Juma's Easy-Peasy Capacitance Multiplier

Founder of XSA-Labs
Joined 2012
Paid Member
I have built many amps and have been working on improving my power supply ripple with the usual CRC or CRCLC circuits. My best effort was actually an Abletec SMPS followed by a CRC where I got 1mV ripple under quisescent bias current. With linear toroidal transformers, I could never get below maybe 5mV and that was with a lot of effort. However, in class A, start pulling a couple of amps and that rises to 20mV sometimes 100mV depending on your trafo and CRC filter setup.

Novi amp sa 2SK2013/2SJ313 - Solid State - diyAudio.rs

Juma's cap multiplier:

585920d1481786375-jumas-easy-peasy-capacitance-multiplier-juma-cx-schematic.png


Here is nice schematic of the basic core cap Mx for a single rail use by Jwksana:
36990363054_f4dbe6e8e2_z.jpg


I recently built a small class A designed by Juma based on the Pass F5 but using the little 2SK2013/2SJ313 MOSFETs in parallel.

F5 with 2SK2013/2SJ313

The amp works well and sounds very nice. It draws about 1.5amps continuously and I am feeding it with my usual class A power supply: a 400VA 18VAC Antek with a 33mF/0.12R/33mF CRC. I was measuring 20mV ripple per rail and audible hum was coming through when no music was playing.

Juma's plans for the little F5 also included a simple capacitance multiplier circuit. It was so simple I had to try it. I have tried others in the past but they were always more complex and typically used BJT's and Darlingtons that I did not have on hand. This one however uses a single IRFP240/9240 per rail, a few resistors and easy to find caps, and a diode. It was so simple I could P2P on the legs of the MOSFET sticking up in the air:

574338d1476275384-very-simple-quasi-complimentary-mosfet-amplifier-f5-j131-k2013-cap-multiplier-closeup.jpg


I powered it up, now fed from my usual CRC that had 20mV ripple under load. There is a 2v voltage drop so you will lose some headroom. However, if you plan ahead, just get a slightly higher voltage secondary rating transformer. Then I hooked it up to the amp and adjusted bias current to usual 1.5amps and now measured the ripple...

Cap multiplier under test with F5 amp and 1.5amps flowing:

574333d1476267068-f5-2sk2013-2sj313-f5-j313-k2013-capacitance-multiplier-works.jpg


I was floored to see the Fluke AC volts drop to 0.000V AC (across +/- rails at 40v total). Sometimes flickered to 1mV then back to zero again (Fluke 101 has 1% accuracy of measured value + 3, the number added to the least significant digit, data valid over 40Hz to 500Hz range). I would love to have an o-scope to see what it really is doing - but there was a significant improvement in the sound quality. The hum was gone. I will try in my class AB amps next but thought folks might like to try this. It is dead simple and so easy. Anyhow, I think I will be making many more of these in the future and a little PCB would be handy. Just wanted folks to know about what a great circuit this is and to give it a thumbs up! Thanks to Juma for designing it.

If you have any experience with cap multipliers, please share. I have seen the ESP one and the one by Keantoken. This one with a single MOSFET is the easiest one I have seen though.

Edit Nov 13, 2016: stereo cap multiplier on one heatsink.

579772d1479032560-jumas-easy-peasy-capacitance-multiplier-juma-cap-mult-closeup-02.jpg


Here is a really nice one by Olafk:

582178d1480173535-jumas-easy-peasy-capacitance-multiplier-psu_juma_ii_02.jpg


Nice 4 cap Mx on 100mm square PCB by Prasi:
640121d1507959350-jumas-easy-peasy-capacitance-multiplier-juma-cm-png


Gerbers on post 245

Here is another nice one by Prasi for use with external bridge block:
640344d1508086056-jumas-easy-peasy-capacitance-multiplier-juma-cm-external-diode-bridge-png


Small SMT/TO-220 Cap Mx (Dual rail) by Prasi:

714135d1541569887-jumas-easy-peasy-capacitance-multiplier-smd-lay-r1-png


720209d1544046993-jumas-easy-peasy-capacitance-multiplier-smt-cap-mx-build-03-jpg


Works very well.
 
Last edited:
  • Like
Reactions: 1 user
Founder of XSA-Labs
Joined 2012
Paid Member
Maybe these other MOSFETs would work even better in Juma's circuit design. They are rated for much greater current and much greater power dissipation.

_

Those are interesting MOSFETs. 80amps - wow! I doubt you can safely pull 80amps through a TO-220 without smoking it. How much heat dissipation is that? Assuming 2v drop x 80amp is 160w!

I don't think there is an issue with current yet - 12amps with IRP240/9240 is quite high already. I don't know too many amps that pull that much current.

