Switch-mode B+ Supply


Occasionally, there is need for a high voltage supply for various electronics projects. A couple of my projects which fall into this category are:
Three Tube Superhet Receiver
Nixie Display

As with any high voltage circuit, a word of caution is in order. These circuits can generate high voltages at sufficient current levels to be deadly. Proceed accordingly, and at your own risk.

A small switch-mode supply, such as presented here, is not capable of delivering as much current as a power supply operated directly from the power mains. However, it can still deliver a serious or even deadly shock. Don’t be misled by its small size!

Here is the schematic for the circuit I developed:

It is a standard boost converter circuit.


Parts values are as indicated. The ICM7555 is a CMOS version of a dual 555 timer. This was chosen over switch-mode regulator ICs, because it’s readily available in a standard DIP package, making it easy to mount on the circuit board.

The IRF730 is a high voltage MOSFET. Other types with similar voltage ratings may be substituted, as long as the turn-on gate voltage is suitably low (5 volt TTL level, or less), and the breakdown voltage is suitably high. The IRF730 was the most inexpensive one that I could find at the time of construction.

The UF4007 diode should not be confused with a 1N4007 diode. These parts are not interchangeable. The UF4007 is an ultrafast rectifier which is required in a high frequency switching power supply.

The inductor may be either a Coiltronics DR127-221-R, or an AlfaMag SWF-1.4-220. Other types will be suitable as well, but the circuit has been tested with these two inductors and they work well, giving good efficiency.


Output voltage is adjustable from below 50 volts to over 400 volts. However, if operation above 200 volts is required, then the voltage rating of C1 must be increased accordingly. The supply can deliver over 25 mA at 90 volts. This was the primary design requirement, because one application is as a power supply for a vintage Zenith Transoceanic receiver which has this power requirement. At this output load, the overall circuit efficiency is 82%.

I have not yet done any exhaustive testing of the power supply to find its maximum voltage/current limitations. In testing so far, neither the MOSFET nor the inductor has shown any detectable temperature increase. The MOSFET is supplied in a TO-220 package suitable for heatsink mounting, but a heatsink is not required in this circuit.

Circuit Notes

The operation of a boost converter circuit is fairly straightforward. A detailed explanation is given on Wikipedia. However, I will provide a more brief explanation here.

The timer turns on the MOSFET, allowing current to flow from the power input, through the 220µH inductor, and through the MOSFET to ground. Because of the inductance, current flow cannot change instantly, and so it starts at zero and then increases steadily. If the MOSFET were to remain on indefinitely, the current would reach a level sufficiently high to burn something out. However, the MOSFET is switched off before that happens. Because the current through the inductor cannot stop instantly, the only path is through diode and the electrolytic capacitor C1. The voltage at the inductor will increase to whatever value is necessary in order to maintain the current flow. This is the principle that causes the voltage across the capacitor to exceed the circuit input voltage. The output voltage is regulated by a feedback circuit consisting of the resistor voltage divider network across C1, and the NPN transistor. This controls the charging current, and hence the time constant, in one of the timer sections, resulting in a variable pulse width signal to control MOSFET on-time. The second timer in the ICM7555 runs at a constant frequency to control the MOSFET switching frequency. The result is a constant frequency pulse width modulation (PWM) signal that maintains the output voltage over a wide range of output load conditions and varying input voltages.

This circuit operates at approximately 24 kHz. This is quite low for switch-mode power supplies which typically operate at hundreds of kHz. However, this frequency was chosen because it is high enough to be above audio range, high enough to use reasonably small value inductors, but at the same time low enough that it won’t generate massive amounts of RFI (Radio Frequency Interference). The low noise requirement is important for radio applications. Even at a switching frequency of 24 kHz, this supply still produces significant harmonics right up to shortwave frequencies. Therefore, RFI mitigation, in the form of shielding and filtering is still required.


This circuit is quite forgiving of construction methods, but best performance is obtained when built on a printed circuit board, following good high frequency design practices. The prototype is shown in the photo to the right. Circuit board dimensions are 1.75” x 1.375” (45mm x 35mm). The 220 µH inductor is the small square black part near the top of the board partially hidden by the MOSFET. It is a Coiltronics DR127-221-R. This is a shielded ferrite core inductor which minimizes radiated RFI. The inductor is intended for surface mounting, and may be installed on the foil side of the board if desired. However, I simply soldered a couple of short bare leads through the board, and then soldered the inductor onto these on the component side. The circuit board foil pattern and component mounting information are available in this zip file. The foil pattern is mirror image, suitable for use with the toner transfer method, of printed circuit construction. The circuit board includes no output filtering to remove RFI, and therefore an external filter may be required depending on the application. RFI is discussed in the following section.

RFI Mitigation

The MOSFET switches extremely quickly, which results in high operating efficiency. When fully off, the MOSFET consumes no power. When fully on, the power dissipated in the MOSFET is i2RDSon. The value of RDSon is extremely low in these types of MOSFETs, resulting in very low power losses when on. However, significant power dissipation can occur if the device switches slowly. During the switching period, both the voltage and current are significant, and lead to significant power dissipation and heating of the device. For this reason, the devices are designed for the fastest possible switching speed. Unfortunately, the extreme switching speed causes switching transients with extremely fast rise times, resulting in the generation of significant amounts of RFI.

To eliminate or minimize RFI, the first step is to install the power supply in a shielded enclosure. This prevents direct radiation of RFI. For initial testing, I mounted the power supply circuit board inside a small tin can. The power leads pass through a ferrite sleeve to reduce conducted RFI. With just this level of RFI abatement, I decided to test it on my Three Tube Superhet. On the medium wave band, performance was better than expected. However, there were several frequencies across the band where the amount of generated interference was unacceptable. I then tested it on the shortwave band from 5000 to 6000 kHz. Surprisingly, the RFI was far worse at these higher frequencies, and most of the received radio signals on this band were obliterated. The reason became apparent when I connected an oscilloscope to the power supply output while under load. There was noticeable ringing on the waveform whenever the MOSFET switched. The ringing frequency appeared to be at approximately 5000 kHz, right in the band where the radio was operating.

The next step was to build a low pass output filter to remove this high frequency noise. After trying of couple of different filter configurations, I obtained the best results with a differential mode choke consisting of a 25 mm diameter by 12 mm long ferrite toroid, with two 9 turn windings of #24 AWG magnet wire. The ferrite is of the type designed for RFI filtering, and has a relative permeability, µR, of about 6500. This results in an inductance of about 640 µH per winding. Shunt capacitors (0.047 µF) were placed across both input and output sides of the choke.

This resulted in the elimination of the high frequency ringing, and reduced the sharp switching transient to little more than a very low level ripple.

It should be noted that both the positive and negative high voltage leads pass through the filter and therefore the negative lead is not common to the negative input of the low voltage power input. My primary concern at this stage was to determine whether the HV output could be made sufficiently noise free to operate a radio receiver. So, I had no immediate concerns about the isolation of the negative leads. The power supply was powered from a six volt lantern battery which had its negative side isolated from the HV negative lead. Given the encouraging results of these tests, the next step is to design a more advanced filter that will allow for a common ground point for both the high and low voltage negative leads.

Since one of the applications of this converter circuit was the B+ supply of a battery operated radio, it would be convenient to have it switch on and off automatically according to whether the tube filaments are drawing power. To that end I designed a current switch circuit that can be found here.

More to follow. Stay tuned...

Back to:



This page last updated: May 29, 2016

Copyright 2014, 2015, Robert Weaver