Keantoken's CFP cap multiplier

TNT

Member
Joined 2003
Paid Member
I had to check mine using the dirtyPCB boards. I could not see such oscillations as your pictures. But I found something interesting... I hooked up my oscilloscope to my playing DAC and saw some traces of the music signal in the DC coming out of the KM. My power path looks as: 230 mains -> 2*Meanwell for +/-18V -> A half decent +/- stab from China 1A (LM317?) for +/- 12 -> The KM -> Soerkris DAC. The China stab board takes away almost all traces of SMPS switching blips. But noise is present ... 100mV p-p. After the KM the noise is a bit lower but the DC varies with music which it does not on the KM inlet.

I know that the KM is not a stab but was surprised it sagged this much.. Is it OK?

Short vid attached... .asc -> .mp4 to see it.

I hit Play 1 second in.

//
 

Attachments

  • WP_20160315.asc
    1.1 MB · Views: 91
Last edited:
What's with these new oscilloscopes having cryptic labels so you can't figure out what the voltage is or the timebase setting?

Check the voltage across R1 on the KM, it should be about 35mV or so. A lot of SMPS noise could get lost in 100mV of noise... I'm not sure it's the KM in this case, I would check the LM317 has capacitors it's stable with and if that's not the problem, replace it.
 
I had to check mine using the dirtyPCB boards. I could not see such oscillations as your pictures. But I found something interesting... I hooked up my oscilloscope to my playing DAC and saw some traces of the music signal in the DC coming out of the KM. My power path looks as: 230 mains -> 2*Meanwell for +/-18V -> A half decent +/- stab from China 1A (LM317?) for +/- 12 -> The KM -> Soerkris DAC. The China stab board takes away almost all traces of SMPS switching blips. But noise is present ... 100mV p-p. After the KM the noise is a bit lower but the DC varies with music which it does not on the KM inlet.

I know that the KM is not a stab but was surprised it sagged this much.. Is it OK?

Short vid attached... .asc -> .mp4 to see it.

I hit Play 1 second in.

//

Did you check for music without the KM in place?
 
I'm just catching up on the recent activities here and understand that you've sent off the pcb design (as per post 785) for a prototype version am a bit puzzled by the missing component/networks of D3, D1+R2, D2+R8, and D4 - are they not necessary anymore or simply tacked on below the board later?

If I follow the same numbering system on the explanation page, your pcb appears to have an extra capacitor on the input with a series R (top edge) to make up a C-R-C with the C3, onto R1, C1 and the base of Q1 (bc337), yes?

So, theoretically anyway, to complete the supply, you'd just need a simple pcb with the diodes and secondary winding snubbers, yes?

Would there be any advantage to move this 'series R' down between the first cap and the C3 side of the pcb, then move the R1 into the now vacant R position and simplify the tracks to the C1 and the base of Q1?

Also, would it any advantage to reduce the ground plane to a thick track that connects the ground terminals of the extra input filtering cap (plus the C3) to the main ground plane in order to provide some extra pcb impedance to input ripple - or am I talking nonsense (again!) ....
 
The protection diodes never seemed to work, since the majority of accidents seemed to be direct short discharge which will probably destroy the transistors anyway. Normal PSU operation doesn't seem to cause any problem for the KM so unless you want to use it as a bench supply in the trenches, the diodes aren't really necessary.

This board doesn't have a rectifier on it so yes, you will need one. Or power it from a regulated supply like some people do.

The amount of filtering you would gain by removing the ground plane will be insignificant compared to the output inductance of the KM and the inductance of the wires you power it with.

I think the traces are already as simple as they can be. The 1k/C2 filter feeds the LED which feeds the transistor base.
 
Sounds like your circuit is overloaded, or maybe Q1 has C and E swapped. Also check the 47R resistor with a DMM to make sure it's the right value.

It seems the BC337 has been damaged, because It worked after I've changed it.
Now it works.

What do you think about changing the 3x1N4148 into 1x1N4148.
Does it any effect beyond decreasing filtering ablility in case of noise with bigger amplitude (circa 0,5Vpp).
 
Using fewer diodes in your version to reduce voltage drop will at some point reduce PSRR by putting Q1 in quasi-saturation, which also makes it nonlinear which is worse than if it were just bad, because then it's generating extra harmonics.

If it works, it works, but it may not be optimal.
 
Using fewer diodes in your version to reduce voltage drop will at some point reduce PSRR by putting Q1 in quasi-saturation, which also makes it nonlinear which is worse than if it were just bad, because then it's generating extra harmonics.

If it works, it works, but it may not be optimal.

So it mean, that the K multi will have more noise on output than in the input?
The sound is still far better than without that.

What do you think can the D45VH10 be used instead of the D45H11?
Because the shop could gave me only VH10.
 
The PCBs arrived and tested fine. The only major problem is the output vias are swapped. They work perfectly fine.
 

Attachments

  • PKM-R1-Top.png
    PKM-R1-Top.png
    440.3 KB · Views: 548
  • PKM-R1-Bottom.png
    PKM-R1-Bottom.png
    399.1 KB · Views: 505
So far the lineup is:

Rcruz 4
Ryssen 4
jamesshillj 4

Let me know if I missed someone, or you want to pull out. Otherwise, please PM me your shipping information.

I can mark out the swapped output via text with a permanent marker or possibly liquid white-out and write the correct labels, unless someone has a better idea.
 
Yes, as I mentioned in the original post. Don't forget the LED.

To fill the board you simply look at the circled pins. They must be the same polarity as your output voltage. So if your output voltage is positive, all the circled pins are positive. It fits terminal blocks and works up to 50V. It could also be used with D4xH11G for 2A output or so.
 
Last edited: