Replacing the ALWSR's pre-regulator by a VBE

[teddy_pardo]

The following mod consists of replacing the ALWSR's pre-regulator by a capacitor multiplier.

A capacitor multiplier AKA VBE multiplier, or Gyrator, is a technique to create a virtual large capacitor from a small capacitor and a transistor. Basically the capacitance of a small capacitor is multiplied by the transistor’s hfe (for more info read here and here).

Why?
The supper regulator does an excellent job in providing low noise low impedance supply. However, as a feedback based design its performance at very high frequencies is somewhat limited (it’s still much better than monolithic regulators though). The ripple after the rectifier and even a 22,000uF capacitor has a saw tooth form of about 70mV at load, which is equivalent to a composition of high frequencies, and thus very challenging for a feedback loop regulator. I have done tests with larger capacitors and CLCLC filters and found that if I can reduce the ripple at the ALWSR input, and especially if I can give it a more sinus like form the performance of the ALWSR improves dramatically. The ALWSR’s on-board pre-regulator works basically the same way as the ALWSR (Jung) regulator itself but is even worse WRT high frequencies. Replacing the pre-regulator by a VBE multiplier gives the best of both words: the VBE creates a very low and smooth ripple (less than one mV) supply, while the ALWSR stabilizes it to the required voltage and provides very low impedance. There is an additional advantage, since the VBE is just a virtual capacitor, it doesn’t require a huge transformer to charge it, in fact I achieved very good (even excellent) results powering my NAC102 with very small transformers in the range of 10-50VA.

Another advantage is that there is no need to have many raw supply channels, many ALWSRs can be connected in parallel to a single channel supply.

And one more: The ALWSR can now be used at lower dropouts (down to around 3.5V). Note however that since most of the dropout is now on the D44H11 it may require a heatsink. (As an alternative the input voltage can be reduced by adding a resistor in parallel to the capacitor, dividing the input voltage).

How?
Fortunately, it is very easy to do it with the ALWSR PCB. In fact it can accommodate the VBE without any change to the PCB itself. Here is how to do it for a positive ALWSR:

1. Replace the 317/1086 by an NPN Darlington power transistor (I have used BDX33C)
2. Replace R3 with a 470-1000uF capacitor (it may be worthwhile to add a 1-2uF film capacitor in parallel).
3. Ignore C8
4. Replace R2 with 1K. I mounted it at the back side of the PCB to leave space for the capacitor.
5. Link TRS2 to TRGND
6. Link TRIN to SRIN
7. Use TROUT as the Input (instead of TRIN)

That’s all…

You can see how it looks in the following pictures:



A closer viwe:



And R2 at the back of the PCB



I than took it one step further and created a compact HiCap replacement for a friend in which I replaced the dual transformer by a simple 2x15V 50VA transformer, and the two 22,000uF capacitors by another VBE. I used this circuit (fig 3), in which I used only the +V part, and a single BDX33C darlington transistor instead of the two transistors.

I also used the same transformer to power the control circuits of the NAC102 (replacing the NAPSC), more on that later.

Here are some pictures of the full PSU:



and



Finished:



and from the back:



The bonus NAPSC connector:



Oh, and I forgot to say, the improvement in the sound is relly great, but who cares...

Now seriously, it's the best thing I've listened to!

Note for NAC102 owners:
In the pictures above you can see that I used a small 317 based circuit to supply the control circuits of the 102. I then used a much better approach. Since my transformer is a center tapped 2x15 providing 41V DC, I just added a 2200uF capacitor between the ground and the center tap to get 20V DC. Simple and effective…

Unlike the HiCap, there is no issue here (at least not to my ears) with sharing the same transformer between the audio and control supplies.

Enjoy, Teddy

[Mus]

Nice write up, Teddy. I don't understand why you replaced the pre-regulator though? Would you not have been better off inserting the gyrator circuit between the pre-regulator and the super-reg? Also, did you try leaving the super-regs intact and simply adding gyrators right next to the circuits being powered i.e. after the super-regs?

