Current Dumping revisited

... The output gain is thus lumpy across the voltage swing and crossover distortion is higher than standard class B (for large swings).
What is the incremental benefit of a lumpy output gain and higher than standard cross-over distortion?
I see the con's ....
But where are the 'pro's ?

Edit: cross-posting.
 
Last edited:
AX tech editor
Joined 2002
Paid Member
What is the incremental benefit of a lumpy output gain and higher than standard cross-over distortion?
I see the con's ....
But where are the 'pro's ?

Edit: cross-posting.

The motivation behind this was to produce an amplifier that has the same performance as a classical class AB but without the need for a factory adjustment and with no issues with thermal stability.

And there is no 'lumpy output gain' and no more xover distortion than in a classical AB. You really should read that analysis I posted above, it's a great piece of engineering ingenuity.

Jan
 
AX tech editor
Joined 2002
Paid Member
Where's that capacitor? Is it that C6 of lumpy 115 pF?
Best regards!

33nF across the opamp supplies.

BTW I understand the thinking for it but it is counter-productive. Since we are only interested in the current coming out of those pins, there is no need to fix the voltage across those pins. And in practice that cap increases distortion and decreases stability as it injects whatever is on one rail into the other thus forming a positive feedback loop. Not good.

Edit: I thought you were referring to the 33nF Mark mentioned, but probably not.

Jan
 
Last edited:
What do diodes D9 and D10 do besides decreasing maximum output swing?

You are right, they aren't necessary.


And yes, I'm also somewhat sceptical about sensing the opamp's supply current and amplifying it's swings. There might be much better ways to translate signal swing from the opamp's +/- 15 V to the rails' level. For instance, look at how RCA has it done in the Power Transistor Manual or how ELEKTOR themselves did it in their 300 W PA amplifier.
Best regards!

Well, the opamp's output current also flows through the rails...

But I'll simulate the Elektor 300W way.

Thanks.
 
33nF across the opamp supplies.

BTW I understand the thinking for it but it is counter-productive. Since we are only interested in the current coming out of those pins, there is no need to fix the voltage across those pins. And in practice that cap increases distortion and decreases stability as it injects whatever is on one rail into the other thus forming a positive feedback loop. Not good.
Jan

It is interesting that the THD is lower with that 33nF. It seems to be a tuning thing, because the best value for that capacitor is 33nF!
 
Do you mind sharing the circuits in plain graphics? Thank you!
Btw, RCA did the level translation with two transistors in common base configuration, see here for instance.

Best regards!

See attached.
 

Attachments

  • Current Dumping Sch.jpg
    Current Dumping Sch.jpg
    180.4 KB · Views: 583
  • Current Dumping PA300.jpg
    Current Dumping PA300.jpg
    169 KB · Views: 598
Not sure what's up but "Current Dumping Sch.asc" isn't working for me. All other examples do.

This one depends on the correct op-amp model including the rail pin currents. The others don't. It is common for 3rd party op-amp models to be incomplete. I have not found a LTC model that had this problem but most of the 5532 etc on line models are bad.

None of these circuits uses feedback to the op-amp output, which can be a useful topology since the delay between the output pin and the rail pins is very quick. But this requires a very accurate spice model.

Has anyone discussed the effect of shorting the output on the op-amp compensation. This can actually be useful because it can effectively cancel the op-amp internal compensation.
 
None of these circuits uses feedback to the op-amp output, which can be a useful topology since the delay between the output pin and the rail pins is very quick. But this requires a very accurate spice model.


The original circuit used that feedback. Some others I saw too.

But every simulation I did showed increased THD when using it.
Even a resistor from the op-amp output to ground increased the THD.
 
I've been rethinking some bit. If that 33 nF cap across the opamp's supply pins really is crucial for the Current Dumping design, neither the ELEKTOR nor the RCA frontend would work correctly, as both feature a fix and regulated opamp supply and are missing this cap. I'm really curious about the sims ;).
Best regards!
 
AX tech editor
Joined 2002
Paid Member
I've been rethinking some bit. If that 33 nF cap across the opamp's supply pins really is crucial for the Current Dumping design, neither the ELEKTOR nor the RCA frontend would work correctly, as both feature a fix and regulated opamp supply and are missing this cap. I'm really curious about the sims ;).
Best regards!

It has nothing to do with current dumping, it is just a normal decoupling of the opamp supply.
Did you read any article about the current dumping?

Jan
 
This one depends on the correct op-amp model including the rail pin currents. The others don't. It is common for 3rd party op-amp models to be incomplete. I have not found a LTC model that had this problem but most of the 5532 etc on line models are bad.

I can understand that but I don't know what TL071 model was used. I tried the TI one that has been around since 1989 and one from ST. Neither work.
 

Attachments

  • TL071_sub.zip
    1.8 KB · Views: 52