The simplistic Salas low voltage shunt regulator

[AndrewT]

pinch off is related to Idss.
I would expect the average pinch off of GR to be about half of pinch off of BL.

Do you want me to measure a few?

Look at p2 graph Id - Vgs the bottom two curves are GR, the top two curves are BL, the middle curve could be either BL or GR.

 

[Salas]

Yes, if you can measure a circa 9mA BL and a circa 4mA GR, please do, so I can see how far off the Spice models are.

 

[fred_com]

I say better use a 2V LED and a diode in series instead of a resistor and delete the FET's trimmer, just connect gate to source. That will give 3.3V more or less, very steady, super low noise, max CCS current. Cathodes facing towards the Vref's CCS FET. No need to change the bypass cap.

I see, can you please look at the circuit attached, is it correct, considering all the changes?

Thanks,
Fedor

 

[Salas]

No mistakes.

 

[AndrewT]

Vsupply 10.0V
1k0 resistor in drain supply.
1M0 source to gate.
Tj unknown
Idss 3.95mA Vpinch off (Vgs) = 510mV, drop across 1k0 0.4mV
Idss 9.05mA Vpinch off (Vgs) =720mV, drop across 1k0 0.7mV

Replacing the voltage setting resistor with diode/s looks a much nicer way of getting super low output voltage.

 

[fred_com]

Thank you so much Salas. Will post back with results.

 

[Salas]

Vsupply 10.0V
1k0 resistor in drain supply.
1M0 source to gate.
Tj unknown
Idss 3.95mA Vpinch off (Vgs) = 510mV, drop across 1k0 0.4mV
Idss 9.05mA Vpinch off (Vgs) =720mV, drop across 1k0 0.7mV

Replacing the voltage setting resistor with diode/s looks a much nicer way of getting super low output voltage.

Thanks for the data Andrew.

 

[Salas]

Thank you so much Salas. Will post back with results.

You wanted the 3.3V and now we find its tricks. Lets see how it goes.

 

[fred_com]

I'm sorry, I don't understand, you want me to wait until some other changes may come up?

 

[Salas]

I meant it proved it needed more changes than simply putting a BD140, and we wait for your testing news for all the changes we thought of and are included OK in your latest diagram.

 

[fred_com]

Oh, I see, thanks.

 

[fred_com]

Salas, I've just tried the new circuit, and the output is 2.94v, LED is shining. Thing is that I've used 2.2v green LED (I hope I measured it correctly), is it responsible for setting Vout? Are there any control points that I can check? Oh, and input voltage is 9v, if it matters.

Thanks,
Fedor

 

[Salas]

Did you use the diode too? Vout is Vref+Vbe. Vref are the elements in the former resistor's place. Add one more diode if its not enough and try some combination with the proper LED to go around 3v3 target.

 

[fred_com]

Yes, used the diode too. Tried with two diodes - output is 3.07v. Voltage across Vref elements is 2.40v and Vbe is 0.67v. So this is correct. And we can change only Vref by using LEDs, right?
Then, I guess, I need to go to the store for some 2.0v LEDs

Thanks,
Fedor

 

[stormsonic]

@fred_com
yeah, try with different LED's.
For Q5 (2SK170): use other 2SK170 with higher/lower Idss. LED's are not perfect voltage references, voltage forward is changing with different current through them.

Another idea:
How much current you want/need to shunt? 50mA? 100mA?
What is your input voltage? 5V, 8V, 10V?

 

[Salas]

With one 2V LED and two diodes its kinda weird it won't go at a higher Vref than circa 2.4. A BL under Vbe should be giving 5-6mA if it is 7-8mA Idss at 9-10V. Either that CCS fet isn't pulling enough, or the LED is sleeping (is it glowing enough?) or there is too much current taken by the load versus the main CCS setting and the reg is starved for current in its own shunt elements.

 

[fred_com]

It works now! The problem was not in LED but in diode - first I used 1N5819, left from other project, but then I remembered that you've recommended 1N4004, found them when looking for some more LEDs. So final circuit is 1N4004 and generic 5mm green LED. It gives out 3.32v. The sound changed to the good side, highs are crystal clear, scene is widened, some microdetails revealed.

Thanks for your support, Salas, you've been a great help. And I'm sorry, that I didn't listen to you from the begin, and started using wrong diodes.
Oh, I changed Q5 and Q6 to GR grade, I think it's ok, considering that everything works.

@stormsonic
as you can see, the problem is solved, but thanks for the support

 

[stormsonic]

@stormsonic
as you can see, the problem is solved, but thanks for the support

Great, I'm happy that now everything is working OK
I was just offering you to change 3 elements for different values and to remove additional 5 elements.

 

[Salas]

It works now! The problem was not in LED but in diode - first I used 1N5819, left from other project, but then I remembered that you've recommended 1N4004, found them when looking for some more LEDs. So final circuit is 1N4004 and generic 5mm green LED. It gives out 3.32v. The sound changed to the good side, highs are crystal clear, scene is widened, some microdetails revealed.

Thanks for your support, Salas, you've been a great help. And I'm sorry, that I didn't listen to you from the begin, and started using wrong diodes.
Oh, I changed Q5 and Q6 to GR grade, I think it's ok, considering that everything works.

@stormsonic
as you can see, the problem is solved, but thanks for the support

Avoid Schottky diodes in the Vref (1A to go 0.6V)... Nice then (horosho?), please post the total final well working cct for any guy's reference who will be interested in a tested 1.2(ish) based 3.3V solution in the future. We like pictures too.

P.S. You mentioned that you had 1.0 based regs there before, correct? Now its a 1.2 5V all Mosfet and the 3.3V with a BD140 in the shunt position we ended up with, right? Was there remote sensing implemented for the 1.0s too? Can you differentiate the before and after regs subjectively?

 

[AndrewT]

I would like to see, in the higher voltage versions, where a resistor is used to set up the voltage reference, a three component string consisting of resistor, LED and Zener.

For low voltage the Zener is optionally replaced with a diode and the resistor can optional be linked out. The LED shows it's working as well as giving us the low noise, low impedance reference voltage that we need.

How about a proposed V1.0a and V1.2a with these modifications?