The simplistic Salas low voltage shunt regulator

[gallon]

The Simplistic Salas Low Voltage Shunt Regulator

Are we really sure that Salas is simplistic?

[AndrewT]

Dare we ask Janneman to test a P2P version to see if sims match real life?

[iko]

Originally posted by gallon


Are we really sure that Salas is simplistic?

The choice of wording is his, so I comply

Andrew, I think the community would definitely benefit from a battery of real life tests; I don't know Jan well enough to impose such a task on him, and also don't know if he's interested in this but would definitely appreciate any help from an experienced man like him. I know salas ran some tests with his super duper sound card. You may find those results in his simplistic RIAA thread. Alas, my equipment is not sensitive enough for such tests.

[iko]

The circuit in post #8 has been built by quite a few people. That's the version that I would recommend to anyone who is interested in this.

However, tweakanitis is a heavy burden for one to bear, and personally am heavily infected. The variation shown below offers a slightly better output impedance in the higher frequency, including audible range. In the bode plots below V(outa) is the circuit of post #8, and V(outb) is the circuit below. Note that higher idle current results in lower output impedance across the (almost) entire frequency spectrum.

Note, this is highly experimental and AFAIK this circuit has not been implemented yet. So, with a 1R resistor, the shunt mosfet passes 676mA, and is supposed to dissipate about 15W. We can perhaps agree that one that is willing to do this may not be all there in the sanity department, but hey, I am planning to try it out soon.




Edit: typo fixed.

[jameshillj]

Back again Ikoflexor,

Yes I did try it both ways ! The CT circuit, that is!
A brief genesis. The first supply was a simple C-R-C with good bits, added the Cmx, then the ZenMod "Shunty" series reg and then the poor Toole shunt - in the middle of this, tried the Minnesotta unit (not bad at all!) and some other variations.
Then the lashed up Salas, removed the Series Reg (you were quite right about that), and set about assembling the DOA unit and the +/- rail shunt (Salas - post 309) and played about with the different caps for sound.
Being quite mindful of Jan's advice about low ESR caps on output, I was quite pleasantly surprised to find the Sikorels, Silmics or FCs didn't cause any problems in the spaghetti patch - may not behave the same with a good pcb..
The sound was extremely good in my opinion, with a NOS dac source and Pass X2.5 preamp (the DOA still needs a buffer for the lightspeed)

Then I added that weird collection of bits called the Charge Transfer circuit between the 2 supply caps and the whole system became dead silent and much more detailed with no "grain" at all. That's when I heard all about the problems in the source - got a loan of a couple of good players to make sure (quite like that Linn CDP!)

So YES, the CT circuit is worth the trouble - I'm surprised we haven't heard from Murcurio before now - he's done quite a bit of work on it and uses it on his power amp, too.

Now, the pcb.
Using the Circuit above on your post #8 as the starting point, (and let's see if we can stick with the same component numbers all the time) there are a few questions to include some variations already tried.
1. One pcb for + and - rails?
2. Using inverted 2skk170s rather than j74? or other devices, pinouts?
3. pads for a k170 option for R10?
4. extra Resistor as per your new Cct in series with C1 or is that a SIM thing?
5. Is there a Resistor between j1 source and 0v?
6. Is there to be a Resistor between the Shunt M2 and 0v?

I would like a small adjustment to the voltage - about 0.5 volts to compensate for zeners, correct for amp offset, etc, etc. Would a small pot (50R?) between zener and rail stuff up the performance?
Could you check this out with your magic circuit cruncher thing!!

Suggestions are MOST WELCOME - I'll start out with the extra tracking that we learned about with the Toole shunt layout and see what happens.

I'll get started into the pcb tomorrow (or today actually!)

For now ... Jh

[Artie]

Originally posted by AndrewT
if you need a Vref between 1.3V and 2.2V.

Two LEDs in series can work out better.
By the time you get to three in series there are advantages to using either.

I'm curious . . . is there any reason not to use any of the dedicated Vref's that are available from Linear Tech, et al., as opposed to diodes or LED's? I realize that there's a small cost issue, but those chips aren't exactly expensive. Are there other issues, (fidelity?), that I'm not aware of?

Thanks.
Artie

[Salas]

Originally posted by jameshillj
2. Using inverted 2skk170s rather than j74? or other devices, pinouts?

2. Yes you can use inverted 2SK170s. Marinos did, and reported.

He used 2N5459s for first test. Only the one under the Zener MUST be a 5-7mA Idss 2SK170. But using 2SKs everywhere does a bit better.

[Salas]

Originally posted by Artie


I'm curious . . . is there any reason not to use any of the dedicated Vref's that are available from Linear Tech, et al., as opposed to diodes or LED's? I realize that there's a small cost issue, but those chips aren't exactly expensive. Are there other issues, (fidelity?), that I'm not aware of?

Thanks.
Artie

I preferred the simple and widely available Zener solution. Somewhere I have read that it is even better, but I have not tested.

Its a finished project as much as I am concerned. Many people are very happy with it. But of course its up to you guys if you want to test it with more elegant additions or performance boosting twists.

[Salas]

Originally posted by gallon


Are we really sure that Salas is simplistic?

Its a simplistic circuit concept. I made it so to pair the simplistic Riaa.
Its good trade off against parts count, cost, time, lends itself to P2P etc. It happens to give statistically good subjective results also.
Its accessible to the simple DIYer. Nothing like a super circuit or a new idea. No claims here.

[Salas]

Originally posted by Lumba Ogir
Salas,
it`s not a big deal after all.

Why would you like the current adj feature for the main current source controller BJT?