Main reason for me is that I literally have a drawer full of them already. My favorite MOSFET. :)
 
Last edited:
Administrator
Joined 2007
Paid Member
These are very effective ripple filters, the only downside is the loss of headroom. This shows a load current of around 650ma. You can see how small the ripple is, but also the loss in headroom.
 

Attachments

  • Ripple.PNG
    Ripple.PNG
    49.9 KB · Views: 4,070
  • Ripple 2.PNG
    Ripple 2.PNG
    49.9 KB · Views: 3,497
Founder of XSA-Labs
Joined 2012
Paid Member
Hi Mooly,
Thanks for the simulation! Wow - that is indeed very effective and confirms my experience and data from the Fluke. I was wondering if it was just missing bandwidth but this appears to be a very sound topology. Loss of headroom probably not an issue for someone wishing to listen to music as lower volumes where hum is audible.
 
Most of Juma's amps have PSU with cap mutiplier. You can see attached some examples.
The last one is with BJT if you need less voltage drop (2.1V compared to 4V drop with MOSFETs).
 

Attachments

  • cmbjt.gif
    cmbjt.gif
    3.9 KB · Views: 4,164
  • C3-PSU-sch.gif
    C3-PSU-sch.gif
    8.3 KB · Views: 4,057
  • post-20-0-57622100-1298065429.gif
    post-20-0-57622100-1298065429.gif
    26 KB · Views: 4,055
  • cubie-cm.gif
    cubie-cm.gif
    10.2 KB · Views: 4,280
  • Cubie_PSU-AMP-PCB.gif
    Cubie_PSU-AMP-PCB.gif
    81.2 KB · Views: 19,866
Member
Joined 2011
Paid Member
Mooly's expression "loss in headroom" can be translated "enormous voltage drop from Vin to Vout; notice that Vout is 4 volts below Vin!!"

This is a result of using "standard" gate threshold MOSFETs whose Vth is about 4 volts. "Logic Level" gate threshold MOSFETs would reduce the voltage drop to about 2 volts, and of course bipolar power transistors would lower it to 0.7 volts. Keantoken's Complementary Feedback Pair ("Sziklai") capacitance multiplier has a Vin-to-Vout drop of about 1.4 volts, as does a two transistor darlington-connection.
 
Member
Joined 2011
Paid Member
IRFP's have a very low Rdson and perhaps have an advantage in instantaneous transient current response for more bass impact?

I doubt that your capacitance multiplier circuit would ever operate the MOSFETs in the Rdson regime (huge Vgs, tiny Vds) but if it somehow happened:

Code:
Device     Polarity   RDSon
-------------------------------
IRFP240    Nch        0.180 ohms
STP75NF75  Nch        0.011 ohms

IRFP9140   Pch        0.200 ohms
SPP80P06   Pch        0.023 ohms
 
................I don't think there is an issue with current yet - 12amps with IRP240/9240 is quite high already. I don't know too many amps that pull that much current......................:)
most 8ohms speakers can demand transients that exceed 12Apk.
4ohms will demand even more.
If the amplifier is any good it will pass that transient demand back to the capacitance feeding it.
 
Founder of XSA-Labs
Joined 2012
Paid Member
Administrator
Joined 2007
Paid Member
Hi Mooly,
Thanks for the simulation! Wow - that is indeed very effective and confirms my experience and data from the Fluke. I was wondering if it was just missing bandwidth but this appears to be a very sound topology. Loss of headroom probably not an issue for someone wishing to listen to music as lower volumes where hum is audible.

Your welcome :)

Here is the same set up but now with a 10 ohm load that is being pulsed. The rise and fall times of the pulse are slow (100ms). Curious that there seems to be some oscillation in vout now.

The current in the load is at the right hand scale.
 

Attachments

  • Ripple3.PNG
    Ripple3.PNG
    58.5 KB · Views: 4,179
Founder of XSA-Labs
Joined 2012
Paid Member
Mooly,
I think this is why I have heard that the K multipliers lose their effectiveness for class AB dynamic loads, but in class A where the current glow is pretty stationary it makes a big difference with line induced ripple hum. Would it help to make C2 33mF and C1 2.2mF? Maybe follow something similar to the circuit in Post #16?
 
Administrator
Joined 2007
Paid Member
Maybe. One thing though, with a FET having a high gate impedance at the frequencies involved, the R/C network can be much reduced. A 1meg and 2.2uf cap seem to give similar results, even a 10meg and 0.22uf does.

(the oscillation is actually nothing of the kind... silly me ! Its just the 100/120 Hz ripple looking suspicious because of the x axis time scale. FET choice is important in minimising this)