diyAudio (
-   Power Supplies (
-   -   Keantoken's CFP cap multiplier (

MRupp 4th October 2010 11:16 AM


if you want shunt reg and don't want to dissipate heat, you can use clever idea from Wenzel Finesse Voltage regulator ...
Or maybe not, after reading this: Reducing Power Supply Ripple and Noise

HiFiNutNut 5th October 2010 12:34 AM

Thank you very much for your responses. I am changing my ISP at the moment and have not had access to the internet for a few days.

I hope I am not stealing too much out of Iko’s thread. This thread needs to be awaken from time to time. For those who are after the ultimate, no-compromise regulator, the Iko reg is one of the best options. I understood I could get better responses here than starting a new thread, but am conscious that I won’t be going too far down.

Most recommended regulators here look fine to me. However, many increase the output impedance. They may behave well in simulations or in “class A” circuit with a constant load. I am not sure how they would behave with “class AB” load with signals of transient / dynamic nature. The Iko reg, Salas reg and Jung reg are reportedly superior in sound quality probably due to their diminishing low output impedance right to high frequencies.

If I had the electronic circuit knowledge, I would probably take Keantoken’s advice and start a new thread on a discrete June reg. My knowledge is too poor for that and I won’t have the time anyway.

The TeddyPardonSuperReg has C-multipliers at the output which increase the output impedance. Iko’s circuit is two C-multipliers in serious. If I were to try the TeddySuperReg, it would be easier to try the Iko’s C-multiplier before the Marantz reg.

Stormsonic’s middle schematic is the closest to the Marantz regulator circuit I have been thinking about replacing or enhancing.

You may be interested in reading this:

I bought my Marantz SA11S1 based on the recommendation from a trusted friend. He had the Marantz, Mark Levinson and Audio Research CD players, out of which he liked the Marantz the most. With a reference level system running, my SA11 is obviously the “cheapest” component in the audio chain at just above $3000 retail. So I have been looking for a better CD player to match the level of performance of my system. I built everything (and designed the active loudspeakers) in my audio chain except the CD player which I could not build.

The stocked SA11 has the typical Marantz house sound. I replaced all the filter caps from Elna Silmic / Cerefine to Rubycon ZL and replaced the output caps from Silmic to Vishay MKP. The Marantz house sound is gone and it sounds a lot more neutral – it no longer sounds like any Marantz though.

A week ago I compared side by side my lightly modified SA11 to the EMM Labs CDSA ($15000) and dCS Puccini ($25000). I also invited the ex-president of the Sydney Audio club, JM, to join me in the audition of the CDSA. We were comparing oranges to apples. The CDSA and Puccini gave a much more detailed and rich sound but also more colourations at the same time. At the end, I did not know which one to choose. On many sound tracks the CDSA and Puccini are more attractive, on others they lost to the SA11. On drums, organ music and sopranos the SA11 killed the CDSA and Puccini. JM put it this way: The CDSA sounds like the best of HiFi. The SA11 sounds closer to the music.

So there you go. Perhaps I don’t need to stretch my wallet to get an expensive CD player. The thing that looks “promising” is that I have measured a few mV ripples on the rails on the Marantz, which means it has a scope to be improved upon. Perhaps if I can get the rails as flat as the Iko reg (which is a thin flat line) it would throw the expensive CDSA / Puccini into the water?

Andrew, no, I can not fit the Iko reg outside the Marantz box.

For now, it seems it is worthwhile to build a C-multiplier and put it in front of the Marantz reg and see how it sounds. I happen to have a pair of D44H and D45H in my part pin.

Any more advices are welcome.


iko 5th October 2010 12:47 AM

Don't worry too much about OT. I think we're fine and we talked about filters and gyrators in the past.

