At 5:1 I'd definitely do that. I use direct step-up for anything up to
10:1, sometimes more, depending on how much power the switcher has to deliver.
LTSpice is really nice for that job. Of course, that means that for all not cost-critical applications I use LTC chips. Which I guess was the purpose why they gave us LTSpice.
Put it on my own web-site? It's not going to make you understand the idea any better than you do now.
"Astatic" - in this context, is just not generating an external magnetic field, nor reacting to changes in an uniform external magnetic field.
Back threading a progressive winding with a single turn around the toroid doesn't make the inductor you produce perfectly astatic, though it's a lot better than it would be without that single turn.
Ayrton-Perry is a good deal closer to prefection, and the "wind half-way around wind back over the full circumfernece of the toroid and then wind back half-way around to get back to the start of the coil is presumably move nearly perfect still.
In the same way that your demand for an example on the web is an evasive question. It would be nice if such an example existed, but absence of - a very particular sort of - evidence isn't evidence of absence.
LT Spice has probably sold a billion dollars worth of chips.
Here's a variant on a boost converter. A standard boost would have a huge current surge at powerup, charging the caps directly through the inductor and diode. I was trying to fix that and came up with the thing below. It's almost a sepic, but the inductors are coupled. C5 isn't necessary but eliminates any leakage inductance effects. It has more copper loss than a plain boost, but uses the core about as well. It can use one of the common dual-winding inductors, who really do have K of 0.99.
This sims at 5 PPM of real time.
Version 4 SHEET 1 1920 940 WIRE -304 32 -368 32 WIRE -176 32 -240 32 WIRE 1296 32 1232 32 WIRE 1424 32 1360 32 WIRE -368 144 -368 32 WIRE -304 144 -368 144 WIRE -176 144 -176 32 WIRE -176 144 -224 144 WIRE 176 144 -176 144 WIRE 464 144 176 144 WIRE 656 144 464 144 WIRE 896 144 784 144 WIRE 1088 144 1040 144 WIRE 1232 144 1232 32 WIRE 1232 144 1152 144 WIRE 1296 144 1232 144 WIRE 1424 144 1424 32 WIRE 1424 144 1376 144 WIRE 1504 144 1424 144 WIRE 1552 144 1504 144 WIRE 1664 144 1552 144 WIRE 656 176 656 144 WIRE 784 176 784 144 WIRE 896 176 896 144 WIRE -368 256 -368 144 WIRE 1552 256 1552 144 WIRE 1664 256 1664 144 WIRE 1424 272 1424 144 WIRE 656 320 656 256 WIRE 752 320 656 320 WIRE 896 320 896 256 WIRE 896 320 816 320 WIRE 1040 320 1040 144 WIRE 1040 320 896 320 WIRE 176 352 176 144 WIRE 1232 368 1232 144 WIRE 656 384 656 320 WIRE -368 400 -368 336 WIRE 1424 400 1424 336 WIRE 1552 400 1552 336 WIRE 1664 400 1664 336 WIRE -176 416 -176 144 WIRE 16 416 -176 416 WIRE 464 416 464 144 WIRE 464 416 336 416 WIRE 656 432 656 384 WIRE -128 512 -384 512 WIRE 16 512 -64 512 WIRE 400 512 336 512 WIRE 544 512 480 512 WIRE 608 512 544 512 WIRE 1232 528 1232 448 WIRE 1232 528 1040 528 WIRE -384 608 -384 512 WIRE -144 608 -384 608 WIRE 16 608 -64 608 WIRE 656 608 656 528 WIRE 656 608 336 608 WIRE 656 640 656 608 WIRE 1232 656 1232 528 WIRE -384 704 -384 608 WIRE -336 704 -384 704 WIRE -192 704 -272 704 WIRE -144 704 -192 704 WIRE 16 704 -64 704 WIRE 464 704 336 704 WIRE 656 768 656 720 WIRE -384 784 -384 704 WIRE 464 816 464 704 WIRE 928 816 464 816 WIRE 1040 816 1040 528 