Not a good way to drive a output buffer, clipping and drive will always have DC offset.
Wont swing well and the opamp is always being pulled up by the bias network,
As warned in the write up, pure junk will cause almost any opamp to oscillate.
Specially with plain wrong compensation.
CFP outputs can turn on no current to turn off with the circuit shown, with the DC offset, like most elliott circuits = oscillators
Is somewhat a good idea since CFP only needs one VBE drop.
Easier to use transistors as diodes to thermal track the drivers. And for opamp buffers drive the diodes centered.
With current sources on both sides, so the swing has no offset and the opamp isn't being pulled up constantly.
I made CFP diamond in one thread one way to do it, can show transistor as bias with old circuit of mine and many others.
Will post can be used for low or high voltage op amps.
With the elliott spirit and his love for capacitor bootstrap. ill toss those in as tribute to him lol.
This is not a bug, it is a feature. It forces the opamp to work in class A. I used this technique to get low distortion with some basic opamp like LM358 in the past. The tradeoff is the unsymmetrical output voltage swing.
Last edited:
Transistors as Diodes, CFP output 2x pairs for relentless drive into 4 ohms.
Using LM1875 as dirt cheap high voltage opamp. Using +/_ 28 volts like LM3886
Does same about 38 watts 8 ohm 65 watts 4 ohm.
T1,T2,T3,T4 all mounted on same heatsink for thermal tracking, like any other CFP
Stable 47 degree phase margin
1K
8 ohm 1 watt .0008 % distortion
8 ohm 35 watt .001 % distortion
4 ohm 1 watt .001 %
4 ohm 60 watt .008 %
Using LM1875 as dirt cheap high voltage opamp. Using +/_ 28 volts like LM3886
Does same about 38 watts 8 ohm 65 watts 4 ohm.
T1,T2,T3,T4 all mounted on same heatsink for thermal tracking, like any other CFP
Stable 47 degree phase margin
1K
8 ohm 1 watt .0008 % distortion
8 ohm 35 watt .001 % distortion
4 ohm 1 watt .001 %
4 ohm 60 watt .008 %
I just dont like designs with a lot of components
Same circuit, 2 diodes and a output section.
Just 2 current sources for symmetry.
Of course it doesn't oscillate, and does 65 watts instead of 12 watts.
.0008% instead of .1%
Not 2 bad for extra 2 resistors and a cap
Just 2 current sources for symmetry.
Of course it doesn't oscillate, and does 65 watts instead of 12 watts.
.0008% instead of .1%
Not 2 bad for extra 2 resistors and a cap
Is this OPA134 a FET or BJT input opamp? Isn't there a capacitor between R16 and GND missing?
Best regards!
Best regards!
FET or no FET you still want that cap to force unity gain at DC. Unless you want to add a servo.
I'm not sure any of the designs posted here asks the important question.
"What does using an OPA as input gives us?" My answer is ...
It needs a Power Transformer/Supply for each channel but this is hardly a problem considering the people here who swear by monoblocks ... oh and a little +/- 15V supply which could just be a little extra transformer. Having 2 big and one little Transformer for a stereo amp would make an amp that looks like a vintage valve amp 😉
I don't think John has taken it to its full potential. There are new supa dupa OPAs which could bring this to 1ppm THD @ 20kHz and full giarnomous power .. but all this needs to be investigated ie tried out. His OPA input could be used without change from small to HUGE power amps.
I'm no longer an AES member these days but a real beach bum. I've lost this and loadsa other stuff in several HD crashes including my own papers
If one of yus AES members is willing to risk the wrath of the AES SchutzStaffel, may I beg a(n illegal) copy of this ... purely for research and education of course. 😊
"What does using an OPA as input gives us?" My answer is ...
- Zillion dB gain so you can get 1ppzillion THD easily without extra complexity
- low input current & matched offsets so you don't have to twiddle offset voltage. BTW, that's why QUAD used TL071, rather than 5534, in the 405. It was for the low input current & offset
- so much extra complexity that you could do a 1ppm @ 20kHz full power design with the extra bits without the OPA
- simple but really crap performance .. especially xover yuckies
A Simple Reliable Power Amplifier with Minimal Component Count
He didn't invent it cos it is essentially Hafler Transnova but stripped to the bare essentials without losing any of its advantages or any performance.It needs a Power Transformer/Supply for each channel but this is hardly a problem considering the people here who swear by monoblocks ... oh and a little +/- 15V supply which could just be a little extra transformer. Having 2 big and one little Transformer for a stereo amp would make an amp that looks like a vintage valve amp 😉
I don't think John has taken it to its full potential. There are new supa dupa OPAs which could bring this to 1ppm THD @ 20kHz and full giarnomous power .. but all this needs to be investigated ie tried out. His OPA input could be used without change from small to HUGE power amps.
I'm no longer an AES member these days but a real beach bum. I've lost this and loadsa other stuff in several HD crashes including my own papers
If one of yus AES members is willing to risk the wrath of the AES SchutzStaffel, may I beg a(n illegal) copy of this ... purely for research and education of course. 😊
Last edited:
Hi Richard
This kind of output circuit does mean that it has to drive the floated power supply to earth capacitance. This always worried me, partly because I don't know what a realistic value would be, so I am concerned it may be just what the amplifier feedback loop doesn't want. Any opinions or data?
What "new supa dupa OPAs" did you have in mind? I still haven't seen any better than the old classic AD797.
Best wishes
David
This kind of output circuit does mean that it has to drive the floated power supply to earth capacitance. This always worried me, partly because I don't know what a realistic value would be, so I am concerned it may be just what the amplifier feedback loop doesn't want. Any opinions or data?
What "new supa dupa OPAs" did you have in mind? I still haven't seen any better than the old classic AD797.
Best wishes
David
I enjoy using the Onsemi D45H11 and D44H11. Great for audio. Great drivers to the output section.
Another good way to use an op-amp is to use the supply pins as outputs. The issue is that the op-amp supply current is critical, so you have to adjust resistors for your choice of op-amp.
This will be in the region of several nF. Extrapolating from Real Life experience 😲 with 'conventional' amps, this is unlikely to be the most critical load for stability. Of course if you are doing full power testing at supersonic frequencies you might have a problem 😲
Anyone want to measure capacitance of their large transformers from secondary to earth/chassis while mounted as in use?
Another big benefit of Vanderkooy's circuit is the efficient use of VFETs. As 'good' power BJTs & VAS drivers are now only available from Ye Olde Unobtainium Shoppe, VFETs are about the only options left.
They are really difficult to drive well in 'conventional' circuits. IIRC, his is a very elegant topology .. far simpler than Hafler's TransNova.
But as I no longer have a copy of his paper, I may be pontificating from the wrong orifice ... hint Hint HINT 😊
Dis beach bum has been out of touch with modern OPAs since the days of pontificating on astx's amps and crossing swords with Guru Wurcer (RIP). A few HD crashes trashed most of my carefully gathered data and there has been loadsa new developments while I've been occupied by beach bum stuff ... and sadly health issues too.
I've revived my interest in OPAs for a completely different application; simple MM RIAA preamps. Not sure why I'm obsessed with such stuff as I'll never build another big amp and no longer have a vinyl collection. About the only thing I've got from this latest round of cerebral wanking is to avoid certain OPAs like LM 4562 ... despite what Self and other (pseudo?) gurus go on about
Last edited: