The simplistic Salas low voltage shunt regulator

[merlin el mago]

+1

Also, I'd like to remind everyone that moderators don't delete or edit anyone's posts due to technical mistakes. Posts only get deleted because they don't belong in the discussion at hand (e.g. off-topic, personal remarks, offending comments). The forum rules are pretty clear on that. Let's be nice and learn from each other's experiences here.

and still Orelli's post3789 appears !

Post 3789 is completely wrong. It offers the wrong advice from someone who has no clue on how that circuit (CCS + Shunt regulator) works.

What's you don't understand AndrewT?

 

[back]

Carbon film: Koa Speer at Farnell, Rs, Mouser, Digikey, etc.

carbon film is easy to find locally.

i wonder if something better is needed.

 

[orelli]

check output voltage vs input voltage.

V in=7.5 V out=5.25


You will find that a very low voltage overhead is required to maintain output voltage.

Now check the CCS current and repeat the experiment. Check ICCS vs input voltage.

CCS current=610mA V in = 7.5

You will find you have NOT got a 600mA CCS.


Not correct: I get over 600mA

Measure and check before you spout off ridiculous claims.

Thanks for your kind words.


ciao, Orazio

 

[orelli]

Hi. The border margins you saw are Vgs related limits. Its good you allowed for some 2V more for mains margin and ripple. Some with more varying mains would be safer at 12V input. JFET or BJT cascode CCT? R10 value? What you had in ithis reg's place before and did you feel any difference? Pictures are welcome.

Sorry , what's CCT?
I,m using your schematic showed in post 3200 (+15Vout) with some changes;

R1=1R
Q5=IRF9530
Q8=K170BL
R6=1k8
R10=26k5(2x13k2 serie)
Rload=12R

all the rest is the same.


ciao, Orazio

 

[Salas]

Hi. CCT means CirCuiT. So you used the one that has Jfet cascode to its left side, with 10V in and 5V out. IRF9530 is not a bad choice, has good reverse capacitance for its transcoductance. Did you put it to the audio test at all by now?

 

[merlin el mago]

carbon film is easy to find locally.

i wonder if something better is needed.

No needs something better.

 

[nedi]

Salas shunt reg V1.2R

Hi Salas,
Is it better to use for 5Volt output R6 2K2 and R10 15K ?

Best regards.

 

[orelli]

Hi. CCT means CirCuiT. So you used the one that has Jfet cascode to its left side, with 10V in and 5V out. IRF9530 is not a bad choice, has good reverse capacitance for its transcoductance. Did you put it to the audio test at all by now?

Not, cause I've soldered a BG 220uF/16V at the PS input terminal of my Buffalo dac, and I'm afraid to damage the four layers board desoldering the cap. So I connected another identical cap in parallel to the Rload ( of course I don't use the 47uF lytic),for testing before to connect the reg to Buffalo,

I know that I risking oscillations, and I want checking whit oscope, maybe tomorrow.
When I'll make the audio test ,surely I'll post the results in the forum.

buonanotte, Orazio

 

[canvas]

Hi Salas,

I read somwehere that you've mentioned the BJT CCS is equal to the performance of the jfet cascode CCS. Since the 2N5459 is hard to find, is the BJT CCS shown at post#3799 workable for 5V or 3.3V out?

 

[Salas]

Can you find J110 or BF245B? Any low noise NJfet > 2V VGS(OFF) > 6ma IDSS. The BJT cascode is equal in PSRR, you should use less than 1K for its resistors when working near 5V so there is Vdrop space for correct current in the LED (near 5mA to reach Vf nom). The way the CCS is arranged it drops one Vbe on R1 and the cascode sees ((Vin-Vbe)-VgsQ1)-VR4. See your Vin will allow at least 3V across the BJT cascode to GND not to totally squeeze it.

 

[Salas]

Hi Salas,
Is it better to use for 5Volt output R6 2K2 and R10 15K ?

Best regards.

Yes. Or R10 27K and 5K trimmer. Better filtering from the 4u7 that way.

 

[nedi]

Yes. Or R10 27K and 5K trimmer. Better filtering from the 4u7 that way.

Thank you Salas

 

[massimo]

Better filtering from the 4u7 that way.

Sorry, which way? 27K + 5K trim or 15K + 1K trim? I'm a bit confused....

 

[Salas]

27+5

 

[Salas]

How do you all think about to compare the Sallas regulator with all this types?

· Superpower compared to a Battery
· Superpower compared to a Jung style super regulator
· Superpower compared to LM317
· Superpower compared to LM7812
· Superpower compared to Dexa NewClassD
· Superpower compared to Burson
· Superpower compared to the Sallas regulator

You see between the trees the forest no more, there are too much choice and every company have a nice marketing strategy.

This is an enormous thread with around 4,000 posts please where are the final definitive schematic of the Salas regulator and post it here or in the other tread please.

You can find this tread here:
http://www.diyaudio.com/forums/power-supplies/185763-belleson-superpower-ultimate-voltage-regulator.html#post2520457

Regards,
Rudy

In which way to compare? You need best PSRR you cascade multipliers, you need low Z you make some shunt, you want good efficiency you make series, tight space is for chip, etc. Then you listen what matters most to the audio application in hand. Not easy to say one type's best IMHO.

 

[AndrewT]

Yes. Or R10 27K and 5K trimmer. Better filtering from the 4u7 that way.

What current do you assume to be passing the jFET Q9?
27k for a 5V output sounds mad ! That would require q9 current ~160uA

I would expect a 1k trimmer (Vr for R10) to be about right for 5V output.
R6 is designed to have 2.8mA passing. That determines the required resistor value.

 

[back]

guys when i reduce r1 to increase current voltage across it drops.

in one regulator i had to go r1=2.35ohm in order to reach 0.460/2.35= 195ma.

is it normal?

 

[AndrewT]

The Vgs of the CCS FET changes with current through it.
The datasheet shows this characteristic clearly.

If you change the current by changing the resistor then you have automatically changed the Vgs. The voltage across the resistor is LED Vref - Vgs - trace Vdrops.

It is far better to add a second resistor in parallel and remeasure the new voltage to calculate the new total current. If you have not hit your target simply swap the added resistor for another value from your resistor stock.

 

[massimo]

.... or select two Mosfets with similar Vgs.

 

[back]

The Vgs of the CCS FET changes with current through it.
The datasheet shows this characteristic clearly.

If you change the current by changing the resistor then you have automatically changed the Vgs. The voltage across the resistor is LED Vref - Vgs - trace Vdrops.

It is far better to add a second resistor in parallel and remeasure the new voltage to calculate the new total current. If you have not hit your target simply swap the added resistor for another value from your resistor stock.

i did that and it worked fine.

i just woried if it was normal.

thanks.