This was a suggested dual-rail supply design from another forum. I was thinking of trying this but would like some comments on it. It's intended use is powering an audio/speech amp with +/- 280vdc regul. rails and 40-50mA load on each rail.
Note that the regulator circuits are identical and the design 'stacks' them, (rather than the familiar split supply with pos. and neg, linear regulator setup such as LM317/LM337). I see two things that don't make sense:
First, in the top image note how the designer implements the splitting and the zero voltage reference. The positive rail has a 'normal' ground return to its reservoir input cap. The negative rail however does not - the return path looks to be through the MOSFET device.
Second, how do you implement a single reservoir ground? The bottom image looks like the 'normal' way - but it too appears to have asymmetrical return paths.
Note that the regulator circuits are identical and the design 'stacks' them, (rather than the familiar split supply with pos. and neg, linear regulator setup such as LM317/LM337). I see two things that don't make sense:
First, in the top image note how the designer implements the splitting and the zero voltage reference. The positive rail has a 'normal' ground return to its reservoir input cap. The negative rail however does not - the return path looks to be through the MOSFET device.
Second, how do you implement a single reservoir ground? The bottom image looks like the 'normal' way - but it too appears to have asymmetrical return paths.
The lower circuit gives a ~-322V unregulated output and the lower pass transistor is not used, feeds nothing. The reason two windings and two rectifiers are required is that both sides use a positive regulator. You can share one rectifier and winding if you build a negative regulator, which requires either a P-channel MOSFET or a CFP (etc) pass circuit. This circuit uses about 6mA bias for a 50mA output, ie poor efficiency. That and the dual winding+ rectifier make this circuit just *** crude and simple. The P-channel MOSFET FQP3P50 is not rated as high as this N-channel, but more than enough for 322V and 50mA, ie 500V and 2700mA.
Thanks. I redrew it with a common center tap (and therefore obvious reservoir ground) and symmetrical N and P channel devices. I am not sure about the zener polarities in the negative side but can probably find it online. I would like to use complementary MOSFET packages. The FQP3P50 appears to be obsolete and lacks a complementary N-channel package. So just shop around for any 400 to 500v, 2.5 to 4A QFET will probably work fine here.
I targeted 5 to 6mA zener leg current because it keeps those devices in their linear region according to the spec sheet. Are you suggesting a different Zener leg current (and package) that is closer to the output current value?
I targeted 5 to 6mA zener leg current because it keeps those devices in their linear region according to the spec sheet. Are you suggesting a different Zener leg current (and package) that is closer to the output current value?
I think the zener on the negative should be the other way.
How much current should this supply deliver?
How much current should this supply deliver?
Yes, the zener 1N4744A in the P-MOSFET
Here are MOSFET which could be an idea to use:
NMOS IRFU310 and the PMOS IRFU9310
They are 400V and 1.1A devices.
The package is IPAK = TO-251
Here are MOSFET which could be an idea to use:
NMOS IRFU310 and the PMOS IRFU9310
They are 400V and 1.1A devices.
The package is IPAK = TO-251
The FET gates need very little current so I was thinking about 1mA would be more efficient. I assumed a <=1W Zener would be cheaper than 5W but maybe not. Marketing? Good catch on the P-channel gate protection diode. If I built something like this, I would add some kind of protection, but maybe it's not worth the complication. A ~220-470 Ohm resistor would take a bit of heat off the MOSFETs and limit the short current. I would be worried about using a TO251 package not easy to heat sink.
The heat sinking definitely needs to planned out in the first steps of board design and layout. In a couple previous builds with those packages, I bent the legs upward at 90 deg. and mount them underneath the board (legs soldered on top component side), and then a mechanical connection via SilPad to a suitable heat sink.
Two MOSFET for high voltage and TO-220 package:
NMOS IRF740 400V 6.3A 125W
PMOS IXTP10P50P 500V 10A 300W
NMOS IRF740 400V 6.3A 125W
PMOS IXTP10P50P 500V 10A 300W
Zener's are correct - one of the 4.7µF electrolytics is wrong though, and the 10nF high voltage caps are wrongly shown as polarized.
Top one will work provided they are fed from separate secondaries.
The lower regulator doesn't care whether it's negative or positive rail is connected to the top regulator's negative.
Attached is the same design using two +ve regulators
used by John Linsley Hood to power his Class A amp.
Last edited: