1. Things you need to know about the new ‘Conversations’ PM system:

    a) DO NOT REPLY TO THE NOTIFICATION EMAIL! I get them, not the intended recipient. I get a lot of them and I do not want them! It is just a notification, log into the site and reply from there.

    b) To delete old conversations use the ‘Leave conversation’ option. This is just delete by another name.
    Dismiss Notice

howto: replacing the ALWSR pre-regulator by a VBE

Discussion in 'd.i.y.' started by teddy_pardo, Dec 16, 2006.

  1. teddy_pardo

    teddy_pardo Trade: Teddy Pardo

    I was asked by many to summarize the long “replacing the ALWSR pre-regulator” thread, and provide a short howto for people who just want to make it without reading the whole thread.

    A quick background
    The ALWSR has a pre-regulator, which is based on a monolithic linear regulator. Through experiments, I discovered that the performance of the ALWSR can be significantly improved by providing it a smoother input containing less high frequency noise. It was first done using an LC filter, and then using a VBE (Gyrator) circuit. Since both the super regulator and the pre regulator are linear regulators, the whole circuit is limited in it’s abilities to deal with high frequencies as exists on the mains, and generated by the rectification process. A VBE circuit consists of a low pass filter, feeding the base of a transistor configured as emitter follower. The role of the transistor is to provide current with lower impedance. This howto explains how to replace the pre regulator by a VBE.

    The VBE
    It turns out that the filter and the transistor have significant effect on the performance. The best circuit so far consists of a second order filter (RCRC), and a FET transistor or a Darlington made of discrete components (more on that later).

    The circuit is described here (many thanks to Hacker for the drawing)


    The Filter
    Best results were achieved using film capacitors. Electrolytic capacitors should be avoided here. For C2 I’m using 6-10uF polyester (MMK, MKS etc), and for C1, the second filter, I’m usually using a 220nF polypropylene. Class X2 capacitors (AKA suppression or AC capacitors) are doing a great job here.

    R1 and R4 effect the dropout, I’m usually using 64K/1.5M for low dropout, or 100K/1M when I can afford a higher dropout.

    The transistor
    Two main options exist, FET or Darlington. FET has a larger dropout and higher output impedance, most people on the forum preferred FET as it is giving a better bass. Personally I found that I like both, and it depends on the application. I also discovered that high impedance at the output of the VBE adds bass, and that may be the cause of the difference in sound. This bass is IMO slightly artificial but nonetheless pleasant to the ear. You can make a simple experiment by using just the VBE circuit without the super-regulator (the SR reduces the output impedance of the VBE). You’ll get a sound which is rich in bass, and very pleasant, something like a tube amplifier.

    With a FET, R3 is mandatory to avoid oscillations. I found that even with a Darlington it provides a slight improvement. Note that in order to use this resistor you’ll have to make a cut on the PCB (see picture below). Smaller values can be used too, I've used 118R successfully.

    In order to make a Darlington I use bc547c and D44H11, together they provide a HFE of about 100,000, allowing relatively low output impedance (few ohms). If your system has sufficient bass I'd personally recommend Darlington (but most other people prefere FET, so make your choice...). Here is how I build a Darlington transistor from discrete components:


    Here are some pictures of the modified ALWSR board:




    This one is made of seven 1uF MKS capacitors:


    Note the cut on the PCB below the blue 64K resistor. This cut is made to allow using R3 to avoid oscillations.

  2. hacker

    hacker Delicious and nutritious

    Great stuff here, Teddy.

    Something to bear in mind for the budding DIY-er who hasn't yet done this mod is that the pins on the ALWSR are used differently with this mod. I've pinched the following from the original thread:

    1. Link TRS2 to TRGND
    2. Link TRIN to SRIN
    3. Use TROUT as the Input (instead of TRIN)

    Failure to do this will result in a less than satisfying sound ;)

  3. natzev_i

    natzev_i pfm Member

    So, isn’t it just time somebody (Andy?, Teddy?) to produce a new PCB or assembled unit called ALW-TP Super-Mega Regulator combining the VBE and the W. Jung final regulator?
    Just a thought.
  4. maxlorenz

    maxlorenz pfm Member

    Thanks Teddy for your valuable effort :)

    I ordered parts for the simpler Darlington BDX33C based VBE, just to have a taste of it (out of stock at Farnell, it seems). I will use it on +5V, digital supplies and +/-15V analog supplies. Maybe use at lower V reduces differences between BDX33C and Fet based regs?

    Where can I find the instructions to do the negative VBE-ALWSR?? BDX34C
    I suppose...

    Many thanks,
  5. teddy_pardo

    teddy_pardo Trade: Teddy Pardo

    I'll try to post these too when I find a moment. It's a bit tricky though, you have to cross two of the transistor legs
  6. teddy_pardo

    teddy_pardo Trade: Teddy Pardo

    I don't have PCB design skills or tools, but I would be very happy if Andy will take the initiative...
  7. maxlorenz

    maxlorenz pfm Member

    Hi Teddy,

    I finally added the VBE yesterday. The ALWSR is feeding the 5V digital PS of a CDPRO-2M transport. I decided on BDX33C because it is easier for my limited skills. Cap is a Wima MKS4 6.8uF and R is a 18K PRP resistor.
    The cap is so long that I soldered it directly on TRGND, saving a few solder points ;-)
    I must have done it right because improvements were heard just minutes after power on: Sound is cleaner; HF and specially lower high freqs. are sweeter and warmer; intruments' textures and timbres are better reproduced and individual instruments are hence more defined; bass was already good but also improved in weight and "expansiveness". Speed did not suffer.
    All this only with a short listening session. I hope it improves further.

