updated 2001-02-11

Note that this is nowhere near complete.

Switching power converters


Data I've been collecting about so-called "switching power supplies" (technically "power converters"). There are 2 kinds: DC-DC converters and AC-DC converters. I've been thinking about using these commercial off-the-shelf chips in non-standard fashion for magnetic levitation for years ... I know I mentioned in in my Senior Design 1 report ....

Possibly related files:


theory: pages that decribe how switching power works and how to design your own switching power devices. Design tools.

power converter modules

power converter components (ICs, inductors, FETs, etc)

National Semiconductor's Simple Switcher series

"If you want to design a DC-DC unit yourself from discrete components, then MAXIM, National Semiconductors, and Linear technology have various DC-DC converter chips, all requiring at least one inductor, a MOSFET, a diode, and a few resistors/capacitors. If I remember correctly both NSC and Linear Technology deliver a program designing the power supply for you, given the required specifications."

International Rectifier

International Rectifier http://irf.com/ sells some interesting high-power, high-current, high-voltage components.


Magic Sinewave and other PWM

From: lis@ict.pwr.wroc.pl (Jaroslaw Lis)
Newsgroups: comp.robotics.misc,sci.electronics.design
Subject: Re: Magic Sinewave Motor Control?
Date: Fri, 18 Apr 1997 20:44:51 GMT
Organization: Technical University of Wroclaw, Poland

On 17 Apr 1997 16:25:19 GMT, albaugh@agames.com (Mike Albaugh) wrote:
>Chris Baron (cbaron@nwrain.com) wrote:
>: I was reading Don Lancaster's site yesterday ( http://www.tinaja.com )
>: about the so-called magic sinewave technique.  This method uses a long
>: string of bits to synthesize a sine wave at very high efficiency.

>	Haven't seen it, but this looks like an opening to mention
>a nifty technique that is probably related and that I have definitely
>used successfully. I saw it in an electronics magazine in the early
>'70s, so even if it was patented, it is probably safe to use by now.
>	The idea is to take shift-register and make a pseudo-random
>number generator out of it. The technique is similar to a CRC, [...]
>	Now, "compare" the value of rng to the (7-bit) value you
>wish to represent. If the value is >=, drive the output high,
>if it's lower, drive the output low. [...]

>	"So what?", you say. "I could have done that with a
>basic ramp-type PWM". Well, yes, but with a ramp, all the highs
>come together, as do all the lows, then repeat. There is a _strong_
>"noise" component at you repetion rate. In this scheme, the largest
>noise component is at the shift-register clock rate, which should be
>well above the frequency of interest, [....]

Is this LSFR method much different in output, than an old trick:
use counter and comparator, but cross counter lines - exchange
LSB with MSB, bit 1 with bit n-1, etc.
This way 8 bit counter output 0, 128, 64, 192, 32, 160 ...
result seems to be similar.

If you do PWM in software, I'd suggest:

     if (state<0)  output 0;
       { state -= range;
         output 1;

performed for each step. This is guaranteed to give highest frequency
value, and most equal space between impulses.

>This makes it easy to filter. Looked at with a scope, the trace resulting
>from sweeping a sine table looks like a broader trace, as compared to
>what you normally get with a conventional DAC, which has obvious (to
>the eye _and_ ear :-) steps. I was using this to do "all digital" sounds,
>but it would be equally applicable to Motor control, with some caveats:
>1) Notice I used an example of 7 bits. 8-bits is a little nasty
>   because it requires 4 taps. Not too much of an issue in hardware,
>   but might be in software. There is a 9-bit LFSR that only uses
>   two taps, but then the period (natch) is 511, so the noise may
>   creep down too much. In general, you trade clock-rate/period/precision
>   off against each other.
>2) As with any PWM technique, but _especially_ here, you need to make
>   sure the switching components can handle the frequency. The very
>   thing that makes this technique nice (high switching frequency
>   compared to sample rate) makes for higher dissipation in the
>   switches.
>					Mike
>| albaugh@agames.com, speaking only for myself

Eos Corporation http://www.eoscorp.com/ sells "power systems" including a 3"x5"x1", AC input, 100 Watt DC output.