WIRE 1040 816 928 816 WIRE 1232 816 1232 736 WIRE 176 832 176 768 FLAG 1424 400 0 FLAG 1552 400 0 FLAG 1504 144 VP FLAG 656 768 0 FLAG -384 784 0 FLAG 1232 816 0 FLAG -368 400 0 FLAG 1664 400 0 FLAG 176 832 0 FLAG 928 816 FB FLAG -192 704 VCOMP FLAG 656 384 DRAIN FLAG 544 512 GATE FLAG 784 176 0 SYMBOL ind2 640 160 R0 WINDOW 0 -70 40 Left 2 WINDOW 3 -80 83 Left 2 SYMATTR InstName L1 SYMATTR Value 8.2µ SYMATTR Type ind SYMATTR SpiceLine Rser=0.06 SYMBOL schottky 1088 160 R270 WINDOW 0 91 33 VTop 2 WINDOW 3 79 35 VBottom 2 SYMATTR InstName D1 SYMATTR Value 10MQ060N SYMATTR Description Diode SYMATTR Type diode SYMBOL cap 1408 272 R0 WINDOW 0 -66 -4 Left 2 WINDOW 3 -65 32 Left 2 SYMATTR InstName C1 SYMATTR Value 5µ SYMBOL res 1536 240 R0 WINDOW 0 -51 29 Left 2 WINDOW 3 -56 62 Left 2 SYMATTR InstName R1 SYMATTR Value 5K SYMBOL cap -64 496 R90 WINDOW 0 -8 65 VBottom 2 WINDOW 3 -35 2 VTop 2 SYMATTR InstName C2 SYMATTR Value 100p SYMBOL res 1248 464 R180 WINDOW 0 77 76 Left 2 WINDOW 3 65 38 Left 2 SYMATTR InstName R2 SYMATTR Value 290K SYMBOL res 1216 640 R0 WINDOW 0 -65 37 Left 2 WINDOW 3 -74 76 Left 2 SYMATTR InstName R4 SYMATTR Value 10K SYMBOL voltage -368 240 R0 WINDOW 0 48 47 Left 2 WINDOW 3 53 80 Left 2 SYMATTR InstName V2 SYMATTR Value 9 SYMBOL current 1664 256 R0 WINDOW 0 -44 90 Left 2 WINDOW 3 -287 220 Left 2 WINDOW 123 0 0 Left 2 WINDOW 39 0 0 Left 2 SYMATTR InstName I1 SYMATTR Value PULSE(0 1 5m 1u 1u 20u 5m) SYMBOL nmos 608 432 R0 WINDOW 0 129 47 Left 2 WINDOW 3 80 83 Left 2 SYMATTR InstName M1 SYMATTR Value FDS4559_N SYMBOL res 640 624 R0 WINDOW 0 61 41 Left 2 WINDOW 3 56 77 Left 2 SYMATTR InstName R5 SYMATTR Value 0.05 SYMBOL PowerProducts\\LT3757 176 560 R0 SYMATTR InstName U1 SYMBOL res -48 592 R90 WINDOW 0 69 80 VBottom 2 WINDOW 3 41 30 VTop 2 SYMATTR InstName R7 SYMATTR Value 10K SYMBOL res -48 688 R90 WINDOW 0 72 79 VBottom 2 WINDOW 3 44 27 VTop 2 SYMATTR InstName R12 SYMATTR Value 5k SYMBOL cap -272 688 R90 WINDOW 0 72 52 VBottom 2 WINDOW 3 44 7 VTop 2 SYMATTR InstName C4 SYMATTR Value 2n SYMBOL ind2 880 160 R0 WINDOW 0 50 40 Left 2 WINDOW 3 46 78 Left 2 SYMATTR InstName L2 SYMATTR Value 8.2µ SYMATTR Type ind SYMATTR SpiceLine Rser=0.06 SYMBOL cap 816 304 R90 WINDOW 0 70 29 VBottom 2 WINDOW 3 76 28 VTop 2 SYMATTR InstName C5 SYMATTR Value 1µ SYMBOL res 496 496 R90 WINDOW 0 70 75 VBottom 2 WINDOW 3 43 29 VTop 2 SYMATTR InstName R3 SYMATTR Value 50 SYMBOL res 1280 160 R270 WINDOW 0 -36 55 VTop 2 WINDOW 3 -38 53 VBottom 2 SYMATTR InstName R6 SYMATTR Value 1m SYMBOL cap 1360 16 R90 WINDOW 0 67 32 VBottom 2 WINDOW 3 70 32 VTop 2 SYMATTR InstName C3 SYMATTR Value 10m SYMBOL res -320 160 R270 WINDOW 0 -40 53 VTop 2 WINDOW 3 -44 53 VBottom 2 SYMATTR InstName R8 SYMATTR Value 1m SYMBOL cap -240 16 R90 WINDOW 0 65 32 VBottom 2 WINDOW 3 72 29 VTop 2 SYMATTR InstName C6 SYMATTR Value 10m TEXT 1416 720 Left 2 !.tran 0 5m 0 1n uic TEXT 1368 592 Left 2 ;D140 BOOST CONVERTER D TEXT 1416 656 Left 2 ;JL June 22 2013 TEXT 712 248 Left 2 !K L1 L2 0.99
--
John Larkin Highland Technology Inc
www.highlandtechnology.com jlarkin at highlandtechnology dot com
Start-up is interesting. Only simulating the first 5 msec does give a sligh tly optimistic picture. V(drain) was creeping up at 5msec, so I ran the sim a second time, for 10msec, and U1 turned itself off again at 5.5msec. and back on again at 6.6msec, off at 7.12msec and on again at 7.65msec.
