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

Salas Shunt Regulator Build

Discussion in 'd.i.y.' started by Bemused, Jan 30, 2013.

  1. Bemused

    Bemused What's this all about then?

    Ive had a Salas Back in Black (SSLV1.1) board for a while.
    Now its time to do something with it, the board came with a partial parts kit but I'm not sure what was supplied as yet.

    I was going to set the Salas to run a Buffalo III main supply VD, currently a Placid is doing this duty.
    The Buffalo manual says VD is 5v25 @ 440mA with the Trident regulator set.
    My Buffalo is currently drawing a 185mA on VD with just one Trident V2.1 running the Avcc rails from the Placid.
    The Tridents for VDD_xo, DVcc and VDd have been replaced with modified versions of Martins Fleas which equates for the difference in current.

    A little further measurement with the AVcc Trident removed showed just 6.7mA on VD, so the AVcc is up at around 180mA and the VD alone is just 8mA.

    This made me reconsider my approach, now thinking to use the Salas for solely AVcc duty either with the Trident or using just the Salas alone. Leaving me with just 8mA to feed into the Buffalo VD terminals from something else. 8mA is bang in standard Flea territory so the new scheme looks to be
    Salas > Trident > AVcc
    Flea > VD
    Your thoughts are welcome, I'm just at the thinking aloud stage here.

    Back to the Salas, first considerations assuming I go with the suggested scheme are either to use the BJT or FET version. Target voltage for the AVcc is 3v3

    From the Salas build manual available here

    a. 2.5-5.5V Vout BJT output reg. 1 red 1.9V LED, 1K trimmer. Other parts jumper.
    b. 5V Vout Mosfet reg. 2 1.9V LED, 220 Ohm R303. Other parts jumper.
    c. 10V-25V Mosfet reg. 2 1.9V LED, 1.8 kOhm R303. 5K trimmer. Other parts jumper.
    d. 38V Mosfet reg. 2 1.9V LED, 33V 1/2W Zener, 220 Ohm R303. Other parts jumper.
    e. 25-40V Mosfet reg. 2 1.9V LED, 6.8k R303, 5K trimmer. Other parts jumper.

    Option a. 2.5-5.5V Vout BJT output reg seems to be the choice.

  2. GWM

    GWM Lost in the 70's

    My understanding is the BJT is no good for higher voltages but both can be used for 3.3V ,theres lots of info on the Diyaudio build thread though its long its still worth wadding through lots of info re-buffalo impimentations.Do you have the 4 green plus 2 red LED version. I am just about to power up my first builds for an active filter supply will probably blow up,do you have a copy of the spread sheet supplied by Salas.
  3. Bemused

    Bemused What's this all about then?

    Thanks, john.luckins has kindly filled me in on lots of documents, I have them all, he suggested the FET
    b. 5V Vout Mosfet reg. 2 1.9V LED, 220 Ohm R303. Other parts jumper.
    But I may have moved the goal posts since his suggestion.

    I think I have the same version as you.

    Nice thing about shunt regulators is they are short circuit proof, but good luck with the power on all the same.
  4. john.luckins

    john.luckins pfm Member

    Hi Tony,

    I run a 5.25 volt Mosfet Salas Reg into my Avcc regs (left and right) of my Buffalo II rather than using it to replace the TP Avcc Regulators altogether. Many report that the Avcc feed is the most sensitive on the Buffalo and this second layer of regulation helps. I've never used the Salas BJT output reg as I've never fed less than 5 volts so I can't comment on how good it is.

    Maybe a red herring but there is another Salas design of shunt regulator, the details of which were buried deep within the DIYaudio thread. It was called the Reflektor as it uses a current mirror to feedback the output voltage to the shunt and reference. The substitution of a NPN in place of M2 makes it suitable for 3.3 volts or even 1.2 volt. no boards that I know of as yet. I was thinking of using this after a higher voltage Salas FET Reg to feed the two sides of the Avcc separately, ditching the TP Avcc Reg as a result Here it is:


    FWIW I went from a Placid BP to a Salas feeding the entire Buffalo II and found it to be an upgrade, after that it just got out of hand! The Zobel and voltage reference caps are critical to the eventual tonal qualities of the regulator as is the load capacitance as is the case with all feedback regulators.

  5. Bemused

    Bemused What's this all about then?

    Many thanks for a detailed post again John.
    Mosfet it is then, I'll punch some data into the spreadsheet next.
  6. GWM

    GWM Lost in the 70's

    Hi John what did you use for the caps I had some wima MKP4 so put them
    in the zobel and polyester for the other spot.
  7. john.luckins

    john.luckins pfm Member

    I managed to get hold of a large number of NOS Shizuki Japanese polypropelenes from Cricklewood electronics before they ran out. They fit nicely and are just a little more bouncy than the Wima MKP4's I had tried. Without them I'd be back to the Wimas. I also tried Amp Ohm FP-CA-AU polypropylene from a cap supplier in Stevenage. They were better than the Wimas and about the same as my NOS Shizuki ones but way too large for the board and too pricey. As I have them I've used the Shizukis for the voltage ref filter when feeding the analogue stages, never tried the Polyesters, though I suspect the difference is smaller here. I use Kendeil 1000uF 50 v for the reference filters on digital circuits which makes them very quiet at low freqs where it matters. I've also noticed that the wimas improve over time and tend to lighten up as they go.