Deltron Inc. http://www.deltronic.com/

the Unitrode UCC2805, a single-chip biCMOS PWM controller that contains all of the necessary circuitry (voltage reference, error amplifier, comparator, MOSFET gate drive, and oscillator) for closed-loop PWM power-supply control. ... can be configured for either voltage-mode or current-mode feedback ... http://www.penton.com/ed/Pages/magpages/oct2298/ifd/1022id.htm

Switching-Mode Power Supply Design by Jerrold Foutz http://www.smpstech.com/

some detailed information on switchmode power supplies http://www.repairfaq.org/REPAIR/F_Repair.html

Darnell Group Darnell.Com/PowerPulse http://www.darnell.com/ "The power electronic industry's premiere web site" "technical articles contributed by industry experts - updated weekly" " Darnell Group specializes in the economic/business analysis of emerging power electronics markets and technologies. ... We provide primary research, news services and original, proprietary information. " UPS inverters, dc/dc converters, Motion Control, Batteries and Portable Power, Switch-Mode Power Conversion, Power Quality Protection.

Valhalla Scientific http://millennianet.com/valhalla/ sells high-voltage, high-power, voltmeters, current analyzers, wattmeters, micro-ohmmeters, etc. Lots of good information on "Fundamentals of Power Measurement", motor/generator testing, "Low Resistance Measurement" (online "resistivity" measurement table of all the common metals, which can be used to adjust the online "copper wire resistance" table for wires of other metals. )

"Unlike conventional ballasts which alternate at 60 cycles per second, electronic ballasts operate at over 20,000 cycles per second. This higher frequency increases flourescent lamp efficacy by 10-15% and completely eliminates hum and flicker. Electronic ballasts also waste less power internally." -- unknown more on electronic flourescent list ballast: http://www.lrc.rpi.edu/NLPIP/Answers/LA-TF5/T5FT_1.htm

Vicor Corp. http://vicr.com/ sells DC-DC converters

$0.85 (in 10 000 s) MC33370 through MC33374 100 KHz switching regulators operating from universal line sources of 85 to 275 VAC; can be turned on or off via secondary-side MCU signal; programmable on or off state upon application of AC power; Motorola SPS http://mot-sps.com/scg

Foresight Electronics http://www.fse-power.com/ is apparently (?) the authorized distributor (?) of Cosel switching power supplies.

Newport Components Limited http://www.newport-comps.com/

International Power Devices Inc. http://www.ipdconverters.com/ sells lots of DC/DC converters

Golden Pacific Electronics http://www.gpelectronics.com/ sells lots of "wall worts" and other low-cost switching power supplies.

Melcher http://www.melcher-power.com/ AC-DC and DC-DC converters.

$4.50 (in 1 000 s) MIC217x tiny surface mount synchronous buck regulator (PWM controller) with integrated high-side MOSFET and low-side MOSFET (2.5 A) Micrel Semiconductor, Inc http://www.micrel.com/217x.html

Linear Technology http://www.linear-tech.com/go/noinductors has some very small (SO-8 and SOT-23) surface mount DC/DC converters. boost, buck, buck-boost, with frequencies up to 900 KHz.

From: James Meyer <notjimbob@worldnet.att.net>
Subject: Re: DC_DC convertor
Date: 14 Oct 1999 00:00:00 GMT
Newsgroups: sci.electronics.design

On Thu, 14 Oct 1999 11:46:42 +0400, Vlad  wrote:

>Can anybody answer what IC I can use in 12V DC to 4.5V DC 700mA converter
>design instead of MAX1771( not available here now)

        The "common as dirt" National LM117 series of 3-terminal regulators can
be used as linear or switching regulators.  The switching circuit requires the
addition of a single transistor and inductor to the basic design.

        Get the datasheet for the LM117 and look at the typical applications
section for schematics.


From: Robert <romapa@earthlink.net>
Subject: Re: DC_DC convertor
Date: 14 Oct 1999 00:00:00 GMT
Newsgroups: sci.electronics.design

Whoa there! These are Schmitt regulators and do not work well with wide load
variations- frequency increases with load and transient response stinks. Also this
type requires a minimum of output ripple voltage [ at least 5mV ] to work properly.
The LM317 configurations in the application notes do not switch the power PNPs very
well so you will have lots of switching losses.If someone is going to go to all
this trouble then they would be foolish to overlook the National LM2575 1A buck
regulators. For this application the poster will want either the LM2575-5.0 or
LM2575-ADJ [ if he has to have 4.5 Volts]. These are available from DigiKey in
several grades for about US$5.00.

Hoping This Helps,-Robert


started 1998-Jan-15

end http://rdrop.com/~cary/html/switching_power.html