[ARC5] Martin Ossman DC/DC Power Supply - very nice!

[Ben Hall]

Afternoon all,

I have built the power supply described by Martin Ossman in the 
March/April 2015 edition of Elektor Magazine.  After correcting a minor 
error in my PCB layout concerning the RT and CT components on the TL494 
switching controller chip, its working nicely.

At no load and 16 VDC input, it's drawing 50 mA.  Output no load is 
219.3 VDC with 1 Vp-p ripple.  Unlike the Chinese DC-DC switcher which 
showed high frequency noise on top of distorted sine wave, the ripple 
here is purely impulse noise as one would expect from the output of a 
switching supply.

Switching frequency is approximately 166.7 kHz.

On my load bank (resistive), I took the following measurements:

219.3 VDC at 0 mA load (500k of bleeder) with ~1 Vp-p ripple
197.8 VDC at 6 mA load with ~1 Vp-p ripple
196.9 VDC at 11 mA load with ~1 Vp-p ripple
195.8 VDC at 25 mA load with ~1.5 Vp-p ripple
194.7 VDC at 38 mA load with ~2 Vp-p ripple (drawing 0.6 amps at 16 VDC)
193.7 VDC at 42 mA load with ~2.1 Vp-p ripple (drawing 0.7 amps at 16 VDC)
192.9 VDC at 50 mA load with ~2.2 Vp-p ripple (drawing 0.8 amps at 16 VDC)

192.9 * 0.050 = 9.6 watts, 16 * 0.8 = 12.8 watts

This is an efficiency of ~75%.

At 24VDC input, it put out 296 VDC no-load.  Because this supply doesn't 
have any regulation, the output voltage is a function of the input 
voltage.  This was by design, as the author was building an adapter for 
his 0 to 30 VDC lab supply to put out 0 to 350 VDC.

After 10 mins at 24 VDC input, with 285 VDC output at 118 mA, I checked 
temperature of all the components.  The warmest thing on the unit was 
the 7812 regulator that powers the low-voltage electronics!  The two 
IRF540's were cool, the output MUR1560 rectifiers were cool, and the 
ferrite transformer was just slightly above room temperature.

Bottom line, I'm pretty darn impressed.  Runs cool, regulation really 
isn't that bad for something that's totally unregulated (and would get 
better from no load to the 6 mA point with lower value bleeder 
resistors), and its got a floating output so using it with something 
like a WS19 or BC348 should be possible.

Being that the design is a good one in my opinion, below is my schematic 
and a photo of my assembled board.

I don't want to revisit the copyright discussions, but I'll say that I'm 
pretty sure I'm safe if I post my interpretation of the schematic vice 
the original magazine article.  :)

The sole difference between my schematic and the article is that in the 
article, there was a SPDT switch on the secondary of the transformer. 
That way one could switch between just the lower half of the output 
transformer yielding half the voltage, and both halves in series 
yielding full output voltage.  In my prototype, while I have the 
provision for the switch, it's jumpered for full output.

<http://www.kd5byb.net/DCDConverter/Ossman_DCDC.pdf>

<http://www.kd5byb.net/DCDConverter/Ossman_DCDC.jpg>

In the photo, you see a pair of 0.47uF capacitors tacked across the 
bleeders.  This was an experiment to see if I could reduce the output 
spikes any by adding them.  The answer is no, they made no difference.

thanks much and 73,
ben, kd5byb