    The effect led me exited... I have a few questions:

    *are my R and C values OK?
    *could you post or help with a schematic to replace 337 prereg with a BDX34C for the negative ALW?
    (I fear a mistake from my part; I've been lucky so far)
    *Can this VBE be used on other PS like DVD's SMPS.
    I have a cheap modded multiformat player on my system nº2 wich sounds good but still has HF noise :-(

    Judging for my short experience I would say that the mod sounds like if I was using a BIG Wima MKS cap, he, he...I will shearch for Epcos MKP or MKS caps. Maybe they are cheaper.
    The only "adverse effect" that I noticed might be a little less "air", but this may be just my ears used to HF noise or lack of burn-in...

    Many thanks to Teddy, Andy and all who contributed to this wonderfull reg.

  8. mjrod99@comcast

    mjrod99@comcast pfm Member

    I'm just getting into the mix.I want to use the modified ALW for 5 vdc and +/- 7 vdc for a cd player.What is the lowest voltage needed to get this mod to work properly?I'm presently using :) 2 batteries at +/-12 ps to the 2) ALW sr.I'm ready to get started.

  9. martin clark

    martin clark pinko bodger

    5-6v overhead should be enough - if those are lead-acid batteries, they will range from c 13.8v fully charged down to 12ish discharged. The operative note of Teddy's is
    - so if you get more than 2.5v dropout across the capacitance multiplier, reduce R1 & R4. The ALWSR itself needs 2.5v of 'headroom'.
  10. mjrod99@comcast

    mjrod99@comcast pfm Member

    I'm also ready to do the negative version with the BDX34C negative version and don't want to screw it up.Also-what is the complementary negative version for IRF610?

  11. PigletsDad

    PigletsDad pfm Member

    IRF9610 is a reasonable choice - pretty much any medium power P channel FET will do; it needs to handle enough current (at least 1A for safety), enough power (a few Watts) and enough voltage (at least 25V), but not be any bigger than needed, as this increases the interelectrode capacitance.

    IRF9610 is 200V, 1.8A, 20W, so fits the bill nicely.

    I think it is now being withdrawn, so may not be that easy to buy. If you have trouble, I will try to think of a more modern alternative.
  12. teddy_pardo

    teddy_pardo Trade: Teddy Pardo

    A BDX33C + MKS6.8 + 18K combination is a less than optimal combination. The BDX has a low hfe when used at low currents. If you want to use it you should rather use a smaller resistance of up to 10K and a larger capacitor
  13. teddy_pardo

    teddy_pardo Trade: Teddy Pardo



  14. maxlorenz

    maxlorenz pfm Member

    Many thanks for your reply, Teddy.

    A pitty that my mod is less than ideal. Anyway it sounds better than before. Bass is deeep.
    Will 6.8uF*2 //, plus 10K be enough?
    (RS-online carry big MMK caps, up to 33uF)

    Excuses for the basic questions :(
    *Wich is better for low V circuits, Mosfet or home-made Darlington?

    *My raw supply is rather high, will a pre-reg placed before VBE be detrimental to sound?

    *Can I use VBE before monolitic regs, like LM317, or in SMPS (if I find the schematics) and still enjoy the sound improvements? I guess yes...

    I will try to digest your mod to (-)SR :)

    Best regards,
  15. teddy_pardo

    teddy_pardo Trade: Teddy Pardo

    *Wich is better for low V circuits, Mosfet or home-made Darlington?

    For high currents, I'd use home-made Darlingtons as they have lower impedance

    *My raw supply is rather high, will a pre-reg placed before VBE be detrimental to sound?

    No, I did it in a similar situation and it works perfectly good

    *Can I use VBE before monolitic regs, like LM317, or in SMPS (if I find the schematics) and still enjoy the sound improvements? I guess yes...

    Yes :)
  16. maxlorenz

    maxlorenz pfm Member

    Thanks! :)

    High currents...mmmhh
    The stock ALWSR had a max current rate wich was limited by the prereg, right? I will study the datasheets to try to deduce how high can the modded units go.
    I have a nice NOS DAC who could benefit from it...:)
  17. martin clark

    martin clark pinko bodger

    The current limit in the original ALWSR is set by the base drive to the pass transistor (D11Hxx), not the pre-reg, because the CCS sets a base drive which the opamp 'steals' to effect regulation. About 7-800mA is possible as standard, maybe 1A with slightly reduced performance. To go much beyond that you will need to increase the CCS current somewhat - which really means you're on your own in terms of performance vs. stability. Should be OK, but pay attention to heatsinking!

    An LM317T can pass 1.5A guaranteed, up to 2.2A usually...
  18. maxlorenz

    maxlorenz pfm Member

    Thanks for that, Martin :worship:

  19. norwood

    norwood Member

    Many thanks for posting the details of this mod, I am a bit confused about the way you have taken the resisitors off C2 in the MMK image i.e between the capacitor terminals where as in the image with the red capacitors you have taken the resistors from the positive output side as per the schematic drawing.
    Am I being thick or doesn't it matter.
  20. teddy_pardo

    teddy_pardo Trade: Teddy Pardo


    Sorry, I don't understand were is the confusion. In both cases one side of the capacitor is connected to the ground, and the other side (the positive) is connected to the resistors.

    Are you asking about the positive or negative version?

Share This Page

  1. This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
    By continuing to use this site, you are consenting to our use of cookies.
    Dismiss Notice