advertisement


Salas Shunt Regulator Build

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.
Geoff

No, I think you are spot on with that Geoff. The big electros are better IMHO got both MC and MM and digital circuits. that has a lot to do with the LF noise and the sensitivity of these circuits to it.

As an aside it is a good idea to also consider the load the regulator is driving. If it is resistive then fine but if there is on board capacitance to earth for local decoupling right at the load then I find it sounds better if it is either given additional artifical "ESR" or a series resistance is put into the feed between the Salas reg Zobel/output and the capacitance in question. With the Legato I have tried both these and eventually removed the 100uF caps on board altogether and replaced them with a 100nF film and a 1 ohm resistor in series to good effect. I did the same on my B4 preamp.

Hope that is useful.
 
As an aside it is a good idea to also consider the load the regulator is driving. If it is resistive then fine but if there is on board capacitance to earth for local decoupling right at the load then I find it sounds better if it is either given additional artifical "ESR" or a series resistance is put into the feed between the Salas reg Zobel/output and the capacitance in question.

(Also an aside, but) +1 to that. The output of all feedback-based regulators 'looks' inductive eventually. In other words at some point you are looking at an R-L-C circuit. Using that to one's advantage to provide some damping only ever improves things - transient control etc.

The reasoning for almost any reg remains essentially similar to this, it's just a matter of how many poles you are dealing with and where they end up/ how you use them.
 
(Also an aside, but) +1 to that. The output of all feedback-based regulators 'looks' inductive eventually. In other words at some point you are looking at an R-L-C circuit. Using that to one's advantage to provide some damping only ever improves things - transient control etc.

The reasoning for almost any reg remains essentially similar to this, it's just a matter of how many poles you are dealing with and where they end up/ how you use them.

So I am using my first regs to power a chip based active filter ,this did have 220uF decoupling on the v-in which I have removed and 0.1 cogs at each chip is this worth a try I do have some 1Rs around spare.
 
Many thanks all.
The labeling of two capaitors for one physical position threw me somewhat.
My first aim is to just get the Salas running, followed by getting the currents right and finally tweeking for performance which probably just means copying Johns build.
Tony
 
Couldn't wait could I :D

Built the first regulator and it indeed regulates and no fires, however I was a little wide of the goals.

Lowest settable voltage was 5v38 using two red leds, 220R (R103) and 500R trimmer so I changed the 220R for 120R and can now set from 4v87 to 6v06, I set it at 5v25 for Buffalo VD.
Very stable output with not drift other dropping by a few hundredths in the first minute or so.

Next missed goal was the CCS current which with 180mA in the dummy load the regulator was shunting only 85mA. I added a parallel 82R 1/4W to the 12R (R101) to give 10.4R and are now shunting a little over 100mA, total CCS is 295mA. I stuck the extra 82R under the board.

Make shift heat sinking is more than adequate, the regulator is only slightly warm.

Looks like I am set to make its twin.

Few pictures.
I had to prattle around trying for the correct dummy load as you can see :D

IMG_1796_zpsa6cc81f2.jpg


IMG_1798_zpse988d779.jpg
 
That's a great result from the off Tony. A couple of things that might help with the tuning of the regulator before fitting. If you are drawing circa 300ma then you probably need a larger reservoir cap. I use at least 4700uF and in my Buffalo II PSU I have used 6800uF. If you are driving digital circuits then that 1000uF Kendeil electrolytic should be where its adjacent Wima cap is. I get them to fit by replacing the pot and series resistor with just one fixed resistor of their combined value which allows the pins of the Kendeil to engage sufficiantly with the board to solder from beneath. I have also done the rectification and filtering on a separate rectifier board and connected to the Salas via a low value (up to 2.2 ohm) series resistor or inductor. I've no proof that this helps having never compared and measured, I just like the idea of keeping high current switching spikes away from voltage references.

