thx Bernhard 🙂
I still have a handful of TDA2030 and 2050, why don't we just use two power operational amplifiers for a practical laboratory test? Wouldn't that make much more sense than ..?
We are making a small audio power amplifier together with our grandchildren according to the well-known principles of feedback and feedforward in combination. A small, completely error-free reference amplifier.
Theoretically, we could also cascade a TDA2030/50 n times to be really error-free at the end of the cascade, with a THD of minus infinite dB.
HBt.
("noch kein Opa")
Unfortunately: No.
The amplifier will never exceed the primary classA amp.
Only the distortion of the classB stage could be canceled completely in theory, not the distortion of the whole setup.
Good opamp + classA buffer as primary amplifier will always be better than TDA...
Most opamps are compensated in a way that their gain roll off 20dB/decade, which is equivalent to the magic C. You don’t need an external C.
You can find the detail maths work from the paper mentioned in below.
Post in thread 'Current Dumping with OPAMP'
https://www.diyaudio.com/community/threads/current-dumping-with-opamp.420292/post-7856231
Last edited:
Do you realize that (then) there will be (also) no magic bridge? The voltage between two potentials is always zero if these potentials are equal. There is therefore no hidden bridge configuration responsible for the acceptably low THD.
I simulated your schematic and it did not work well for me, but maybe I used the wrong opamp.
About -85db, compared to -120 dB with C.
CD25W
If you like,
try integrating the magic bridge with the circuit from the link above so that we can finally put the matter to bed.
About this or that OP-Amp. Model you really don't have to worry about. I get identical results.
If you like,
try integrating the magic bridge with the circuit from the link above so that we can finally put the matter to bed.
Do I understand that correctly? An improvement of 35dB through the use of a C, i.e. a kind of integration or a hidden low-pass filter n. order? For example, let's say f0=20kHz with f3=60kHz ... and our hidden filter already attenuates by xdB at 3*fo. No, you can't mean it like that.
So there is a miracle bridge after all, as propagated by the other side, for example. Because our faulty buffer, with its amplification factor of less than 1, is located precisely in this bridge, connecting the two lateral posts, the identical potentials of the complex voltage dividers, “complex” because there are reactances.
But the foundation of this bridge is our Rload. The bridge (posts) stands on the load resistor, our loudspeaker.
10mA in the output stage.
Depending on small circuit details the THD plot can get very bumpy.
No more switching.
Schematic:
Depending on small circuit details the THD plot can get very bumpy.
No more switching.
Schematic:
Another quick look at the Malvar Paper of Brazilian University.
Ra=180Ohm
Rb=500Ohm
I hope everything is understandable, self-explanatory and error-free. 😱
Here is the circuit of jxdking post#1
-95dB for 15V0p seems credible but low level is bad.
I don't have the right models, so I used the BDs again.
And I tried to optimize the values without success.
-95dB for 15V0p seems credible but low level is bad.
I don't have the right models, so I used the BDs again.
And I tried to optimize the values without success.
My circuit has the classA buffer which isolates any load from the opamp.
I think we are missing the point in this discussion (apologies if I'm treading on anyone's toes here). If you look at the published specs of the Quad 405, they are not spectacular by modern standards where on this forum ppm or sub-ppm distortion levels at high power in practical, built amplifiers have been demonstrated by numerous practitioners. However, the Quad CD amp had no adjustments in production! Further, due to the design, it would never need trimming during its working life (let's assume 20 years for a quality amp). This covered output bias current (typically 20-60mA in a class AB amp with all the attendant thermal tracking requirements etc) and DC offset.
If you read Self's amplifier design book, you can see how much effort goes into the thermal stability/tracking of a linear class AB amp. And if the designer/manufacturer doesn't want to trim out any DC offset with a pot, they will have to incorporate a DC offset servo. These requirements add cost (labour and material) and complexity. As I mentioned earlier, the Quad design dispensed with this stuff at a stroke, and that's the true genius of the concept.
I would expect that if someone went all out on a new clean sheet CD design today, the real performance of a modern version could be at the single-digit ppm level. The LM318 used in the original Quad amp is a dinosaur compared to modern devices, and the availability of modern sustained beta transistors (drivers and OP) would also pay dividends.
As to whether the technique works or not, the correct test of this is to drive the load with the unbiased output transistors and no FFWD resistor. The distortion will be spectacularly high. Insert the FFWD resistor and it will drop to 0.01% or better - a very respectable figure for a mid 1970s amplifier!
🙂
If you read Self's amplifier design book, you can see how much effort goes into the thermal stability/tracking of a linear class AB amp. And if the designer/manufacturer doesn't want to trim out any DC offset with a pot, they will have to incorporate a DC offset servo. These requirements add cost (labour and material) and complexity. As I mentioned earlier, the Quad design dispensed with this stuff at a stroke, and that's the true genius of the concept.
I would expect that if someone went all out on a new clean sheet CD design today, the real performance of a modern version could be at the single-digit ppm level. The LM318 used in the original Quad amp is a dinosaur compared to modern devices, and the availability of modern sustained beta transistors (drivers and OP) would also pay dividends.
As to whether the technique works or not, the correct test of this is to drive the load with the unbiased output transistors and no FFWD resistor. The distortion will be spectacularly high. Insert the FFWD resistor and it will drop to 0.01% or better - a very respectable figure for a mid 1970s amplifier!
🙂
My NE5532 model doesn’t model its class B output stages. I would say my simulation is over optimistic. In a real life, -70dB to -80dB thd is more realistic with class B feed forward stage. To get more than that, you really need to build a dedicated class A stage for feed forwarding.