 is a very neat way of making a high performance true class A opamp with the freedom to easily set the idle current to suit expected loads.
What exactly is a class A opamp? FWIW the old school single supply discrete transistor audio paths were "class A" and not very low distortion.
Back in the very early days of op amps like the 709 and the ubiquitous 741, the most pressing problem for audio circuits was their poor noise performance and gain bandwidth product. Both of these were significantly improved by adding a discrete transistor (or two) at the input (one of the better designs being the Tweed Audio mic pre designed by David Rees). Later op amp deveelopments improved the noise performance to the point where an external transistor was no longer necessary. I wonder why this was never extended to the provision of class A output stages?

Cheers

Ian
+1 not just the input, but adding discrete bipolar devices to the output side could be beneficial (i've added discrete devices before and/or after).

"Class A" is a poorly understood buzz word. AFAIK there is no such thing as a class A op amp (operational amplifier). There is such a thing as a "class A" power amp (that doubles as a room heater because it is grossly inefficient).

JR
 
Last edited:
"Class A" is a poorly understood buzz word. AFAIK there is no such thing as a class A op amp (operational amplifier). There is such a thing as a "class A" power amp (that doubles as a room heater because it is grossly inefficient).

JR
There are plenty of class A discrete op amps - the Calrec range for instance or the Neve BA440.

Cheers

Ian
 
There are plenty of class A discrete op amps - the Calrec range for instance or the Neve BA440.

Cheers

Ian
Hi,
strictly speaking, the Neve BA-440 shouldn't really be included as an op-amp since it lacks the differential amplifier on the front end.
The BBC designed replacement: BA-512 had this as well as better compensation. IOW it's a real op-amp.

Either also ran with considerable class-B action on their outputs as was originally used in desks for driving a 50 ohm 10dB step-up transformer.

We would need about 170mA bias to do that in class A with the Neve rail, whereas the 440 and 512 were biased for about 12mA.

Isn't the PQ-14/15 the only class A Calrec?
 
Last edited:
I wonder why this was never extended to the provision of class A output stages?

Cheers

Ian
I know the 5532 runs pretty warm with even "modest" +/- 18 V rails. My hunch is that a Class A output stage would cause the chip to be able to cook a pork chop! Just add a tiny cast iron heat sink on top. <g>

Seriously though, the earlier Neve desks using the 5532 chips are notorious for creating enough internal heat to require regular replacement of the 'lytics and those oddball multi-conductor "jumpers" that interconnected the various PCBs inside the module.

Bri
 
There are plenty of class A discrete op amps - the Calrec range for instance or the Neve BA440.

Cheers

Ian
I didn't try searching "Calrec range" but I found a schemo for the BA440.

What I found was not an "operational amplifier". https://www.technicalaudio.com/neve/neve_pdf/0440_BA440_class_AB_line_amplifier_doc_4pages.pdf
FWIW the link calls it a class AB line amplifier... Neve calls it a "quasi-operational amplifier".

Nor was the output stage class A, it was very much class AB.

wiki said:

Operational amplifier


High-gain voltage amplifier with a differential input
An operational amplifier is a DC-coupled electronic voltage amplifier with a differential input, a single-ended output, and an extremely high gain. Its name comes from its original use of performing mathematical operations in analog computers.

Sorry to be pedantic... old school 9 transistor radios were class A, modern audio paths are not.

JR
 
Hi,
strictly speaking, the Neve BA-440 shouldn't really be included as an op-amp since it lacks the differential amplifier on the front end.
The BBC designed replacement: BA-512 had this as well as better compensation. IOW it's a real op-amp.
I agree but it is debatable if an operational amplifier is what you really want for audio. They were originally designed for use in analogue computers for military applications
Either also ran with considerable class-B action on their outputs as was originally used in desks for driving a 50 ohm 10dB step-up transformer.

We would need about 170mA bias to do that in class A with the Neve rail, whereas the 440 and 512 were biased for about 12mA.
I think you need a lot less bias current than that. The BA283 output stage runs at about 80mA and that is single ended so a push pull class A should require no more than 40mA quiescent but that still equates to nearly 1W of dissipation with a 24V rail. My tube class A output stage runs at 6mA quiescent and will output +26dBu into a 600 ohm load via a 2:1 step down transformer. The Neve needs a 10dB step up so its quiescent needs to be 16dB higher which would be just under 40mA. Even so, the real reason there are no class A op amps is clearly the quiescent dissipation is way too high.
Isn't the PQ-14/15 the only class A Calrec?
I am not sure. I am at the wrong PC right now. I will check later.

Cheers

ian
 


Write your reply...
Back
Top