Post by thread:[OT] POWER SUPPLIES

I bought one of the little Power Trends switching supplies from DigiKey. It takes voltage around the range you say and provides 5V @3A. It is very small and requires a 100uF external cap. It boasts a 90percent efficiency. At $21.00 per part, they aren't the cheapest. I bought mine for a robot power supply. After playing around with the thing, I may buy another just to hawe banging around in the shop. Its really cute! ok..jef On Sun, 23 Apr 2000, Brian Kraut wrote: {Quote hidden}> Does anyone have a suggestion for a low cost low parts count 5V DC 1A > power supply that will work with a 10-40V input. I am looking to > replace 7805 regulators in some designs I have that are in very small > enclosures and don't have room for a heat sink. A 7805 without a heat > sink in a small case only driving 100mA will burn up in no time with a > 30V input because it has to dissipate 2.5W of heat. > > I am looking for a switching type regulator that costs no more than a > few bucks and takes minimal board space. Preferably no external parts > other than an inductor and a cap. > Jeffrey D. Spears University of Michigan College of Engineering ``Double-E, can't spell gEEk without it!'' -Captain Gerald M. Bloomfield II, USMC (my brother)

I know exactly what part you are referring to. They are wonderfull except for the price. They cost as much as the rest of the components, the circuit board, and the enclosure. Jeffrey D Spears wrote: {Quote hidden}> I bought one of the little Power Trends switching supplies from > DigiKey. It takes voltage around the range you say and provides > 5V @3A. It is very small and requires a 100uF external cap. It boasts > a 90percent efficiency. At $21.00 per part, they aren't the cheapest. > I bought mine for a robot power supply. After playing around with > the thing, I may buy another just to hawe banging around in the > shop. Its really cute! > > ok..jef > > On Sun, 23 Apr 2000, Brian Kraut wrote: > > > Does anyone have a suggestion for a low cost low parts count 5V DC 1A > > power supply that will work with a 10-40V input. I am looking to > > replace 7805 regulators in some designs I have that are in very small > > enclosures and don't have room for a heat sink. A 7805 without a heat > > sink in a small case only driving 100mA will burn up in no time with a > > 30V input because it has to dissipate 2.5W of heat. > > > > I am looking for a switching type regulator that costs no more than a > > few bucks and takes minimal board space. Preferably no external parts > > other than an inductor and a cap. > > > > Jeffrey D. Spears > University of Michigan > College of Engineering > > ``Double-E, can't spell gEEk without it!'' > -Captain Gerald M. Bloomfield II, USMC > (my brother)

LM2575T-5 Switch mode 5pin TO220 package 7V-40V input and 5V @1A Output is about £1.80 10off LM2576-5 Switch mode 5pin TO220 package 7-40v input and 5V @3A Output is about £2.10 10off These devices need only two capacitors and one Zener diode. Regards, Jay At 11:25 23/04/00 -0400, you wrote: {Quote hidden}>Does anyone have a suggestion for a low cost low parts count 5V DC 1A >power supply that will work with a 10-40V input. I am looking to >replace 7805 regulators in some designs I have that are in very small >enclosures and don't have room for a heat sink. A 7805 without a heat >sink in a small case only driving 100mA will burn up in no time with a >30V input because it has to dissipate 2.5W of heat. > >I am looking for a switching type regulator that costs no more than a >few bucks and takes minimal board space. Preferably no external parts >other than an inductor and a cap. > > >

Sounds like it may be just the ticket. I will check it out. Thanks. "Jay.R.Vijay-Indra" wrote: {Quote hidden}> LM2575T-5 Switch mode 5pin TO220 package 7V-40V input and 5V @1A Output is > about £1.80 10off > > LM2576-5 Switch mode 5pin TO220 package 7-40v input and 5V @3A Output is > about £2.10 10off > > These devices need only two capacitors and one Zener diode. > > Regards, > > Jay > > At 11:25 23/04/00 -0400, you wrote: > >Does anyone have a suggestion for a low cost low parts count 5V DC 1A > >power supply that will work with a 10-40V input. I am looking to > >replace 7805 regulators in some designs I have that are in very small > >enclosures and don't have room for a heat sink. A 7805 without a heat > >sink in a small case only driving 100mA will burn up in no time with a > >30V input because it has to dissipate 2.5W of heat. > > > >I am looking for a switching type regulator that costs no more than a > >few bucks and takes minimal board space. Preferably no external parts > >other than an inductor and a cap. > > > > > >

