Class A bias opamp

Dear Good Internet headphones friends. I have read a few places about biasing an opa627 into class A to get better performance out of it but notice some strange discrepancy.

http://www.alicemagicbox.com/lasercollection/lovelycubeclassabiasmod.html

This site and really useless headphonia link (somehow marv was a part of this shennanigans) show 2 resistors used with pins 1 (offset trim) and 4 (v-) and 7 (v+) and 4. Nothing about V- to output as described here...

https://tangentsoft.net/audio/opamp-bias.html

http://www.diyaudio.com/forums/chip-amps/38984-opa627-sings.html

I probably trust the latter two guys over the former, but wth why are these different approaches?

 

I've seen the jfet bias used with the Millett Butte

http://www.pmillett.com/butte.htm

 

There is a range of different approaches to achieve the same (or similar) outcome. There are pros and cons. They all result in shifting the bias of the output stage. It was always a tweak with the OPA627, but newer opamps are OK (better?) without it.
Personally, I would be mostly concerned by all the bad effects of those multiple stacked IC sockets topped off with a SMD adapter PCB. Stacked sockets and resistors/capacitors with long legs (antennae) can get you unwanted HF oscillations.

 

The sockets are great for convenience but you could always solder the resistors straight to the opamp pins or bottom of the pcb.

 

Mystery solved.... the right answer was always Tangent/Carlosfm way of pin 4 and 6 for single opamps like the opa627. Those morans at headphonia (or maybe just me for never realizing it) never stated they did the mod for dual opamps on adapter of the opa627. Now it all clicks.

 

The rule these days is that biasing into class A is becoming meaningless with modern opamps. The most extreme improvement is achieved with LM324 which has an ultrahigh crossover distortion.

I have tested an OPA627 and an AD8610 with absolutely no improvement according to my ears. The design was the QRV04.

 

How does the OPA627 stack against newer TI designs? I'd go with something like 1612 for line level and 1688 for lower impedance loads.

 

Gut feeling is that newer chips are better.

 

I'm finding that the OPA627 is everything people say it is and very literally so; it can be transparent, it can be dark, it can suck at bass, it can better most anything else, it is overpriced and overrated. In the circuits I have dropped the OPA627 into it has exhibited all of these traits. It's most definitely a try it and see. The AD8610 was a big meh for me; it also does not like a 15v supply voltage. The AD4627-1 is supposed to be "far superior" but in my m^3 amp I ended up preferring the opa627's for L and R and 4627-1 for G.

I think the opa627 is maybe still unsurpassed in medical/scientific applications where the cost is trivial if that means anything.

The tweak here for biasing I have tried with a resistor across pins 4 and 6 has been slightly positive; I didn't even realize its benefits until removing the socket with the resistor on it.

 

The price of the OPA627 has more to do with the manufacturing technics of this old design than of demand. The OPA827 is the modern and cheaper alternative that newer designs are using.

So yes, the price is just silly.

 

Has anyone actually produced data on performance vs bias?
Not because I think we'd get the be-all-end-all blah blah, but the number wanker in me does like to see pretty graphs.

If I even remotely had the knowhow, I'd do something like x-axis = frequency, y-axis = bias, z-axis = THD (or some overlaid combo of different types of distortion whatever). Like something hacked to look similar to a spectral or waterfall plot since we kinda understand those. Of course, this would be at some randomly standardized load and operating value, otherwise I'd have a bajillion different charts. I'm a numbers geek, but I'm not a masochist. At least, not unless someone's paying me to do it.

 

Sorry had to bump this thread, I've now used this mod in a few different applications and I hate to admit it but the opa627 sounds far better with this mod than without. Previously I had tried lower biasing but now I've pushed up to 6ma @ 12v and 8ma @15v with 1.8k resistors. The chips do start to run hotter but not alarmingly so, nowhere near the datasheet limits. The sound becomes smoother up top and less bloated in bass each time. Pulling this off requires a few resistors and dip sockets and the soldering is a pain in the ass but at least with the dip sockets the modification is a plug and play trial.

I wouldn't recommend judging opa627 as-is without giving the bias modding a try first. Yeah the mod looks stupid but opa627 now sound like ass to me otherwise.