The simplistic Salas low voltage shunt regulator

Status
Not open for further replies.
Jump to Latest
#6,222
Salas said:
Not 0.1uF only, small super low ESR films in general will interfere with the fixed Zobel termination of the Bib if there is not much inductance in between to really decouple and will cut a lot of phase margin. They have caused oscillation even. We had a couple of practical examples in this thread where they have caused instability that fixed with their removal. Harsher flat sound like when with EMI interference is a clue. Reflektor is more tolerant but its better without too. You can try without any film on the load's rail, short run of 4 wire cabling, or tantalum rail bypass if it has something to really do there. The shunt reg will damp the load's rail differently than a series one, especially on 4 wire connection.
Click to expand...

i'm using bib's 1.2V (with D45) and 3.3V (with bjt) for Ackodac, that has 100uF (electrolyt) and 0.1uF (film) as decoupling on all analog and digital supplies. The wirelength from bib to ackodac is about 2 cm.

What would be the best option.
- remove 100uF an all supplies
- remove 100uF AND 0.1 uF on all supplies?
- drop the zobel filter on bib?
- use 47uF electrolyt cap on bib, and remove 100uF an all supplies?

- different meassures for analog and digital?

Is there any other method to meassure what is best - except listening-test and theoretical reasoning?

Peter
 
#6,223
First you should use the oscilloscope. By looking at the rails in AC coupling input switch mode. If you will see no sine wave like oscillations then the system is OK as it is.
If its not I would remove the very small value capacitors first.
 
#6,227
pchw said:
Lookie lookie, my first reflector powered up, and a little adjustment at the pot, it was dead on 5V, so happy 🙂 Will swap out the Bib with this later for a comparision.
Click to expand...
Will need a little advise from the expert.
The reflector has been working flawless for 6 months, but when it is compared to my BiB 1.1, I always feel the high is not as crisp. I think I can use some expert advise here. The circuit is based in the a variation of the 5V as in the attachment.

The major changes (as of now) are:
1. The Zobel Cap (C5??) is a 34nf Mica instead of the 0.47uF
2. The 2 x 820uF is replaced by a 2700uF bypassed by a 0.22uf Epcos polypropylene.
3. M2 is IRF610 instead of IRF640.
4. The 2 x 560R are replaced by 1 x 1150R.

Any suggestion to components change to try? The load is no mor ethan 75ma, and the CCS is set to deliver approx 250ma, and the output is 5V.

Thanks!!
 

Attachments

  • Reflector-5V.png
    Reflector-5V.png
    26.5 KB · Views: 585
#6,236
:worship: Salas :worship:
It is no kidding. Just 2 seconds in the intro, I knew the man's guidance was what I had been seeking for the last 3 months!! I have been using the IRFBC40 in my capacitor multiplier a few times but never even thought one second that was what I needed 😱

I am still in denial of how a recycle (yeah, I recycle it from a broken computer power supply) device can provide such dramatic difference!! Even at this early stage, I can say the reflector is at least as good as the BiB.

FWIW, this devices does have lower Vgs than the IRF610. I think it is around 3.7V. So, I needed to adjust the voltage upward. Otherwise than that, it was pretty much clip and solder affair.

Again, thank you, Salas!!
 
#6,238
The STP/MTP are in my Mouser shopping cart. So, they will be in my next order once I accumulate enough to commit.

One educational question, why BiB doesn't need a high Gfs shunt device. The irf9610 is pretty low, but it works just fine in the BiB. So, is the choice of the shunt devices a function of the error amp topology? The current mirror requires higher gain device to work wth?

Thanks!!
 
#6,239
There is the IRF9530 in the shunt position officially designated for BIB not the IRF9610.

The reason that Reflektor appreciates better more transconductance in M2 is that it does not have an error voltage amplifier to push it but a mirror "bridge" instead to tell M2 how to current compensate the voltage output to Vref balance. So M2 has to listen easier, needs be sensitive. Its a simpler elegant notion and has no strong phase consequences.
 
#6,240
Thanks for the explanation, Salas.

It is IRF9530, my memory really isn't what I want it to be 🙂
Sorry for all the questions, sometimes I just have the urge to want to know the why even though things are working. Learning never stops at any age 🙂

Have a happy Thanksgiving to all!!
 
Status
Not open for further replies.
Top Bottom