Great heatsink and four wire sensing already! Woa, steady on there boy. I've not got round to that yet on my Buffalo even after 6 months of Salas regs.

John
 
Hi John, thanks for keeping an eye on me.

A couple of things that might help with the tuning of the regulator before fitting. If you are drawing circa 300ma then you probably need a larger reservoir cap. I use at least 4700uF and in my Buffalo II PSU I have used 6800uF.

Yes habit of fitting Kendeils in reservoir positions, never gave a thought to where it should go. I usually do PSUIII, in fact I have a mind to run some of Carls rectifier boards as a Salas feed section as a latter addition so could get quite silly. As a temporary measure I have some 4700uF that could fit in.

If you are driving digital circuits then that 1000uF Kendeil electrolytic should be where its adjacent Wima cap is. I get them to fit by replacing the pot and series resistor with just one fixed resistor of their combined value which allows the pins of the Kendeil to engage sufficiantly with the board to solder from beneath.

Woops and you did mention this in your email, I have to admit to being a little brain dead of late. The 1000uF Kendeil is one hell of a Vref filter. Shame on me for not realising.

I intend to feed VD in the first instance but do remember that 90 odd percent of my VD load is just the Avcc circuit as all the other Tridents are fed from Fleas and separate transformers. My VD load is now just Avcc plus just 6.7mA digital.

Once the Salas regs are something like I will move them to feed the Avcc Tridents directly and make up some other regs for the 6.7mA VD terminals. Maybe Flea but probably a TPR in the interim as I have a pair about right for a temporary job.

Hope I am explaining my self, its a sort of staged implementation and saves me time building lots of bits. If SQ screws up along the way I can easily revert and then have to build all the bits in one hit.

I have also done the rectification and filtering on a separate rectifier board and connected to the Salas via a low value (up to 2.2 ohm) series resistor or inductor. I've no proof that this helps having never compared and measured, I just like the idea of keeping high current switching spikes away from voltage references.

I will get the scope out once they are in position and also do a PSUIII model just to see whats happening, I have my transformer data somewhere so it should be quite accurate.
Good point about getting the rectifier pulses away from Vref.

Great heatsink

Heat sink looks like a large alloy bar in the photo but its actually alloy angle about 10x40mm from B&Q


Tony
 
Heat sink looks like a large alloy bar in the photo but its actually alloy angle about 10x40mm from B&Q


Tony

Your accidental trick photography had me completely fooled there. I thought you had a nice chunk af Alu bar. It is still man enough for the job I should think, though that will depend on how many volts are across your CCS FET. You will need at least 5 volts across the FET which is a minimum of 1.5 Watts and then there's a further 3 volts headroom required by the current setting resistor(s) (which dissipate nearly 3watts BTW so the lower value one will get toasty).

Still can't believe how quickly you crack on with this stuff!
 
Yes it is very cool to the touch, no were near the temperature the Placids run at althogh Placid load is maybe 500mA and this is 300mA.
Planing on upping the voltage headroom so it nay get hotter yet, we will see.

I find the build very fast, maybe 30 minutes for stuffing the board and a two or three hours all told with pratting around with heat sinks and testing, of course its not fitted to the Buffalo yet.

Its the brain side of things that slow me down, this Salas is all Greek to me :D
 
Looked at the ripple with different reservoir capacitors.
Rough figures and 300mA load.
1000uF reservoir gives 2v79 ripple.
33000uF reservoir gives 126mV ripple.
47000uF reservoir gives 100mV ripple.

