The simplistic Salas low voltage shunt regulator

[samoloko]

It Is almost a copy of your prototype pcb layout published

 

[Salas]

Try a Zener reference. Any, even 18V to check. If its not oscillation and the cap does not have a problem it should go Vz+VFled+2Vbe=Vo. Zener cathode orientation vice versa to LED.

 

[Mull3t]

Hey all,

I've read good things about this shunt regulator and decided to use one as a supply for a Joachim Gerhard Buffer. The buffer requires +/-15v and 25mA per channel. The things I'm looking to accomplish are: what should I use BJT or IRF? and would this... http://www.digikey.com/scripts/dksearch/dksus.dll?vendor=0&keywords=TE62055 ...work for a transformer? Amveco makes a dual 18VAC version as well, but from my reading of the BiB guide it looks like the dual 22VAC version would be a better choice.

@Salas I saw your response... http://www.diyaudio.com/forums/powe...-voltage-shunt-regulator-116.html#post3402287 ...about how to pick a transformer. I'm still trying to wrap my head around this. I'm a bit of a beginner on the design side of things and have had my hands dirty for quite a bit when it comes to "paint by numbers" electronics. This is my first foray into trying to get out of that box.

Thanks,
Scott

 

[dimkasta]

Alright, see with what cap where does least possible Vpp of whatever circulates or emits and you can't do much more, then leave it alone.

I have added a 1000uF FR cap on the shiga entry and things are down to

No patience right now to experiment with lower values and different types...
When I start wrapping things up, I will also up the CCS current a bit

 

[Salas]

Hey all,

I've read good things about this shunt regulator and decided to use one as a supply for a Joachim Gerhard Buffer. The buffer requires +/-15v and 25mA per channel. The things I'm looking to accomplish are: what should I use BJT or IRF? and would this... http://www.digikey.com/scripts/dksearch/dksus.dll?vendor=0&keywords=TE62055 ...work for a transformer? Amveco makes a dual 18VAC version as well, but from my reading of the BiB guide it looks like the dual 22VAC version would be a better choice.

@Salas I saw your response... http://www.diyaudio.com/forums/powe...-voltage-shunt-regulator-116.html#post3402287 ...about how to pick a transformer. I'm still trying to wrap my head around this. I'm a bit of a beginner on the design side of things and have had my hands dirty for quite a bit when it comes to "paint by numbers" electronics. This is my first foray into trying to get out of that box.

Thanks,
Scott

The SSLV1.1 (Back In Black) guide has most stuff you need to know. You just get a GB board and you are sorted. Both 18+18 or 22+22 VAC Tx are good for your purposes. That example is about 35V scale so choosing VAC about equal as VDC needed covers a 10V drop which is good for lower capacitance in the CCS MOSFET and covers logical mains variations from place to place or hour to hour services.

 

[Salas]

I have added a 1000uF FR cap on the shiga entry and things are down to

No patience right now to experiment with lower values and different types...
When I start wrapping things up, I will also up the CCS current a bit

It looks good VS what you initially got. Its no instability and you ate away much weird stuff you could pick so its a good thing. Plus the counter shows nothing specific.

 

[dimkasta]

It looks good VS what you initially got. Its no instability and you ate away much weird stuff you could pick so its a good thing. Plus the counter shows nothing specific.

To sum everything up and build some further understanding, we had ~200mV EMI that was playing tricks with the sense lines.
By capping the sense lines we slowed them down and saw the actual raw EMI as plain random noise, so we had to cap it to eliminate it.
We basically "limited" the sense functionality?

 

[Salas]

No. You just filtered crap emissions on the rail line locally at the Shiga.

 

[dimkasta]

So should I remove the 47uF from the sense lines?

 

[Salas]

Remove it to see.

 

[vgeorge]

Try a Zener reference. Any, even 18V to check. If its not oscillation and the cap does not have a problem it should go Vz+VFled+2Vbe=Vo. Zener cathode orientation vice versa to LED.

Here is one I made with the 3055VL. Target was to get 35 Vout. Vin was 43V. I had a problem in the beginning to get more than 15V with the 5K trimmer and I had to go up to 15 - 20K to reach the desired voltage. But could not get the voltage to be stable and not to drift. Maybe due to thermal issues (4404 too close to the heatsinks?), or due to my stash of 4403 which seemed to be higher Hfe. Changed the trim and led you see on the pictures (also removed the zip tie on the lower pair) with some combination of zeners and diodes and is stable right now. Also tried an 640 in place of the 3055VL but here was no difference in the drift.
Could not listened to it yet as I am too busy at the moment, will see when I get some time.

 

[samoloko]

thank you Salas and vgeorge

4403 are far away from Mosfets - I don't have heatsinks at all because I test only for 20 sec
I am almost sure that there Is no oscillation - why It will work to 14 V and not more
will see with a led string or zener

BTW I removed Vref filtering cap

 

[vgeorge]

20 seconds maybe not enough time to reach the desired voltage, even without the Vref cap. Put some sinks even small ones and test for a few minutes.
You can even do a combination of leds zeners and diodes.

 

[Salas]

@ Vgeorge:
3055VL is under 2V VGS at that bias so it would take a big resistor. Still it climbed.
@ Samoloko:
There was a connection error in that Eagle test layout around the output Mosfet pins so follow the schematic and not the vias blindly.

 

[samoloko]

Salas
you already told me about that - thank you
I think output fet Is connected correct for positive - drain to other fet drain, source - to gnd
now I have 9 leds In string + 1 K as a Vref
output Is 14 Vdc leds does not light
there Is an error but don't know where

 

[AndrewT]

I have added a 1000uF FR cap on the shiga entry and things are down to

No patience right now to experiment with lower values and different types...
When I start wrapping things up, I will also up the CCS current a bit

I would use inspired guesswork to suggest the Shiga has not got properly designed decoupling built in to the correct locations and of the correct sizes/types.

 

[vgeorge]

@ Vgeorge:
3055VL is under 2V VGS at that bias so it would take a big resistor. Still it climbed.

What I meant was that even with the 640, the resistor value was higher than what samoloko used.

 

[dimkasta]

I would use inspired guesswork to suggest the Shiga has not got properly designed decoupling built in to the correct locations and of the correct sizes/types.

You would probably be right...

I plan on separating some more stuff with their own PSUs... we ll see...

 

[AndrewT]

It's not PSUs, it's decoupling.
The faster signals change the more important decoupling becomes.
Digital signal processing cannot operate properly without a lot of both MF and HF decoupling.

 

[dimkasta]

Still with such complex devices with fast twitching motors and digital controllers divided in many subsystems, decoupling should be able to fix some stuff but probably not all when there is only 1 PSU...