Regulated B+ ?

[rock soderstrom]

I use almost the same circuit in the majority of my tube supplies. However I don't use ZD1 because I'm not interested in clamping/regulating the voltage.

Thanks for your input!

I've found in practice, a bunch of RC filtering prior to the MOSFET isn't needed, and didn't result in much (if any) measurable improvement in ripple at the output. I always use 47uF/450V prior to the regulator, and the same 47uF/450V sitting on the gate (C1 in your picture), and get uV ripple on the output at modest currents (<100mA).

This roughly corresponds to the experience of member @Kingston in the linked thread. I wouldn't mind saving some real estate on the PCB. Good info, thanks! Then I will also increase C1 to 33uf then I can use one size of capacitor for the whole B+ power supply. My PCB should be able to cover everything from a 0.5mA microphone to a 50mA HPA. In front of the IRF840 I will place a CRC filter with two 33uF, one 3W resistor (as required) and a 100nF MKP in front of the whole thing.

I would also increase C2 to 33uF, nice to have or counterproductive?

Note that when the pot VR1 is set to half, you have 500k feeding into the 100K R1 that charges C1, which means that the MOSFET gate can see time constants over 10 seconds, meaning that it can take a full minute for the output to establish itself. For a tube circuit, this can be nice, as it gives time for the heaters to warm up and the tube to begin operating. But it can also mean that you twiddle the pot and the output doesn't change for a long time.

If you have a high incoming voltage, and a large adjustment range, then heat-sinking the MOSFET becomes mandatory, doubly so if you think the circuit might be operating into short circuit conditions.

I had already expected this, a heat sink is definitely planned.

 

[Matador]

I would also increase C2 to 33uF, nice to have or counterproductive?

The output impedance seen looking to the drain of the MOSFET is pretty low - as a result, increasing C2 doesn't really reduce the ripple measured at the output. In my testing, 100n was ample, especially for audio circuits.

If you want to increase the uF, it's better for the caps that will be multiplied (e.g. sitting on the MOSFET gate) rather than on the output.

 

[rock soderstrom]

The output impedance seen looking to the drain of the MOSFET is pretty low - as a result, increasing C2 doesn't really reduce the ripple measured at the output. In my testing, 100n was ample, especially for audio circuits.

Okay, I understand. I've also seen this in other circuits, they all used rather small film capacitors, I'll just try it out. Thanks.

 

[ruffrecords]

Many thanks for the comprehensive explanation, Ian!

Now I know roughly how to fish! I'm still not quite sure what the Kingston variable voltage reference adds in terms of ripple reduction.


I don't quite understand fully this circuit yet, I see it more as a regulator, the zener diode (plus the trimmer) gives the voltage reference for it. If this is a capacitance multiplier, then I can easily increase C1 (in the Kingston circuit) to 33uF, right?

It is a regulator. I had not noticed the zener!. A1W 200V zener will have a typical impedance of about 1K5 at 1mA so with the 15K feed resistor this will attenuate the ripple by an additional 20dB. Then there is there is the 100K/10uF RC of the capacitance multiplier which should drop ripple by another 2 X Pi x f x R x C times = 628 times = 56dB

To optimize my circuit a bit more I suggest to add another 33uf/1K RC filter, simply because I don't have 100uf capacitors at hand, so according to your calculation I would get about 78dB ripple filtering before it goes into the Kingston circuit. That should be enough for a clean B+ voltage, right?

That will help. With the zener plus cap multiplier you should have nipple in the sub microvolt region.

Thanks again for your support!

My pleasure. None of this is very complicated and with a little bit of arithmetic anyone can do it.

Cheers

Ian

 

[rock soderstrom]

It is a regulator. I had not noticed the zener!. A1W 200V zener will have a typical impedance of about 1K5 at 1mA so with the 15K feed resistor this will attenuate the ripple by an additional 20dB. Then there is there is the 100K/10uF RC of the capacitance multiplier which should drop ripple by another 2 X Pi x f x R x C times = 628 times = 56dB


That will help. With the zener plus cap multiplier you should have nipple in the sub microvolt region.

My pleasure. None of this is very complicated and with a little bit of arithmetic anyone can do it.

Cheers

Ian

Thanks again, that really helped me. I've already put my new knowledge into practice a few times today. I can see now!

As you have probably already read, I am still wavering a little about the amount of Pre IRF840 filtering. I think I simply have to test it out...

WIP: ...it's not so easy to get the changes onto the same board again. It will take ages again, but eventually it will fit. I'm so lame at designing PCBs...


Single-sided circuit boards don't make things any easier.