>> LM2575T-5 Switch mode 5pin TO220 package 7V-40V input and 5V @1A Output is >> about #1.80 10off >> >> LM2576-5 Switch mode 5pin TO220 package 7-40v input and 5V @3A Output is >> about #2.10 10off >> ---------- >> >I am looking for a switching type regulator that costs no more than a >> >few bucks and takes minimal board space. Preferably no external parts >> >other than an inductor and a cap. > >You also might look into Nationals newer stuff like an LM2675N-5 --260kHz so >smaller externals. They also have have some software to calculate component >values for it. You can source them through http://www.pioneerstandard.com (around >$2.66 usd ea.) as it seems digikey doesn't have them yet. > ---------- This thread has been a very good learning experience - [for one unversed in buck regulators]. I downloaded the datasheets from Nat'l, and summary follows: LM257x: $5.15-7.28, qty 1, Digikey. Cap types: requires large values (100-2000 uF), high voltage on input (50+), *very* low ESR (< 0.25 ohms) and ESL, plus very high ripple current & ripple voltage capablility. Inductors: must be designed to confine magnetic fields and not saturate at high load currents, else serious EMI and lack of regulation result, expensive $2-7 from Digikey. Layout: somewhat critical, to confine magnetic fields and minimize lead inductances. Output ripple: serious, may require post-filtering - another inductor plus 100 uF cap. [I assume ripple increases with Iload]. All in all, it looks as though a 1-off design would cost $10+ in parts, would probably have a substantial ripple problem, and would probably generate enough EMI to make its use with analog circuitry questionable. Also, due to the stiff requirements on the caps [ie, very low ESR and high ripple voltage/current capable], I wonder about the long-term reliability of the design in general. Also, given that large caps/volts are required, the ckt would probably not be all that small. All in all, this is a beautiful concept, but you pay a real price, in cost and size, to increase efficiency from < 40% [ie, linear 7805 type] to 70-90%. Heat-wise, the buck regulator certainly rules. Could anyone with practical real-world experience on these issues make comment? best regards, - Dan Michaels Oricom Technologies ===================

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 >Cap types: requires large values (100-2000 uF), high voltage on input > (50+), *very* low ESR (< 0.25 ohms) and ESL, plus very high > ripple current & ripple voltage capablility. The voltage is strictly a function of what your input supply will do. The requirement would be the same switcher or not. >Inductors: must be designed to confine magnetic fields and not saturate > at high load currents, else serious EMI and lack of regulation > result, expensive $2-7 from Digikey. Proper selection helps a lot. Take your maximum current, and find inductors that won't saturate there. Then select for L value, and series resistance (a loss term) Toroidal is reccomended, but pot-core is fine too. >Layout: somewhat critical, to confine magnetic fields and minimize > lead inductances. >Output ripple: serious, may require post-filtering - another inductor > plus 100 uF cap. [I assume ripple increases with Iload]. > >All in all, it looks as though a 1-off design would cost $10+ in parts, >would probably have a substantial ripple problem, and would probably >generate enough EMI to make its use with analog circuitry questionable. The "simple switchers" are anything but, for the reasons you state here. Their relatively low frequency makes inductor selection problematic. Look into something that can run at around 200-300 kHz. Life gets better. You have a bit more loss in the diode and switch, but it's worth it imho. Layout in a switcher consists of some critical tracks, and the rest are just tracks. If you do it right, they are quiet. EMI? That's a problem I've never seen. My conducted line scans have always been "flatliners" and radiated emissions has never been even enough to notice. I've done one board, 3 x 6 inches that had a 5W switcher, plus a current mode stepper controller at 700mA (another pair of switchers, effectively) and a processor, and an amplifier for a mag head, located very close to the stepper, with about a 1uV output (from the head) The only source of "noise" was inductive coupling between the motor and the head. Power supply noise was measurable, but not on the op-amp's outputs. The part 15 scan on this one was on the order of "Are you SURE it's on?". Output ripple is very controllable. Simple layout tricks can get you an additional 6dB or more, just by making the current go where you want it... Example: Output side diode, cap, and output point.. Route the track from the diode to the cap with maybe 100 mil, and return the cap to the main ground point the same way. Then route another track to the output. (rather than taking the output off the diode and hanging the cap off on a branch, or making them all one large "blob") >Also, due to the stiff requirements on the caps [ie, very low ESR and >high ripple voltage/current capable], I wonder about the long-term >reliability of the design in general. Also, given that large caps/volts >are required, the ckt would probably not be all that small. Check Digi-Key. Low ESR caps aren't that bad. Check the specs, don't automatically run for the "super-duper-low ESR" listings. Today's average cap is the low ESR cap of 10 years ago. You want caps rated at or below your calculated ripple current, at 2000 hrs and 105c. There's a complex equation for lifetime, but if you double the hours for every 10C decrease, you'll be pretty close to right. - -- Are you an ISP? Tired of spam? http://www.spamwhack.com A pre-emptive strike against spam! Where's Dave? http://www.findu.com/cgi-bin/find.cgi?kc6ete-9 -----BEGIN PGP SIGNATURE----- Version: PGPfreeware 6.5.2 for non-commercial use <http://www.pgp.com> iQA/AwUBOQSfhIFlGDz1l6VWEQKgsACfUBB81kPkzA/BgYbfuKtSNDctPjYAniw6 HtBu9wXpzOUDOwUXcm3e9EYt =HqER -----END PGP SIGNATURE-----

