Audio Asylum Thread Printer
Get a view of an entire thread on one page
|For Sale Ads|
Some people have posted about forcing op amps to run in class A mode. I did not think that this was possible. If it is, I would very much appreciate someone enlightening me on how it is done. Thanks in advance.
As has been noted, the purpose of running an op-amp output stage in Class A is to run it single ended, using only the (usually faster) NPN transistors.
By drawing a current referenced off the negative supply, from the output stage, the op-amp will be biased to output positive current equal to it to achieve a net voltage output of 0 volts with no signal present. This forces the NPN portion of the output stage to carry the entire signal without reaching the zero current point, as long as the signal output current does not reach that of the load current amount .
The less this current load varies, the less it will be modulated with the signal loading. Hence, a true current load is going to sound and work better. A simple resistor load, while electronically simple and fool-proof, has the least steady current loading, as it's loading varies anytime the supply voltage varies. A FET based current load will have a much steadier current under varying conditions, and therefore work and sound better.
Some notes on implementing a current load on an op-amp:
Figure the loading on the op-amp now, whether internal or if it is the source amp for the output of a component. What is the next stage or component input impedance?
Figure how much current will flow with a 2 volt output (maximum for a CD player, and typically the hottest line level signal in the chain), lets say as an example, that a CD player is faced with a 10 kohm input impedance for a preamp. 2 V out into 10 kohm is 0.2 mA. Now, you want to bias the op-amp with no more than about twice this current level, or approx. 0.4 mA. To draw 0.4 mA from an op-amp, you would connect a 37.5 kohm resistor to the negative rail, assumming a 15 volt PS.
This much bias will keep the op-amp output in Class A up to a 2.8 V RMS output level. Now, there is the issue of peak RMS voltages, and so on, and not getting too close to zero output current when the signal drives against the current load, so twice as much current as indicated for a 2 V output covers this nicely.
Remember, the total load on the op-amp will be the input Z of the next stage/component, in parallel wiuth the effective current load resistance. In the above cited example, the 10 kohm component load in parallel with the 37.5 kohm load will end up at approx. 7.9 kohm total load, so make sure that the op-amp can handle this amount of total loading in terms of the output stage.
Of course, just like many power amps, the harder/deeper you run an amp into Class A, up to 1/2 it's full current output, the sweeter it tends to sound (usually). But for op-amps, there is a very real issue of heat dissipation in the IC die, and keeping the thermal situation under control, so somewhere between the above recommended loading and 1/2 the max output current is going to be the place to experiment.
Using the above recommended 0.4 mA for most op-amps will usually sound better than operating Class AB, and not cause any over heating problems.
Personally, I have not found too many op-amps that need 2 mA to sound their best, as this level of bias is a bit much in most cases. Use of a 5 kohm ressitor to the negative rail will result in approx.3 mA of load current, and an effective loiadon the op amp fo the 5 kohm in parallel with the next circuit/component load. If this were 10 kohms, then ther total laod will be on the order of 3.3 kohms, which is a difficult load for most op-amps, in terms of maintaining absolute linearity and freedom from HF distortion.
So I would try to temper how much current load is placed on any given op-amp, especially op-amps that do not have a robust output capability.
One trick that can be used, is to help stabilize the load current, and you can do this by using a split resistance, say a 10 kohm resistor from the rail, then a series of by pass caps, and then from that node, a 27kohm resistor to the output/s. This filters any HF noise off the PS rails, and loads the op-amp with a 27 kohm AC load, and a 37 kohm DC load. This should not be necessary with a FET based current load or current diode, etc.
This issue is one of the main reasons the Walt Jung's special combo of Analog Devices AD744 and AD811 (or some other buffer) seems so interesting to me. One of the AD744 compensation pins presents an output signal BEFORE it goes though the internal class B output stage. You can feed this signal into another buffer stage of your choice for power. Why try to bias the output stage into class A with whatever problems that could possibly bring? Just don't even use the internal output stage at all! Walt's circuit takes advantage of the availability of this signal and feeds it into a AD811 high speed, high power video amp. This combo measures quite well and sounds quite good as a line stage or headphone driver. As far as I know, the AD744 is unique in having this output versatility.
Thank you for taking the time to give a very detailed description of trying this mod.
Actual implementation is trivial. Connect a 4.99 kohm resistor between the op-amp output and the -15V line. This comes from the POOGE Chronicles. I've tried it and it works well.
Also what are the offset null pins used for in single op amps? Does the value of resistor need to be played with to any large degree? What wattage rating on resistor? I assume metal film and matched pairs. Thanks in advance for answering these questions.
If I am using a dual Opamp (BB OPA2604) I connect a 5 kohm R between pin 4 (-V) between/across both pin 1 (output A) and pin 7 (output B)? Therefore 2 resistors required for each dual opamp?
Thanks in advance.
I've added a 0.25 Watt 5Kohm 1% resisitor on the opa2604's of my preamp. That is one resistor per 2604 chip, from pin 4 to pin 7. I didn't mount any heat sink, but up to now about 3 hours, everything is good.
Amazingly sweet !!
Thank you, all of you.
unless you get some heatsinking on that. It'll just burn out sooner on you, and still could just overheat on you soon. Get the heatsink and be safe. Op-amp's dont like to run very hot anyway.
They run really hot. Thank you Andy.
Your advice is reasonable. I'm now building some tiny heatsinks using metal fuse holders. I'll stick them to the chips using heat dissipating gel.
Thank you Andy !!!
The idea is to force the output stage of the [monolithic] slopamp out of class AB and into class A mode, by pulling sufficient current to cut off one side of the output stage and keep the other side cooking enough, but not too much. Works well from a technical standpoint, provided the output stage dissipation is set within reasonable limits, and lowers thermal tails [or memory effect] in the input and gain stages of the opamp, by providing a local thermal bath.
From a perceptual standpoint, most monolithic slopamps have enough other issues IMHO that there are diminishing returns. But worth the effort if within the constraints of keeping the same topology and not doing a more radical redesign. Generally, any audio function that uses opamps can be implemented without them, often with spectacular improvements, although more expensively, and therefore not appropriate as a mass consumer item. But for DIY, who cares about production issues?
Share and enjoy...
Please read this post
But you wouldn't be able to force an IC op amp to run in Class A. I mean say an OPA637 will never run in class A, right?
Yes you can. Pulling a fixed current from the output to one of the
supply rails effectively turns the IC opamp's output stage into
a single ended stage with constant (external) current source (or sink). Which is a class A topology.
For current source you can use, in ascending order or quality:
-monolithic current source IC
-FET or transistor current source, cascoded
This post is made possible by the generous support of people like you and our sponsors: