Keantoken's CFP cap multiplier

Member
Joined 2006
Paid Member
It really depends on the layout, I think the most important thing is that there is a short path from the driver collectors to the zobel and drain grounds. So much of this is context-dependent, I can only make the most general comments, and I still may be wrong.

I am using separated paths for the zobel gnd and other grounds... they will all join star ground on the psu board. Somehow I believe zobel return should not be connected directly to other gnd. Am I wrong ?

In fact I see the zobel as another load to the power amp and as we normally route speaker return to the psu star, I decided to do the same for the zobel.
 
Last edited:
Member
Joined 2006
Paid Member
I rearranged the layout using upright diodes.

Please let me know your opinion.
 

Attachments

  • P300 KMult 3.PNG
    P300 KMult 3.PNG
    144.7 KB · Views: 436
That webpage was nice more than 5 years ago but I think it is a bit out of date.

PSRR will be fine if your load current is not high enough, but output impedance will be terrible - 2R or higher.

Also, note that the important thing is really that Q2 is turned well on. The 2SA1930 might just have 0.56R Vbe(on), which would be fine. If your load current is more than 25mA, it will be fine.

Hi Ken,

I´m planning to build a 20W class A power amp and I saw you thread and web page on K-mult.

May I use equivalents of D44/5 trannies?

Is the R8/9 use for high current surges in your web page valid in this case?

regards,

My guess is that equivalents will work, but I cannot guarantee it. R8/R9 are not used in the PCB version because I found them to be unnecessary.
 
Last edited:
I am using separated paths for the zobel gnd and other grounds... they will all join star ground on the psu board. Somehow I believe zobel return should not be connected directly to other gnd. Am I wrong ?

In fact I see the zobel as another load to the power amp and as we normally route speaker return to the psu star, I decided to do the same for the zobel.

I see the issue of layout as a bunch of intersecting and nesting current loops. You look at which current loops are both important for operation and sensitive to layout (or current loops which would emit large EMI if not controlled). The Zobel current loop is one such important loop. This loop goes through the MOSFETs, through the rails and through the closest decoupling cap to the zobel ground which completes the loop. A physically small loop here is good for stability. Feedback should come directly from the zobel, otherwise the amp sees the zobel as inductive, which is also bad for stability.

I see the layout of a power amp as a 2-wire system. The wires should stay together. The two wires start at the feedback resistor and the feedback ground. Next the pair of wires goes to the zobel (which is connected to the output stage and decoupling). Then the pair of wires go to the speaker. I think the zobel ground makes a good star point.

Sometimes people think it's better to put one ground away from another thinking if it's "over there" that it won't get the other one dirty. The real solution tends to be the opposite when we are talking about stuff on a power amp PCB. It is often better to put such a node in the center of the design, where nothing is too far away and it can be controlled.
 
Last edited:
That webpage was nice more than 5 years ago but I think it is a bit out of date.

PSRR will be fine if your load current is not high enough, but output impedance will be terrible - 2R or higher.

Also, note that the important thing is really that Q2 is turned well on. The 2SA1930 might just have 0.56R Vbe(on), which would be fine. If your load current is more than 25mA, it will be fine.

My guess is that equivalents will work, but I cannot guarantee it. R8/R9 are not used in the PCB version because I found them to be unnecessary.

Tks, Ken!
Best regards,
 
Member
Joined 2006
Paid Member
I see the issue of layout as a bunch of intersecting and nesting current loops. You look at which current loops are both important for operation and sensitive to layout (or current loops which would emit large EMI if not controlled). The Zobel current loop is one such important loop. This loop goes through the MOSFETs, through the rails and through the closest decoupling cap to the zobel ground which completes the loop. A physically small loop here is good for stability. Feedback should come directly from the zobel, otherwise the amp sees the zobel as inductive, which is also bad for stability.

I see the layout of a power amp as a 2-wire system. The wires should stay together. The two wires start at the feedback resistor and the feedback ground. Next the pair of wires goes to the zobel (which is connected to the output stage and decoupling). Then the pair of wires go to the speaker. I think the zobel ground makes a good star point.

Sometimes people think it's better to put one ground away from another thinking if it's "over there" that it won't get the other one dirty. The real solution tends to be the opposite when we are talking about stuff on a power amp PCB. It is often better to put such a node in the center of the design, where nothing is too far away and it can be controlled.

Thank you.... I must modify the layout according to your input.

I have been studying the circuit corresponding to your boards. I have built those and they do perform very well indeed.

Can I use the same circuit in my power amp (+- 62v rails) ?

I breadboarded the high voltage circuit (the one in your site with a lot of diodes) and it works also but I am having trouble including it in the power amp board... (too many components) so I would like to use your board circuit instead.
 

Attachments

  • kmult board sch_edited.jpg
    kmult board sch_edited.jpg
    154.5 KB · Views: 340