The simplistic Salas low voltage shunt regulator

[Ryssen]

Learned hard way that more important than single super regulator is local bypass for each chip and to "isolate" each chip to prevent him from interfereing with other chips. For "digital" chips.

How is this "isolation" done,best way?

 

[RCruz]

IMO, best isolation is obtained by using different TX + rectifier for each chip.

I have used a dedicated TX + rectifier for each reg. There are chips (like a DAC or DECODER) that need several +5vin and I got best results powering each +5Vin line with it´s dedicated TX + rectifier + superreg.

 

[fred_com]

Capacitors at power supply pins.

Sorry, I can be little slow sometimes You mean decoupling electrolytic caps on the supply pins of the device?
And can you please elaborate a bit more on your modification of 3V3 reg with DN2540? You are using it to feed digital parts too, right?

Thanks,
Fedor

 

[stormsonic]

Do you compensate for stability at the output, or with lead lag internal, or with both?

P.S. Have done once that page 4 one in the link for tube anode CCS. Hiss went up so much I went back to a resistor.

Latest version is opamp based. Very fast opamp. Bootstraped, opamp will self regulate his own power supply over 1R resistor + capacitor to ground. This 1R resistor have multiple function; to isolate opamp Vcc from output, to clean opamp power supply by forming RC LPF with cap connected to opamp power supply pins, to measure current drawn by opamp and to provide some R for low ESR cap.

Yes, there are some drawbacks, you have to watch on CMIR and output voltage range.

Sense wires can be used, no problem with this. When acting as 3-pin regulator replacement into existing design & very close to load, then no sense wires.

And no sense wires when output is set to 4V and additional NPN at chip supply pins is used to isolate this chip.

Sorry, I can not show regulators schematic, it is similar to comercial design.
Anyway, it is similar to other opamp based designs

Walt Jung have exceptional designs on his Webpage

@Ryssen
Figure 1
Another nice example for low loads
http://www.audiofil.net/data/upload/20091207_190756_Mini_shunt.gif

@RCruz
Yes, if you have enough space, $$$ and doing your own design.

@fred_com
look on picture posted, where CCS MOSFET is replaced with DN2540

 

[Salas]

Is it still a parallel reg or you changed to series? When you say additional NPN you mean a local capacitance multiplier?

 

[fred_com]

look on picture posted, where CCS MOSFET is replaced with DN2540

Thank you, maybe will try that later on.

 

[merlin el mago]

Will be good to supply hiFace clock 3.3V?

 

[Salas]

Mind you its not purpose made to be a BJT under 5V shunt reg. Its the needed fix for it just to work decently since it was assumed wrongly initially, plus a depletion Mosfet instead of the more elaborate CCS that Storm proposed as good enough for digital. A curious hybrid that got born along fixing Fred's...

 

[merlin el mago]

Salas, so your point of view is that is better use MJE350 for voltages under 5V, I suppose that your schematic ShuntTT DC motor 2.5V out can be configured Remote Sensing but for me it's a little complicated to get, could you show me how do the RS?

 

[Salas]

Without a buffer, all BJTs will give good frequency for Zo in the shunt position, but not very inspiring Zo value. MJE is only 500mW, a good TO-220 choice would be the one we used with Disco for the DHT tube's heater No26 preamp 1.5V reg. D45H8.

http://www.diyaudio.com/forums/tubes-valves/151421-26-pre-amp-22.html#post1996668

http://www.diyaudio.com/forums/powe...-voltage-shunt-regulator-162.html#post2014178

There you can see remote again, its exactly like you remote a V1.0 for Fred's too.

 

[merlin el mago]

Thank you.

 

[fred_com]

Without a buffer, all BJTs will give good frequency for Zo in the shunt position, but not very inspiring Zo value. MJE is only 500mW, a good TO-220 choice would be the one we used with Disco for the DHT tube's heater No26 preamp 1.5V reg. D45H8.

Hi Salas, may it be better to use D45H8 in my case, instead of BD140?

 

[Salas]

In your case no, since the current you use is enough, BD 140 has better Ft. I think Merlin runs his regs for digital very heavy, and then a TO-220 is taking it. Also the D45H8 has half the Vbe of BD140 and with it it was feasible to go 1.5V 1A.

 

[sandbasser]

I've been studying this thread (and the other similar ones on this forum) and I'm completely puzzled about using LEDs as VRefs. Can someone please give me some references of where to look to gain more knowledge and understanding.

Thanks...

Also, I'm wondering about measuring the voltage drop of an LED. I saw a discussion in one of the threads about a using 9v battery a 1K resistor and a DMM reading the voltage drop across the LED. I'm not sure I fully understand this. Can anyone offer an explanation.

Thanks again.

I've learned a lot, I think I just need a little more knowledge to pull it all together.

 

[Salas]

The problem for someone coming late in these threads is that they are mainly about build feedback and help desk, beyond the initial concept's ironing out etc. Kinda live reports, then historical. The body of posts is inversely proportional to the simplicity of the concept. But thats how it gets when something has good fidelity to complexity ratio. Not amazing or too brilliant but helpful for many, and doable by many. You get my drift I hope.

Now about LEDs. Imagine them as low Vf, very low noise, glowing Zeners.
You would need some appreciable current to get them in a good region for best noise, nominal Vf, low impedance. 5-10mA is good. So you simulate that with a battery and a resistor makeshift quick test if you need to short them by Vf. Its one way. Another one may be fed by NJFET with G&S shorted. A 2N5459 is fine. That's all.

About references, there were discussions in DIYA, some are LED masters around here, search a bit, I don't have anything bookmarked right now. Avoid blue and white color LEDs, just that. IR, Red, Green, Yellow, are OK.

 

[fred_com]

Thank you Salas.

 

[sandbasser]

Thanks... time to do more reading.

 

[stormsonic]

Something about voltage references: LINK

Thermal drifting for LED: LINK

 

[albertob]

Hello Friends
I've built a B1 and I am very happy with the result. I want to upgrade the power supply with low voltage shunt regulator Salas, to see differences. With the current power source is a bit of noise.

I built the following schedule posted in the forum since my knowledge is basic:

An externally hosted image should be here but it was not working when we last tested it.

I measured and I have a 26V output.

What I have to change for 24V?.

I've tested with no load (with multimeter) and within seconds Q4 gets very very hot, and that having a sink.

Is it normal?, How I can simulate a load to test it before connecting it to my B1?.

Thanks for the help.

Excuse my English. I used a translator

 

[Salas]

Hi,

First of all delete the 0.1uF parallel to C4. It probably gives oscillation.
Without any load all current goes through Q4 and its normal to get hot. Don't test for long without a power resistor 470R 5W as load. Make R1 10R 2W. You run about 300mA right now with 6R8 and it must be very hot for it.
Measure mV across R3 and tell me, so we will change the Zener to a resistor for getting 24V and lower noise. Also change R6 to a 2SK170 connected as Q2, from +24V pin 1 to BC550's collector pins 2,3 for better performance.