So I have decided to have a whole new front end using Carls generic rectifier boards.
Transformer
Nuvotem Toroidal 50VA 2x9V o/p
49Ω 230 volt primary
2.778 amp per secondary
10.2 volt no load per secondary
RS 223-7888

Capacitor
Cornell-Dubilier SLPX 85degC Snap-in cap 33,000uF 25V ESR 0.018Ω
RS 744-1142

Inductor
Ferroperm - 1585 500uH 3A INS - Inductor, Axial, 500uH, 3A
500µH / 0.2Ω.
Farnell 1186803

All in CLCLC configuration giving a ripple of <2mV

Again RS have no stock, transformers not in until 08/02/2013 and capacitors 28/02/2013
So alternative capacitor looks to be
Panasonic TSUP Al electrolytic cap,33000uF 35V
ECOS1VP333EA
RS 127-492

Lastly I got a little worried about voltage headroom so thought to increase the transformer from 2x9V to 2x12V
Transformer
Nuvotem Toroidal 50VA 2x12V o/p
Mfr Part No. 0050P1-2-012
49Ω 230 volt primary
2.083 amp per secondary
13.71 volt no load per secondary
RS 223-7894

Changes in transformer characteristics now give a ripple of 1.5mV, terminal voltage of 17v45, peak inrush current of 6amp and an extended start-up time of around 2 seconds although time is not a problem for its ultimate intended position of Avcc.
Perhaps a little overkill so I like it :D:D:D

Tony
 
I've subscribed to this thread, this is what i'm eventually going to get round to I think, not the time now though so will 'save the thread for later'

Thought that powering each side of the legato similarly would be a good idea as well.

Stefan
 
Hi Stefan, yes I would expect they would improve on Placid for the Legato.

They really don't take long to build once you have all the parts. Not much additional kit required if you had the kits from Teabag unless you get carried away with rectifier stages. :D

I now have some twins ready to take to the plain for battle with a herd of Buffalo :D

IMG_1800_zps0b159453.jpg
 
Has anyone on here built the V1.2R regs? I built mine (one per channel) on a single piece of verobaord for my Salas phono stage. The component count is quite a bit higher than the V1.0 and V1.1 regs but if is for a low-PSRR use then the V1.2/V1.2R are better but you have to be careful about the layout (so I made mine as compact as I could). Here is my phono stage/regs:

http://imageshack.us/f/515/dscf3305k.jpg/

Looks good on that build, I have an Arcam Alpha CD (original) sat waiting to be filled with Salas regs when I get round to it.

The V1.0 are dead easy to build on verobard, I have a few other components that I use them to power the audio circuits with.
 
Managed to squeeze it in.

IMG_1803_zps752ac209.jpg


Its a dirty job and needs more work, feeding into Buffalo's VD to feed 180mA Avcc regs and a 6mA of miscellaneous unknown digital.

Plenty more work needed on the implementation, this being the most basic.

SQ is good, very good, lots of bounce. Bass and low bass is noticeably improved along with some clearing in the upper frequencies.
The increased bass is quite dramatic as I have never felt it rumbling in my chair before, double bass string work is much improved.

Only one slight down side in so much as stereo image seems a little smaller but would live with this. I need more time with the new sound to be sure and still need an install tidy up.

Thanks all
Tony
 
Has anyone on here built the V1.2R regs? I built mine (one per channel) on a single piece of verobaord for my Salas phono stage. The component count is quite a bit higher than the V1.0 and V1.1 regs but if is for a low-PSRR use then the V1.2/V1.2R are better but you have to be careful about the layout (so I made mine as compact as I could). Here is my phono stage/regs:

http://imageshack.us/f/515/dscf3305k.jpg/

Looks good on that build, I have an Arcam Alpha CD (original) sat waiting to be filled with Salas regs when I get round to it.

The V1.0 are dead easy to build on verobard, I have a few other components that I use them to power the audio circuits with.

Not seen them before.
 
That's getting crazy now. I've hardly done anything to mine. I loved bodging the gigadac but this one seems rather to expensive to risk killing :-(

I'd love to get a good look at some of the builds up close, be good to hear the difference as well. I'm hoping to get to Johns soon, only about 30 miles away. I'd like to listen to the to Tony's as well. Alas my house just isn't big enough to fit all of you in for a good social. Anyone fancy hosting?

Stefan
 


advertisement


Back
Top