The cap multiplier won't give you low Zout if that's what you're looking for, but then... to get that you already know what kind of heat you need to dissipate with a shunt. It will, however, give you surprising line regulation (filtering). Perhaps you may not need to get such low output impedance for that load, and it's something that you can literally put together in a few minutes. This double cap multiplier beat the pants off the finnese and variations, and it's what I use in my low noise measurements preamp. Because the preamp schematic had such poor psrr I really needed this filter and got as low as about 0.4nV/rtHz preamp self-noise.

Sometimes a simple solution is enough. However, maybe a v1 Salas shunt is still doable with less idle current to avoid the heat issue? Then you get low output impedance too.

At least you have options :D

HiFiNutNut 5th October 2010 03:13 AM


Originally Posted by ikoflexer (
Because the preamp schematic had such poor psrr I really needed this filter and got as low as about 0.4nV/rtHz preamp self-noise.

0.4nV/rtHz at idle or when there is a dynamic load? I hope it is the later.

Anyway, given that the Marantz has such large capacitance in the load on board I guess higher impedance may be OK. I would add the Gyrator before the Marantz series regulator, in contrast to what TeddyPardon does to the output of the regulator.

iko 5th October 2010 03:54 AM

This is a low noise preamp that I built to make low noise measurements with, and the idea is that you can't measure low noise if the preamp itself has self-noise. There are others who have built such preamps with even lower noise, but for me it is an achievent even so.

Bill, it's a simple circuit which you can do without a pcb, to test. If not good enough, chuck it and go to the next level. Or if you have doubts, definitely try something else.

Filter before the regulator, and then measure at the load. But remember to measure with very short ground lead on the scope probe.

keantoken 6th October 2010 10:27 AM

1 Attachment(s)
Even the C-multiplier can be improved upon with a few components. In this case the output impedance is decreased, and bass response increases by a very large margin. Using a complimentary Darlington/CFP arrangement, the drive needed at the input is very small, in which case we can increase the bias resistors, increasing the RC time constant.

Using the CFP improves output impedance. It may improve HF impedance as well but it depends on the slave transistor, and the PNP version of the D44H11 has worse HF specs I believe which is why HF impedance is worse in this case. In this case you could probably avoid clunky low-voltage amp type devices and use old CRT flyback BJT's (cut out of monitors/TV's) which are fast with low Cob even if they have lower gain, and should be bulky enough for your application.

There is stigma surrounding the CFP because of its reputation for oscillating in amplifiers. I don't expect a big ordeal, but I haven't seen many do this so I can't be sure. The first thing I would try would be a base stopper on the power transistor.

- keantoken

keantoken 6th October 2010 10:36 AM

Another thing to consider with the CFP-multiplier is that if the load current drops below the quiescent of the master (Q3) the slave (Q2) will turn off. This is usually not an issue, especially if electrolytics are used in bypass, but if it is a problem you can use a bias resistor to draw several mA at all times.

Quiescent here is 4.6mA. For a 12V output, 1.2k will provide 10mA bias, good enough for me.

- keantoken

stormsonic 6th October 2010 10:53 AM

1 Attachment(s)
Keantoken, can you simulate with reverse polarities, Q2=NPN, Q3=PNP? Like pic bellow

Or replace 2SA970 with JFET (J202)

AndrewT 6th October 2010 11:29 AM

you have posted a regulator proposal, not a capacitance multiplier.

Using LEDs fixes the base voltage and changes the behaviour of the circuit rather a lot.

stormsonic 6th October 2010 11:54 AM

no, I'm just just asking to simulate his circuit with Q2=NPN and Q3=PNP.

Or to simulate his circuit with Q2=NPN and Q3=JFET

sorry for bad expression

All times are GMT. The time now is 01:07 AM.

Search Engine Optimisation provided by DragonByte SEO (Pro) - vBulletin Mods & Addons Copyright © 2021 DragonByte Technologies Ltd.
Resources saved on this page: MySQL 16.67%
vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2021 DragonByte Technologies Ltd.
Copyright ©1999-2021 diyAudio