Need help: MOSFET-boosted High Voltage regulator?

[km-r]

Im currently constructing a tube amp with two EL84 running SE at about 40mA each. Now my power transformer has output at about 240VAC, silicon rectified to about 340V. with limiting values of about 300V, i think running that HOT will shorten the tube life.

i would place a large-ish resistor to drop about 60v off the max voltage but this will dissipate about 6watts while not providing regulation.
i would go for an active way with a heatsinked MOSFET/BJT-boosted zener regulation.

heres a rough sketch of my proposed PSU setup.



I KNOW THIS WILL NOT WORK AS IS. i have not tried using MOSFETS as boost elements for Zener regulators.
i would like to ask for additional components to add here?
most schematics ive seen are complicated with slow-ramp features which i dont really need.
how about the zener resistor R?

thanks in advance!

[stephanovitch]

Regulation with mosfet:
http://www.dissident-audio.com/RegulHT/Regul.html

Explanation in french

[km-r]

haha guess someone didnt do a thorough search... thanks man!

[merlinb]

You should add a protection diode between gate and drain (anode to gate), and a 12V zener between gate and source (anode to source) but apart from that it will work as shown. Choose R so that a couple of milliamps flows down the zeners. You don't need much, since this is not a critical application.

[defaced]

Check the datasheet for the exact MOSFET you plan on using.  Some have some/all of the protection diodes inside of them.  But as Merlin has pointed out, yes, this schematic will work just fine as is.  I'm using a similar setup (though my reference voltage comes from the junction of some series connected filter caps) on a preamp test rig I'm working on. 

[PRR]

> i think running that HOT will shorten the tube life.

Nothing lives forever.

EL84 is very inexpensive to replace when/if it does cook to death.

EL84 at 13W Pdiss can last a VERY long time.

RCA ran 6BQ7/EL84 at 17W Pdiss in their stereo tuner/amps.

Since the amp is SE, drop about 10% in a resistor and add a second filter-cap to get cleaner B+. 250 or 500 ohms 10 Watts and another 40uFd-100uFd.

In my personal opinion..... regulation is rarely good for tube audio amps. It's more stuff to buy and wire and fail. GOOD regulation has to start with MUCH more voltage, or else it will fall out of regulation when wall-voltage sags. That means it works and sounds different between evening (low line) and midnight (high line).

I guess 280V from 340V meets my idea of "good regulation" overhead. There's 5 Watts wasted in the MOSFET (same as a resistor dropping the same voltage); more at high line voltage. You won't dissipate 5 Watts on a 99 cent heat-sink, it needs thought and space and air. The heatsink will be bigger (sink and air-space) than a resistor dropping the same Watts because a resistor can be run much hotter than a Silicon device, needs less surface and air to carry the same Watts off.

The plan will work except it may pop the gate oxide at turn-on; as Merlin says, some diodes are often wise.

> how about the zener resistor R?

The gate draws "no" current.

The R-C product reduces buzz. Since big caps cost more and big-ohm resistors don't cost more, you want a large R.

The Zeners have a knee current. A 100V Zener may be 100V at 2mA, 99V at 1ma, 95V at 0.1mA, and just an uncertain 50V-80V at 0.01mA. Expect to run your Zeners at 1/10 to 1/2 of rated Power to keep them fat and happy and near nominal voltage. You will probably use 0.5 Watt Zeners and need 4 to 6 to get to 280V. That gives 1mA to 5mA. Your "340V" may really go 300V to 380V on different days. 300V-280V is 20V, at 1mA gives 20K. 380V-280V= 100V, in 20K gives 5mA. With these assumptions, a number like 20K seems good.

20K and 0.1uFd (100nFd) gives a low-pass at 80Hz. Assuming 50Hz power, 100Hz ripple, this is hardly any ripple rejection. True, the Zeners will clamp ripple.

And after all that, _I_ would still put a big resistor in series with the MOSFET. Shorts happen. The 33uFd on the output is "a short" at turn-on. The MOSFET will try to pass infinite current, and can fail "for no reason". With 100 ohms 10W in series, the maximum current is like 340V/100 or 3.4 Amps, which is short-term safe for any large MOSFET. And a long-term short is putting 1156 Watts in the 10W resistor, it will smoke quickly and leave a good clue for trouble-shooting. There is also transformer resistance so the maximum current is less. Maybe transformer resistance is ample to protect against short-term shorts, but I don't know what iron you have.

[km-r]

> i think running that HOT will shorten the tube life.

Nothing lives forever.

EL84 is very inexpensive to replace when/if it does cook to death.

EL84 at 13W Pdiss can last a VERY long time.

RCA ran 6BQ7/EL84 at 17W Pdiss in their stereo tuner/amps.

Since the amp is SE, drop about 10% in a resistor and add a second filter-cap to get cleaner B+. 250 or 500 ohms 10 Watts and another 40uFd-100uFd.

In my personal opinion..... regulation is rarely good for tube audio amps. It's more stuff to buy and wire and fail. GOOD regulation has to start with MUCH more voltage, or else it will fall out of regulation when wall-voltage sags. That means it works and sounds different between evening (low line) and midnight (high line).

I guess 280V from 340V meets my idea of "good regulation" overhead. There's 5 Watts wasted in the MOSFET (same as a resistor dropping the same voltage); more at high line voltage. You won't dissipate 5 Watts on a 99 cent heat-sink, it needs thought and space and air. The heatsink will be bigger (sink and air-space) than a resistor dropping the same Watts because a resistor can be run much hotter than a Silicon device, needs less surface and air to carry the same Watts off.

The plan will work except it may pop the gate oxide at turn-on; as Merlin says, some diodes are often wise.

> how about the zener resistor R?

The gate draws "no" current.

The R-C product reduces buzz. Since big caps cost more and big-ohm resistors don't cost more, you want a large R.

The Zeners have a knee current. A 100V Zener may be 100V at 2mA, 99V at 1ma, 95V at 0.1mA, and just an uncertain 50V-80V at 0.01mA. Expect to run your Zeners at 1/10 to 1/2 of rated Power to keep them fat and happy and near nominal voltage. You will probably use 0.5 Watt Zeners and need 4 to 6 to get to 280V. That gives 1mA to 5mA. Your "340V" may really go 300V to 380V on different days. 300V-280V is 20V, at 1mA gives 20K. 380V-280V= 100V, in 20K gives 5mA. With these assumptions, a number like 20K seems good.

20K and 0.1uFd (100nFd) gives a low-pass at 80Hz. Assuming 50Hz power, 100Hz ripple, this is hardly any ripple rejection. True, the Zeners will clamp ripple.

And after all that, _I_ would still put a big resistor in series with the MOSFET. Shorts happen. The 33uFd on the output is "a short" at turn-on. The MOSFET will try to pass infinite current, and can fail "for no reason". With 100 ohms 10W in series, the maximum current is like 340V/100 or 3.4 Amps, which is short-term safe for any large MOSFET. And a long-term short is putting 1156 Watts in the 10W resistor, it will smoke quickly and leave a good clue for trouble-shooting. There is also transformer resistance so the maximum current is less. Maybe transformer resistance is ample to protect against short-term shorts, but I don't know what iron you have.

i love the math but oh you did it for me hahaha thanks man! ive looked into some zener knee currents like they need at least 0.25mA?? but then more should be better, like two-four times? 18K-20K zener R. okay so a gate-source 12v zener and a drain series resistor 100ohm 10w...

wow.