It uses the the bias network to do error correction. It is equivalent to CFP with MOSFET output, but it can output from rail to rail, with help of the bootstrapping.
The example below is +-14Vp into 8 Ohm with only +-15V supply voltage. You can see the distortion is low. Note that you could put opamp at the front and wrap NFB around it to reduce distortion farther.
The example below is +-14Vp into 8 Ohm with only +-15V supply voltage. You can see the distortion is low. Note that you could put opamp at the front and wrap NFB around it to reduce distortion farther.
The correction circuit looks a little different from that of Cordell's....
And you do without drivers for the HEXFETS - is that really OK?
I realize that this is not a finished design, though! 🙂
And you do without drivers for the HEXFETS - is that really OK?
I realize that this is not a finished design, though! 🙂
Right, the original design from Bob Cordell is rather complicated. I don't expect I can fully comprehend his design intention. There is only 2 options, one is stick with the original design which is proven working, the other one is start from scratch based my own understanding.
In this circuit, I would leave out the driver for HEXFET. There is NFB back going on. I would like to keep the components count inside NFB loop minimum so that it could be unconditional stable without any compensation.
It is equivalent to CFP with MOSFET output. You don't really need adding driver transistor in CFP. It will work without the driver.
What changes have to be made to have a decent output power, say 50W or more? Since the OPS is inherent rail-to-rail, I guess a +/- 15V OP is still sufficient to give any output swing?
+-15V is only good enough for about 14W with 8 Ohm.
+-30V would be good for 50W.
What power supply are you going to use?
The only thing that need to be changed is the bootstrap portion R8~R11. Keep the current of them around 7mA according to your supply voltage.
24V-26V-28V maybe...
So 30-some volts and 1K8...2K2 for the resistors? I guess they remain equal value also with raised supply voltage?
The circuit is so simple, it is tempting to give it a try in KiCAD! But no promises....
So 30-some volts and 1K8...2K2 for the resistors? I guess they remain equal value also with raised supply voltage?
The circuit is so simple, it is tempting to give it a try in KiCAD! But no promises....
Looking forward to your design - if it is published!
Is 1A quiescent current ok for the shown buffer? Maybe 0,5A?
What were your operating conditions?
Is 1A quiescent current ok for the shown buffer? Maybe 0,5A?
What were your operating conditions?
Thanks.
As I said, it works like CFP, Q5, Q6 sense the difference between input and output (AC signal), and convert it into current, and use the current to drive the output device.
I found it bias like CFP, too. 50mA or so is the best. Over biasing it worsens the distortion as it widens crossover region.
There are some minor fixes that has to be done. It is a very good topic so i would increase the current in the Q5678 to get a higher slew rate.
Gate protection zeners for Q3 and Q4 are also good to have. And then it would be good to consider current limiting resistors in series with Q7C and Q8C.
How does it work with Q7C and Q8C to the output?
Gate protection zeners for Q3 and Q4 are also good to have. And then it would be good to consider current limiting resistors in series with Q7C and Q8C.
How does it work with Q7C and Q8C to the output?
This looks nice! All transistors are at hand except for the ZVP4424A and ZVN0124A. Any substitutes for these?
They can be replaced with 2n5551/2n5401 for prototyping. They have the same pinout as the ZVN/ZVP ones.
I want to keep R13-R16 in quarter watt package.
I found push pull mosfet configuration was pretty robust in the past of my design. It is not easy to overvolt the Vgs as the voltage is aways clamped down by the bias network. Unless, here, Q7, Q8 blow up at first. Even though, the fuses on the power supply may blow before Vgs protection kicks in.
👍, I will investigate this.
The IRFP140 I have starts to conduct at 3V. The IRFP9140 starts to conduct at about 3.5V. Thus, it won't work if I tie them to the output.
Here is take 1 on the schematics, added a few caps and diodes.
I couldn't figure out direction of the bootstrap caps. Advice?
I couldn't figure out direction of the bootstrap caps. Advice?
Looks good.
PS: I won't bother to additional power filtering for input stage. The Power Supply Rejection Ratio is high enough already.
They are vertical MOSFET. They tend to thermal run away by themself. You might get away without them with perfect thermal coupling. I would put them in the design. You can always short them on PCB later on.