U1 was still switching at 10msec, but the circuit was not entirely settled down. I was tempted to run a third time, it for 20msec, but it's your can o f worms ...
Au contraire, I have the feeling you don't understand :-)
I suggest to take a closer look at what an Ayrton-Perry winding is and what it is not. This is the real deal:
formatting link
If you wind that onto anything, then ideally the resulting component has no inductance. How is that doing any good on a toroidal inductor?
No, I am simply stating there is no inductor with an Ayrton-Perry winding. Because it makes no sense. It would be like a car with two drive trains but each pulling in opposite direction.
Not sure if it's that much but in my case (or my clients') alone it's in the six digits by now. LTSpice takes a ton of risk out of designs.
a
It is a SEPIC. Most of my SEPICs have coupled inductors, aside from smaller magnetics that let's you play with the EMI footprint:
formatting link
The real deal would be to design one that's synchronous. LTC has a synchronous boost chip but I haven't yet figured out a way to efficiently use that as a synchronous SEPIC. It is possible, of course, but the schematic gets very busy and big. Never had enough available real estate to do that.
uses
who
Dual cores have less air gap, it's tough to get small flyback transformers that are substantially better than k=0.95.
The LT3757 is a nice chip. I used it a lot, including for some very unorthodox non-switcher circuits.
That is the web-site I found for you. I wasn't paying enough attention, bec ause it is a Ayrton-Perry non-inductive/astatic resistor and not the Ayrton
-Perry astatic/no-external-field inductor.
Ayrton and Perry were two historical characters who got their names attache d to two different components - an astatic resistor and an astatic inductor .
A web-site for the resistor doesn't make it the only "real deal" in town.
Why would you apply a scheme devised to make a non-inductive resistor, to an inductor? Ayrton and Perry invented a non-inductive resistor and a parti cular way of making an astatic inductor. Their names are attached to both c omponents but this doesn't make the two sorts of components interchangeable , or even all that similar.
Sadly, you are making a false statement. The Ayrton-Perry resistor and the Ayrton-Perry inductor are two distinct and different devices. The underlyin g idea is the same, but the realisation of it in the resistor and the induc tor are necessarily different.
If you were to confuse a resistor with an inductor, you might think that.
I've developed an aversion to synchronous switchers. Too many of them have created astounding EMI problems. The crossover is inherently nasty.
uses
who
I tried charging a 1000 uF cap in the sim. It ran for 10 hours, created a 10 gbyte .raw file, and it was only halfway up to my target voltage. I stopped it. Got to finish my book, anyhow.
LT3757 is a nice chip. Have you seen any funnies? It has enough external components to let me tune everything, but it's not too smart like some other chips.
We recently had a few failures in LTM8023's. We sent one back to LTC and insisted on a failure analysis. The results finally came back:
It's our fault.
They will change their die attach method as part of their "continuous process improvement" program.
--
John Larkin Highland Technology Inc
www.highlandtechnology.com jlarkin at highlandtechnology dot com
I've done several fairly simple SMPS designs, which worked fine, but now I'm going to have to get good at it. My usual way of doing that sort of thing is to get a good upper level undergraduate text and read it like a novel, maybe do some of the exercises, and then just start in doing the work.
Ideally it would cover the ins and outs of the major converter families, and talk about when and why you'd use one or another, and what to watch out for. (A lore book, in other words.)
Any suggestions? (Amazon reviews of Maniktala's books are very positive, but he seems to have written about 5 of the same book.)
I don't have any books on switchers. I guess I've just learned stuff from data sheets and appnotes and other guys and experimenting.
Some lore:
Iron powder cores can get really hot.
Switching speeds in IC converters are getting crazy, a couple ns rise/fall. The amps/second are getting scary. External fets are good because you can add gate resistors to slow things down.
Synchronous switchers can be really nasty. My un-favorite waveform:
formatting link
I just spun a PCB to take out a couple of LTC3411's that were making more RF than a customer liked in his system. The big spectral lines were harmonics of the switcher frequency, in the 5-9 MHz range.