Dave Van Horn wrote: ..... >The voltage is strictly a function of what your input supply will do. The >requirement would be the same switcher or not. .... >The "simple switchers" are anything but, for the reasons you state here. >Their relatively low frequency makes inductor selection problematic. >Look into something that can run at around 200-300 kHz. Life gets better. >You have a bit more loss in the diode and switch, but it's worth it imho. ..... >Layout in a switcher consists of some critical tracks, and the rest are >just tracks. .... >If you do it right, they are quiet. EMI? That's a problem I've never seen. >My conducted line scans have always been "flatliners" and radiated ...... Hi dave, thanks for the reply. I am a little confused. Were your comments related to "switcher" supplies in general, or to the LM257x buck regulators [that I was commenting on] in particular? Many of the same considerations apply to both, I am sure. Your quiet "zero-EMI" device was a bucker, or an AC-powered switcher with coupling xfrmr/PWM generator chip[3842..], etc? I was mainly fishing for real-world input from people having used the buckers, like the LM257x. Thanks and regards, - Dan Michaels

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 >I am a little confused. Were your comments related to "switcher" >supplies in general, or to the LM257x buck regulators [that I was >commenting on] in particular? Many of the same considerations >apply to both, I am sure. Mostly generic. Try to avoid low freq switchers, and the NS "simple switchers" >Your quiet "zero-EMI" device was a bucker, or an AC-powered switcher >with coupling xfrmr/PWM generator chip[3842..], etc? 24VAC input, bridge rectified, into the motor control chips (PBL3717s) and the flyback switcher (3842 IIRC) with Zetex mosfet and custom inductor on an RM-6 core. + and - 5V output. >I was mainly fishing for real-world input from people having used >the buckers, like the LM257x. Done that too, Though not with the "simple switchers". An associate did, he had lots of noise problems, and eventually gave them up for linear tech devices. I've only used the one NS part as a buck, and I forget the part number.. An 8 pin dip, works acceptably well. I can dig that up if you like. - -- Are you an ISP? Tired of spam? http://www.spamwhack.com A pre-emptive strike against spam! Where's Dave? http://www.findu.com/cgi-bin/find.cgi?kc6ete-9 -----BEGIN PGP SIGNATURE----- Version: PGPfreeware 6.5.2 for non-commercial use <http://www.pgp.com> iQA/AwUBOQSxl4FlGDz1l6VWEQKekACgiUsShsdCrKHf9RdqNXixfD0Uw7UAoIG+ g91LqOxWFFM8Ze1tN+tcmABV =07Pv -----END PGP SIGNATURE-----

Dan Michaels wrote: [snip] > Could anyone with practical real-world experience on these issues > make comment? Low ESR means to have power to delivery in high current peaks, as the switching FET needs to load the biggest possible magnetic field at the coil it needs low impedance current source. If you are using NiCad cells, they help a lot because they have very low impedance. I use Max641, 586 or 587 at my devices (the first one has an internal catch diode). EMI? yes, but not enough to cause any problems at the 22.5 bits ADC installed less than 20 mm from the coil. Just for info, this ADC resolution is about 0.12uV (120 nano Volts), the interference should exist but it is not noticeable. About the coil, all switcher chips manufacturers supply formulas to calculate the current and coil size. Just use a 20 to 30% bigger coil and sleep well. I already made several, several tests with those switches, using more than 10 different coils from 10uH to 1000uH at the same circuit, same load, little changes in efficiency are noticeable, of course there is always the "best fit". There are units working below 100kHz and above that. Smaller components are required if running 600kHz and above, but then the first noise harmonics would be close to the microprocessor clock signal... :) I really don't think twice, not anymore. Portable units *will* have a switching supply to extend battery life, except if requested by the customer not to do it, but then it will be his problem to carry a pocket full of new batteries. Wagner