  8. GWM

    GWM Lost in the 70's

    Didn't know kendil did 1000uF always used Panasonics at that value where did you get them?
  9. john.luckins

    john.luckins pfm Member

    Mine were from umut1001 on Ebay. A reliable seller of Kendeils. I have some new ones spare if you want to save a bit of time and money. Hisa shipping times can be lengthy.

  10. Bemused

    Bemused What's this all about then?

    John, I would be most pleased to take a pair of Kendeils should GWM not take up your offer.

    I dont have the board to hand but assume the Filter shown on sheet 1 of the spread sheet is contained within the Salus.
    What filter combination do you recomend for my load of 185mA and is 100mA overhead giving a toal CCS of 285 sensible

  11. GWM

    GWM Lost in the 70's

    Thanks John I'll keep that in mind,bought some Pannies only a few weeks ago.
  12. GWM

    GWM Lost in the 70's

    Tony no the pre filter is not on the board just a pair of bridges and single cap.
  13. john.luckins

    john.luckins pfm Member

    Hi Tony,

    PM me your address and I'll pop a couple in the post for you to try. They are a tighter fit than say 470uf Panasonics but did improve things when driving the Buffalo digital stages IMHO.

  14. Bemused

    Bemused What's this all about then?

    Thanks John Ive sent an email.
    No rush.
  15. PigletsDad

    PigletsDad pfm Member

    I've looked at that circuit, and can't see how it works. Is S+ connected to the output at the load, and S- to earth at the load?

    The trimmer and diode combination sends a current into the mirror, and an equal current comes out of the other arm, regardless of S+ voltage. I can't see where the shunt feedback is.

    Can somebody explain it to me?
  16. john.luckins

    john.luckins pfm Member

    Having never built it I don't know for sure that it does PD, however this is how I think it should:

    As you suggest S+ is connected to F+ at the load, likewise with S- and F-. If the load current draw reduces and the voltage at S+ increases so the voltage across the sequence of two series e-b junctions, the 1S1588 diode (which I'm assuming isn't a constant current diode) and the 1k trimmer must also increase. The dynamic impedance of all components in this string apart from the 1k trimmer is low and so most of the voltage increase appears across the 1k trimmer. Current through the trimmer increases therefore. Current across the other leg of the current mirror increases by roughly the same amount as a result and develops an increase in the voltage across the two 560 ohm resistors and thus the gate of the FET. The increase in FET gate voltage causes it to conduct more and so reduce the current available for the load, so reducing its voltage.

    I can see that the electrolytics will considerably slow down the response of the reg and that there are issues of having sufficiant voltage across the FET but that's about as detailed an examination as I have done.

    Salas may have built it and know how it performs but I thought it a reasonable candidate for a low voltage shunt as with an npn instead of M2 it can operate down to output voltages of around 3-4 diode drops depending upon components , i.e. circa 1.8-2.4 volts and with a bit more flexibility at 3.3 volts.

    Maybe I'm about to find out that either my explanation/analysis is flawed or perhaps that the circuit has performance issues. Probably a good thing too in case someone has just lept off to build one on the basis of my post, so critiscism gratefully received here!

  17. PigletsDad

    PigletsDad pfm Member

    The FET will keep working well below it's pinch-off voltage, just acting in the voltage controlled resistor mode, so lowish voltages should be OK.

    I think your explanation makes sense; it seems a very complex way to get roughly unit gain. Why not just connect the diode from S+ to FET to top of trimmer, bottom of trimmer to ground, and wiper to the gate?
  18. Sparts155

    Sparts155 pfm Member

    That's closer to the original Salas shunt, which also has an extra BC transistor in-between the reference part and the output FET to increase open loop gain. Not sure how the new circuit increases performance .... ?

    I've built a version of the Salas shunt too, by the way, and would highly recommend it.
  19. Bemused

    Bemused What's this all about then?

    Most parts for the first Salas are either found, begged or ordered.

    I have now selected the jfets
    Out of five I had
    Idss 4.17mA
    Idss 4.50mA
    Idss 5.43mA
    Idss 5.82mA
    Idss 5.87mA

    So I designated them as
    Q102 - Idss 5.43mA
    Q103 - Idss 5.82mA
    Q105 - Idss 5.87mA

    Other parts additional to the kit are

    C105 - 1000uf Kendeil K05 105 degree (Thanks John)

    R103 - 220R metal film 1% 1/4W

    D111, 112, 113, 114 - Qspeed LQA30T300 M006J.P.1 0911F

    R301 - Vishay PR02 Resistor,A/P,AXL,5%,2W,12R
    RS Stock No.683-5650
    The spreadsheet calculates 12R9

    R105 - Bourns 3296Y top adj cermet trimmer,500R 10mm
    RS Stock No.522-0209

    X101, 102 - TE Connectivity 282841-2
    RS Stock No.710-0362

    The capacitor C101, 102, 103, 104 I am a little unsure of, I van see the positions on the board and schematic.
    IE C101, 102 are drawn as parallel film and electrolytic but the board only has positions for one capacitor although both are labeled at that position.
    Is one to be fitted under the board?
    Due to uncertainty I have only ordered MKS4
    MKS4 Polyester Cap,100Vdc,4.7uF
    RS Stock No.108-3034

  20. GWM

    GWM Lost in the 70's

    Tony C101 or 102 are either or, use a big electro for sensitive loads such as phono pre or a film 10uF for tonal quality for less demanding loads,C104 should be best quality film between 4.6-10uF I used a 4.6MKP4 here but had to bodge the board.I am sure John will be along soon to correct me though.

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