B+ regulator with variable reference

My3gger said:

What values have input filters in ver.4 at 20mA? Have you also tried 1meg pot at 300v in ver.4?

Click to expand...

That 1 meg works for all variations and voltages, no need to change anything there. I've tested this at 250VDC now (input about 280VDC) 30mA too. I'm sure that 4.7uF filter for zener reference is enough for all cases too. When testing I've had input ripple in the range of 0.5-1VAC. That will be gone completely. Output is very clean indeed.

300VDC should work fine as well. See PRR's post how to calculate the watts and the input resistor for the zener, that's the only thing you need to change. Pay attention to MOSFET cooling when passing a lot of watts.

My3gger said:

Btw, Fearn uses simple zener shunt regulator and it seems ok there.

Click to expand...

They work, but this solution easily scales to much more than just 100mA while I wouldn't trust a zener there. This has far less noise too.

I thought about R1,2 (with filter caps) as in first schematic, before Analag's post. If input into the regulator was 280VDC for 250VDC output, you somehow dropped ~40VDC before regulator if transformer remained the same. This is what i would like to understand better. Sorry for confusion.

Ask the ohms law how much voltage will drop through a resistor if current is 100mA.

Yes, Ohm's law tells me that values on 2nd schematic give 40V drop. I asked about it because Analag's post made me think there is something more to it than usual dropping/filtering resistor/cap(s). Thanks!

I'm interested in building your B+ Regulator for the capsule polarisation and tube anode of a microphone. But I'm unsure about a few things and would like to understand it better.

Am I right, that R1 sets the min and max current for the circuit?
I have a relay for disconnecting the back diaphragm. The relay needs a switch on current of about 5 mA. The capsule/Anode needs about 0.5 to 1 mA.
Now the question is, which Zener should I use for ZD1? If I use a 200V 3W (ON Semiconductor 1N5956B) Zener, which has a reverse current of 1.9 mA then R1 should be set for about 8 mA, right?
So if I would like to have a range of 110-210V and current of 8 mA R1 should be 2k5 Ohm and input Voltage 230V.

Are my calculations right?

Is there anything else I should change in your circuit when using it for such a small current consumption?

Murdock said:

I'm interested in building your B+ Regulator for the capsule polarisation and tube anode of a microphone. But I'm unsure about a few things and would like to understand it better.

Am I right, that R1 sets the min and max current for the circuit?
I have a relay for disconnecting the back diaphragm. The relay needs a switch on current of about 5 mA. The capsule/Anode needs about 0.5 to 1 mA.
Now the question is, which Zener should I use for ZD1? If I use a 200V 3W (ON Semiconductor 1N5956B) Zener, which has a reverse current of 1.9 mA then R1 should be set for about 8 mA, right?
So if I would like to have a range of 110-210V and current of 8 mA R1 should be 2k5 Ohm and input Voltage 230V.

Are my calculations right?

Is there anything else I should change in your circuit when using it for such a small current consumption?

Click to expand...

For clarity let's assume we are talking about the latest version pictured in https://groupdiy.com/index.php?topic=51232.msg651854#msg651854

I think this article gives good tools to calculate and select R1 and Z1 for minimum current consumption.

https://www.electronics-tutorials.ws/diode/diode_7.html

This part of the circuit serves as reference for the MOSFET and you can mark that current down as "almost nothing". Choose a Z1 slightly above the desired maximum voltage required for the regulator output. This "crude" reference is  further RC filtered by VR1/C1, dropping the reference down to the actual desired output voltage. VR1 can be a high voltage pot,  stepped rotary or just static resistors.

You can optimize further from the picture. R1 and the high-watt zener was simply selected so that it can work in a wide range of voltages. As in, Z1 could be 50V or even 320V.

Also note that Z1 can just as well be a series of zeners to reach the desired crude reference.

For capsule polarization you can completely skip the MOSFET and feed it directly from R3/C1 reference. Add some sensible load - let's say 100k - paralleled with C1 to stabilize exact voltage since polarization itself draws "none". Probably works just as well for the relatively small plate current - even without the artificial load of course.

You only really need the MOSFET to offload heat to a heatsink for higher currents.

Kingston said:

Since it was you who pointed this direction in the first place, here's rev4, thanks!

Click to expand...

D1 is not essential; the MOSFET body diode performs the same job.
This circuit is similar to a quick and dirty stabiliser I mention in my book:
https://books.google.co.uk/books?id=VMjkDQAAQBAJ&printsec=frontcover&dq=merlin+blencowe&hl=en&sa=X&ved=0ahUKEwjlkOytjtngAhUHrxoKHYDXAusQ6AEIKjAA#v=onepage&q=Fig.%2011.46&f=false

Google's Look Inside can be funky. Here's a low-res copy:

Would there be any advantage to having R3 (in merlin's design) replaced with a CCS?

I've mocked up the basic MOSFET-follower topology listed here, and observed that the ripple reduction wasn't much better than could be had for a few cascaded RC stages.

As a test, I wired up Kingston's circuit as a 48V supply using an 80V Zener reference.  I took an existing 48V DC supply, and injected a 1V peak 120Hz sine wave on top to simulate  ripple from a rectified supply.  I trimmed the supply to 24V with a 15mA load (1.6K load resistor), and looked at the DC output across the load resistor.  The ripple was about 15mV peak-to-peak, which means this supply has a ripple reduction of about 42dB.  Two, 100R/100uF passive RC filters are about 50dB down.

I'm wondering if the limited transconductance of the MOSFET along with the high drain to source capacitance are limiting performance.  Maybe a feedback/error amplification stage might help in this regard?  Last time I tested, several of the discrete BJT designs I found on the internet that incorporated voltage feedback were capable of 70-80dB of ripple reduction, which is 5-15dB better than most integrated voltage regulators (at the cost of double or triple the quiescent current).

Matador said:

Two, 100R/100uF passive RC filters are about 50dB down.

Click to expand...

Just that they are terrible to control, don't work as bench supply and any significant variance in current makes a static config like that useless. Completely different - and very specific application for any plain RC.

Kingston said:

Just that they are terrible to control, don't work as bench supply and any significant variance in current makes a static config like that useless. Completely different - and very specific application for any plain RC.

Click to expand...

Of course, I wasn't suggesting that the regulated circuit doesn't have advantages onto itself.  However if you are powering noise sensitive circuits one might think to apply some RC filtering pre-regulator if you can afford to throw away some of the voltage before the MOSFET.  Then you'll get the best of both worlds.

This may become critical for circuits with poor PSRR like vanilla "common anode" gain stages, etc.

Matador said:

However if you are powering noise sensitive circuits one might think to apply some RC filtering pre-regulator if you can afford to throw away some of the voltage before the MOSFET.  Then you'll get the best of both worlds.

Click to expand...

Exactly. This is just a building block. The very first post was an attempt at a complete example application, optimised for a specific job.

Thanks for the measurements and investigation. I never bothered further analysis due to the fact in my task-optimised implementations it has delivered flat noise floor. Naturally, if we can come up with significant improvements for this MOSFET dropper, everyone benefits.