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 At 03:22 PM 4/24/00 -0400, Wagner Lipnharski wrote: >Dan Michaels wrote: >[snip] >> Could anyone with practical real-world experience on these issues >> make comment? > >Low ESR means to have power to delivery in high current peaks, as the >switching FET needs to load the biggest possible magnetic field at the >coil it needs low impedance current source. If you are using NiCad >cells, they help a lot because they have very low impedance. Low impedance batteries are some help, but you're putting all that fast current on the battery wires. BAD DOG! Put a small low ESR cap AT the fet, and a murata/panasonic three lead EMI filter in series. Now the battery lead current is practically DC, and the circuit gets what it needs as well. Actually, in these circuits, the low ESR cap is on the output. The input demand is a current that starts at zero, and ends up at some finite and not terribly high value. On the output stage, I have used a low ESR (Panasonic HFS) for the output charge well, and then a series 1 ohm into a 1000uF cap. The 1 ohm was as effective as any reasonable inductor, and cost a lot less in every way. (Obviously this is assuming a low current output) >There are units working below 100kHz and above that. Smaller components >are required if running 600kHz and above, but then the first noise >harmonics would be close to the microprocessor clock signal... :) And why would that matter? - -- Are you an ISP? Tired of spam? http://www.spamwhack.com A pre-emptive strike against spam! Where's Dave? http://www.findu.com/cgi-bin/find.cgi?kc6ete-9 -----BEGIN PGP SIGNATURE----- Version: PGPfreeware 6.5.2 for non-commercial use <http://www.pgp.com> iQA/AwUBOQS+XoFlGDz1l6VWEQIJRACeLdzhFrSesW60wqUKuKHaZBkHvXUAoIow SNPARlP6nUSMmyft3/hOGaWy =f7Jq -----END PGP SIGNATURE-----

Dave Van Horn wrote: > >>I am a little confused. Were your comments related to "switcher" >>supplies in general, or to the LM257x buck regulators [that I was >>commenting on] in particular? Many of the same considerations >>apply to both, I am sure. > >Mostly generic. Try to avoid low freq switchers, and the NS "simple switchers" .... .... >>I was mainly fishing for real-world input from people having used >>the buckers, like the LM257x. > >Done that too, Though not with the "simple switchers". An associate did, he >had lots of noise problems, and eventually gave them up for linear tech >devices. I've only used the one NS part as a buck, and I forget the part >number.. An 8 pin dip, works acceptably well. I can dig that up if you like. > Soooo, if I perceive this correctly, you are definitely warning off of the LM257x chips. They do, in fact, generate too much EMI. >From the datasheet, I clearly got the impression if the layout is non-optimal (leads too long, ie inductive, and traces too small), the cap ESL is too high, wrong diode is used, or the inductor is not properly chosen (so as to confine magnetic fields to the core), then EMI definitely will be a problem. Lots of criticality here. So, do you think your associate's troubles emanated from one of these selections, or is really from use of the LM257x, no matter what else you do? Also, which LT chip did you guys finally select? Thanks mucho grande, - Dan Michaels Oricom Technologies ===================

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 >So, do you think your associate's troubles emanated from one of >these selections, or is really from use of the LM257x, no matter >what else you do? For him, it was the SS chips. I don't know why exactly, maybe the long leads on the TO-220-ish packages? >Also, which LT chip did you guys finally select? Several, I don't remember any problems with LT parts at all. - -- Are you an ISP? Tired of spam? http://www.spamwhack.com A pre-emptive strike against spam! Where's Dave? http://www.findu.com/cgi-bin/find.cgi?kc6ete-9 -----BEGIN PGP SIGNATURE----- Version: PGPfreeware 6.5.2 for non-commercial use <http://www.pgp.com> iQA/AwUBOQTG4oFlGDz1l6VWEQLZWQCg2zgt49N24DycKkbIZRjSauMmtvAAoNOv F5lmgCmzQ/mS33B6MV2Cu4f2 =4ORS -----END PGP SIGNATURE-----

Joe Dowlen wrote: .... >Don't let the other people out there convince you that the simple switchers >are noisy either. In my current power supply design my noise level on the >output is -109.3 dBv @ 2 kHz (18 bits). Noise is undetectable at the >switching frequency. I only have a 2 stage pi LC filter on the output of >the switcher. The low frequency noise is caused by the feedback control of >the switcher and poses more of a problem for us than the high frequency >noise. Granted, care was taken to use shielded inductors, keep all leads >short, choose the right capacitor and diode etc. but, the design is not far >from the one generated by the simple switcher software. > Joe, I moved this response from the 24-bit thread to this one. When you referred to "simple switchers" and "s.. s.. software" here, were you referring to the LM257x "Simple Switcher" (TM) Natl chips and their design s.w., specifically, or to a basic SMPS design you are using? Other people mentioned having lots of noise problems with those particular chips. Did you crack their nut? Cheers, - Dan Michaels Oricom Technologies ===================