LTM8023s are nice and quiet. As are the older National Simple Switchers.
PCB loop currents really matter.
Ferite beads around switchers are often useful if you have any low-level stuff on the same board.
You can charge pump off switching nodes, or make voltages off windings added to the main inductor.
Continuous mode is great.
Tantalums explode. Polymer alums and ceramics don't.
Switchers have negative input impedances, so can have startup problems. Some wall-warts can't kick off some switchers. Use soft-start if it's available, and reset it!
It's fun to make your own switchers.
There must be a lot more.
--
John Larkin Highland Technology Inc
www.highlandtechnology.com jlarkin at highlandtechnology dot com
Only on badly designed chips. LTC's are among the best, but even there I found bloopers, missing leading-edge blanking and such. Sometimes I take care of cross-over myself, with two resistors, a cap and one diode per side and then through a muscular gate driver chip. One only has to be careful that the added prop delay won't cause the leading-edge blanking to become ineffective.
Whenever I can I design switchers synchronous. Never had much of an EMI problem but they sure are efficient.
uses
who
That's why things like soft-start also have to be cut short. Or with "trickle-on" primary switchers one has to send a huge jolt into the VCC cap so they start before the computer developed cobwebs.
Are you writing a book?
Never seen any funnies or unruly behavior with that as far as I can remember. And I have also used them as pulsers and in all sorts of odd situations.
process
That's one of the things I like about LTC, they are honest and they fess up when they screwed up. Like when I found a boo-boo in the design of the LT6700 series. It resulted in an apology, thankfulness, and immediate corrective action. In today's corporate world that is remarkable. I have seen very different behavior in larger semiconductor companies.
I had to get into switchers, big time, in the early 90's. What I obtained:
a. Good face and eye protection.
b. A large fire extinguisher (no, that's not a joke).
c. A bench kill switch for power (all power).
d. Several pounds of ferrite parts.
e. A big angle grinder.
f. The 1990 "Unitrode IC Data Handbook". That was pretty much the only switcher book I ever had, Unitrode was "the" place where the experts hung out. Meantime they were swallowed by TI and it is possible that the collection of app notes in there is available as PDF.
I have heard that a book by Pressman is considered the master book for EEs getting into switchers these days but I don't have it. Regarding university or academic literature about SMPS, I was never very impressed. But this site is great if you want to kick around architectures:
They can also spectacularly disintegrate. I learned that lesson as a teenager, trying to squeeze 1200 watts of RF through a T200-2 because I "invested" the money for the 2nd core I should have stacked on into a crate of Pilsener beer. The result ... *KABOOM*
The
formatting link
Couldn't it be dithered?
And nothing beats 4oz copper :-)
to
Yup. And those make for interesting conversations in design reviews.
Having seen orange and green clouds from exploding tants, I wonder whether I might have breathed stuff that has already shortened my life span ...
and
Switchers can make other switchers go totally berserk even after start-up.
Some of it is etched as scars into the Formica of my lab bench.
Thanks, I've gleaned quite a few pointers from you and others on SED. The switchers I've made previously were either roll-my-own hysteretic bucks or LM2594-based bucks and inverters. Probably the most fun one was a hysteretic buck regulator for driving heaters in a high-performance temperature control loop for a passive-cavity stabilized laser. I got about 30W out of it with nanovolt ripple and nanoamp current noise injection into the upstream supply. Good medicine, but of course the peak efficiency was only about 80%.
I've become a fan of using computer wall warts for instrument power, because they have easily enough output for just about anything I do, and they're cheap and reliable. Since they produce 16-20V, the switcher efficiencies aren't too bad, and using some nice Coilcraft dual-winding toroids I have a range of things I can do with them without dying from interference.
I read a good book on photocathodes the other day, and learned a lot. I really like doing that.
One idea I had was to taper a fiber really small and deposit a photocathode on the end. Shoot a lot of light into the fiber, and you get a really small electron emitter, like for a cheap student SEM.
--
John Larkin Highland Technology Inc
www.highlandtechnology.com jlarkin at highlandtechnology dot com
That sort of thing has been done in the optical microscope world, e.g. in Eric Betzig's papers on near-field scanning optical microscopes (NSOMs) from the late 1980s and early '90s.
Around the same time, some of my fellow IBM Yorktown folks (Jean-Marc Halbout and co.) used photocathodes deposited on fibre facets illuminated by picosecond lasers, to make stroboscopic SEMs.
It's a cute idea. The basic issue IIUC is that even way up in the gun, SEM vacuums are too dirty for caesium-activated photocathodes to last very long.
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.