From: Robert Baer
X-Mailer: Mozilla 4.75 [en] (Win98; U)
Subject: Re: 87% All that vector calculus paid off
Date: Sun, 22 Dec 2002 06:54:05 GMT
NNTP-Posting-Date: Sat, 21 Dec 2002 22:54:05 PST
Organization: EarthLink Inc. -- http://www.EarthLink.net
Chris Carlen wrote:
> I built my first new flyback power supply prototype using my new
> Ferroxcube 19/8/5 E cores in 3C90, and a UC3525A PWM chip.
> I have yet to run it closed loop, so I'm just manually setting the duty
> cycle right now. This chip sure produces better gate rise/fall times
> than my useless function generator.
> I had my wife wind 32 turns of 26 ga. on the spool last night, and she
> did a real nice job for her first inductor. This morning I soldered the
> leads onto the bobbin pins, and gapped the core with several layers of
> packing tape, measured to exactly 0.18+/-0.01mm with a digital caliper.
> So that gives an effective gap of 0.36mm. That should produce an
> Al=100 and ue=140. So I figured I can sustain 1.14A before exceeding
> 160mT, and with switching around 180kHz, I'd need 47V input to pump 12W
> through the thing, which is my goal.
> Later I'll have to re-design the inductor to get the thing to produce
> the power throughput I want with only 11-25V input. But for now I just
> wanted to put the 3C90 through some tests, and see if my gapping would
> yield results close to what is predicted.
> At the moment the thing is driving a 10k resistor with 351.5V, pulling
> 0.315A from a 45.1V supply, for 87.0% efficiency. I'm running at 150kHz
> with 40% duty. Not a bad efficiency for a flyback boost converter on
> a breadboard, huh? Sure better than the results that stupid toroid
> inductor gave. I'll update my www with this design once the whole thing
> is done.
> The core, coil, and FET are very slightly warm to the touch. The whole
> thing is built on a breadboard with moderate attention paid to
> minimizing parasitics, and providing ample bypassing reservoirs.
> Oh, my analog scope shows the drain spiking up to the 600-700V range
> before the HER105 diode turns on. Is this bad? The inductor only has
> about 65uJ per cycle, and the IRF730a is able to handle repetitive
> avalanche energy of 7.4mJ, so I can't imagine that the thing will break
> under these conditions. But I don't know much about this avalanche
> stuff. I'd appreciate some lectures on the subject. Should I try a
> faster diode?
> I will try closing the loop next, and then think about what additional
> features will be needed in order to make the supply usable as an
> adjustable output voltage supply, with some open circuit and short
> Oh, I'll probably experiment with the TL3842 current mode controller
> before I settle on a final design to commit to PCB. That might be a
> good thing for making a supply with short circuit protection.
> Good day.
> Christopher R. Carlen
> Suse 8.1 Linux 2.4.19
It might be a good idea to check the resonant frequency of the
inductors on your transformer.
Scatter winding will raise that considerably; "pi" winding is ideal
I choose a FET so that it will not avalanche, or modify the windings
so there is no problem.