Regards,

Mus

[teddy_pardo]

Yes, I tried several combinations and found this one to be the best sonically, but that's me, I'll be interested to hear what others think. In addition having the VBE on the ALWSR allows connecting many of them to a single raw PSU. I have the intention to connect 8 ALWSRs inside my NAC102:

TA front end
TA output
Gain front end
Gain output

All these 8 ALWSRs will be powered by a single 30VA transformer!

Teddy

[S-Man]

Over on DIYAUDIO various people concluded that one of the benefits of a tracking pre-reg is the reduced modulation of the LED based current source by changes in output voltage.
(Can't this myself, since the pre-reg can't keep up with the SR at HF anyway).
I am not a great fan of op-amp based regs, but when I tried the ALW SRs I thought it sounded rather better without the pre-regs. If I had not sold my ALWs I would certainly try Teddy's suggestion - the logic seems spot on to me!


BTW, anyone tried Coffin's reg? (I have).
http://www.diyaudio.com/forums/showt...hlight=minireg

[trancera]

Teddy,

Are you saying that that PSU provides the sonics of a hicap and the control (napsc) power all in one ?!

Btw, top bodging award, looks good and a very discrete PSU, must be excellent vfm

[Sparts155]

Originally Posted by S-Man

BTW, anyone tried Coffin's reg? (I have).
http://www.diyaudio.com/forums/showt...hlight=minireg

I've got a Coffin board, but not got round to building it yet. What do you think of it?

[S-Man]

I found the Minireg slightly better than the ALW. For my taste all the opamp based regs I've tried are a bit too analytical, the Minireg less so than the others.
Please take into account that I am powering either opamps (Jung nested f/b) or a discrete opamp so may get different results to Naim type circuits.

I realise I am in a distinct minority here but IMO 78/79XX can sound better than op-amp regs if they are carefully applied! (I guess I have zero cred at this point).

FWIW I measured <5mOhms o/p impedance with a +/-24V coffin reg and it took about 4uS to recover from a load transient (measured right at the o/p terminals, 109mA transient into a resistive load). It is also very quiet <0.2mV pk-pk.
My favourite shunt reg is much noisier at around 2mV pk-pk noise, with a similar output impedance and 2uS recovery time - the recovery is also somewhat cleaner and better damped.

[Napsac]

Interesting thread, looking to try this one.
Any tips / advice on calculating the the voltage dropped by the gyrator, read the links from ESP which mentions adjusting the 12k resistor.......
Is this a case of suck it and see ?

[martin clark]

No, you can work it out. Bit of discussion of gyrators in this thread, with the numbers coming in post 11:
http://www.pinkfishmedia.net/forum/s...ad.php?t=17825

[peranders]

Just a note: A gyrator or C-multipier can't store any energy so the result isn't always the real thing simulated.

[peranders]

Originally Posted by S-Man

I found the Minireg slightly better than the ALW.

I'm sorry to say but how did you compare the performance? The Minireg has major design faults and is the lowest performing super regs I ever have come across. The major fault is the grounding and this leads to unstability when you are using faster opamps.

If I compare the ALW super regulator and the MiniReg, the winner is obvious if we talk electrical performance.

[ariegur]

This PS runs my 102 with regulators that teddy installed in and the sound is great. This is a little box that replaces the napssc and the Hicap and it does it with great results.

The price is about a quarter of an s/h napssc and Hicap so is the place that this box takes compared to the other two boxes.

Arye

[Linnik]

Very nice thinking and application Teddy!
Thanks for your comments too, Arye!

This kind of new thinking is always welcome.

Andy will obviously comment after he has got time enough to get deeper into this.

BTW Teddy, in the photo your t-former is 2x12V and not 2x15V as you tell.

Oz

[trancera]

Arye,

Are you able to compare this to a Hicap?
It looks great and I have some Sreg boards unused as yet so very interested!

Brian

[markt]

Brian, your inbox is full.