Class-A Opamps

[tiefbassuebertr]

Originally Posted by dimitrij

class A biasing for opamps seems a nice idea. I will give it a try for my new preamp that i'm gonna build soon.
A (maybe) stupid question: will it work also for power opamps?
I'm talking about LM1875 and similar devices.
Has anyone tried this yet?

If a voltage gain is absolutely necessary (e. g. in cases, where there is only a buffer stage with unity gain as power amplifier), then this is a very good idea.
In all the other cases (i. e. if the power amplifier uses voltage gain stage factors between 10 and 100) I would prefer discrete preamps with only one stage in the signal path (source follower e. g.) - for an example go to
http://www.passdiy.com/pdf/B1%20Buffer%20Preamp.pdf
or for more output current by use of multi-ampping systems
http://www.audiodesignguide.com/my/Follower_99c.gif
(use an idle current in such cases between 100mA and 200mA)
If there is a short connection to power amp with high input impedance above 20 K-Ohm, then I would prefer a passive pre-amp with only volume control and source select rotary switch.
check out therefore this thread:
Potentiometer Overview - Alps, Penny+Giles, TKD and Vishay and more, what's your fav?
most guys like for preamps circuits like this:
Jfet audio buffer
but I am not a friend of such topologies for the use in normal line stages applications where low gain factors realized.

[dimitrij]

well, don't know about the sensing circuit, but i have a stereo LM1875
amp (based on datasheet circuit), but it looks to me that it does not boil at all.
mine is built with decent heatsink and runs very cool. I never
measured its temperature, but i'm sure it runs at around 40° C...

Unfortunately i'm not brave enough to test if we can increase its bias with
a resistor... (read: i'm not well aware of what will happen, so i don't want to risk to blow a beautiful sounding amp)

[ontoaba]

Most OTA is class A, but all low open loop gain; (CA3060, LM13600). They has highest linearity at open loop.
LM13600 is buffer attached inside and could be instantly paralleled.

[janneman]

Originally Posted by AndrewT

I don't think so.

let's use a simplified example of an opamp that has the 3 stages of the Lin topology and the last stage is push pull.

The first stage draws Quiescent current Iq1
The second stage draws Quiescent current Iq2
The third push pull stage draws Quiescent current Iq3 = Ibias

Let's further assume that no other quiescent current is drawn by auxiliary circuits.
Iq1 & Iq2 could be feeding single ended ClassA stages.
The total quiescent current (Iq) is Iq1+Iq2+Iq3
The maximum ClassA output of the Lin amp is 2*Iq3
We have many combinations of varying current in the 3stages as the output current is increased towards IclassA max.

One must account for Iq1 & Iq2 before one can draw conclusions about IclassA max.

Then the opamp will have a non zero Iauxiliary. More complication

Opaq's method relies on the not unreasonable assumption that the rail currents of a ClassA amplifier follow the ClassA output current.
It appears he has detected the change in rail currents to include the half wave rectified versions of the ClassB output current.

Until now I had assumed that the bias current in an opa output stage would be about half the total supply current. OPAQ's results seem to indicate otherwise.
Anyway, an easy way to increase an opamp class A output current is to follow it with a buffer like the OPA634 or LM49600. You can set the buffer bias current often up to 20mA which means that it is in class A up to full output in less than 1k.
You take the opamp feedback from the buffer output so the whole thing acts like a hefty class A output opamp.

jan didden

[jacco vermeulen]

Originally Posted by janneman

easy way to increase an opamp class A output current is to follow it with a buffer

At the risk of getting poo-shot, the LME49600 is much better than the BUF634 with raised BW connection.

(a bag of BUF634 by the cheap in multi-parallel/bridged mode would be nice though)

[jcx]

my parallel amp push-pull biasing scheme shown earlier in this thread should work with power chip amps as well as op amps

there is also Doug Self's "Class XD" which could be applied to power chip amps


on the smaller scale, some early generation A/DSL driver op amps have bias current setting/boost pins, presumably most of the extra bias is going to the output stage, I believe a few 10s of mA Class A output is possible with some of these chips

they were designed for low distortion (some down to -100 dB) at 100 KHz - low MHz into 25-100 OHm loads

newer ADSL chips have apparently been designed for much lower quiescent current, according to Scott Wurcer the industry has come to accept ~80 dB distortion numbers as adequate so their outputs are low bias again