I just put its little brother (the 3103, 750 mA vs 2.5A ) into a board...I'll let you know how that turns out.
Cheers
Phil Hobbs
-- Dr Philip C D Hobbs Principal Consultant ElectroOptical Innovations LLC Optics, Electro-optics, Photonics, Analog Electronics 160 North State Road #203 Briarcliff Manor NY 10510 hobbs at electrooptical dot net http://electrooptical.net
The spikes from that switcher freaked out opamps all over the board. Adding a parallel schottky helped, but not enough.
The LTC sync switchers work great, but they are expensive.
-- John Larkin Highland Technology, Inc lunatic fringe electronics
Well, hopefully the diffusions are different as well as the device sizes....
Cheers
Phil Hobbs
-- Dr Philip C D Hobbs Principal Consultant ElectroOptical Innovations LLC Optics, Electro-optics, Photonics, Analog Electronics 160 North State Road #203 Briarcliff Manor NY 10510 hobbs at electrooptical dot net http://electrooptical.net
Interestingly, the LT Spice delay line element acts like a DC-coupled
1:1 transformer.-- John Larkin Highland Technology, Inc picosecond timing precision measurement jlarkin att highlandtechnology dott com http://www.highlandtechnology.com
Please let us know if that chip snaps, too.
-- John Larkin Highland Technology, Inc picosecond timing precision measurement jlarkin att highlandtechnology dott com http://www.highlandtechnology.com
We'll see in a week when the boards are stuffed.
Cheers
Phil Hobbs
-- Dr Philip C D Hobbs Principal Consultant ElectroOptical Innovations LLC Optics, Electro-optics, Photonics, Analog Electronics 160 North State Road #203 Briarcliff Manor NY 10510 hobbs at electrooptical dot net http://electrooptical.net
Schottky can probably be expected to behave better in series, as they don't have a slowly-oozing charge recovery period, though they do have enough excess capacitance that it looks almost as bad*.
*One sometimes encounters circuits where a junction diode has less "recovery" current than a schottky, and thus runs cooler with it, despite the higher Vf.And for the same reliability reasons, schottky have avalanche capability built in. Occasionally you even see avalanche energy in the datasheet! It's much less than, say, the energy you'd get in a MOSFET of similar ratings, but it's there.
The explanation is: guard rings.
Around the schottky junction itself, PN guard rings are employed, to soften the electric field as the junction gradually works its way to the surface. Thus allowing higher breakdown voltages. It's not too noticeable on the lower voltage (40V and below) parts, but a necessity above there. The schottky junction itself doesn't have avalanche capability (as far as I know), so this is why the avalanche rating (if present) is so small: only a little ring is actually handling it.
Which leads into another interesting wrinkle... pulsed recovery. Again it's not too big a deal for low voltage parts, but high voltage ones (Superjunction and TrenchMOS Si-schottky, and SiC) have enough internal resistance where, for currents above the rating, the guard ring gets forward-biased and therefore exhibits all the crappy behavior of a junction diode.
Curves usually hide this by showing a rather small region, but a few dare to stroke their nose suggestively. Infineon for example.
When you see it, it looks like this: for low voltages and nominal currents, the Vf(I) curve goes up as usual. By about 3V, it starts kicking up with a suddenly much steeper slope -- and with a huge tempco. Of course, the PN guard ring has a modestly high voltage drop (~3V for SiC), so you don't notice it within ratings, but it's the limiting factor under surge duty (well, that and frying the thing).
Normal junction diodes don't appreciate surges, either. I've caused "silent" (i.e., non explosive-psst-BANG) failures of diodes under repetitive surge conditions. Absolutely no bulk thermal heating; my best guess is it was causing electromigration, that eventually resulted in a micro-toasting of the diode, and subsequent failure.
So alas, if you need to handle big surges (polarity or surge protection, snubbing), you need to use diodes rated for the peak, not the average. Schottky will go into guard ring forward bias, and exhibit nasty reverse recovery (which still might not be a problem for a snubber, but you'll be better off with a junction diode instead); and junction diodes have to be rated appropriately, otherwise they'll go toast.
Inversion table? You should find the plans and build a few more; it sounds like a product that would sell well in your neighborhood. Especially the Mission district, and all that.
Tim
-- Seven Transistor Labs, LLC Electrical Engineering Consultation and Contract Design Website: http://seventransistorlabs.com
Last two LTC "medium power" parts (i.e., ~50W class) I took a good look at, made a horrible mess.
Namely, LTC3810, and a uModule.
Both shoved spikes into the input rail (and output for the uModule, because it was a one of the "flying inductor" versions), with pulse widths comparable to (and therefore probably even taller and shorter) the risetime of the scope I was using.
It's a sad and constant truth that applications engineers and development kits are utterly ignorant of EMC issues. We called up LT; best their FAE could suggest was put more ferrite beads on the input and output. Nevermind that it's common mode (ground injection too), and nevermind that even the largest ferrite beads saturate at 10% of the DC bias flowing in the circuit.
Good thing I'm eggspurt at EMC. That project would probably still be lingering if I hadn't been involved. ;-)
I could write an article about the ways in which the '3810 appnote is wrong. I really should; at least the nice thing about external-switch controllers is, you can actually deal with it responsibly.
Now, monolithic controllers, you have no choice, because they never put the power switch drain/collector on a separate pin from the VCC. If they did, you could add a snubber, extra filtering, whatever. Alas...
Tim
-- Seven Transistor Labs, LLC Electrical Engineering Consultation and Contract Design Website: http://seventransistorlabs.com
Where did you expect to see the HV, at the positive rail? The drain-coil junction?
What circuit did you use in your "misspent childhood"?
Mark L. Fergerson
The LM3103 has an internal regulator to produce its VCC, with a pin for a filter cap. I've got a bunch of filtering on the input, which isn't too hard at that power level, so we'll see. It's the first sync buck I've ever used--I needed the totem pole to make a pair of flybuck isolated supplies. (Also a first for me.)
Cheers
Phil Hobbs
-- Dr Philip C D Hobbs Principal Consultant ElectroOptical Innovations LLC Optics, Electro-optics, Photonics, Analog Electronics 160 North State Road #203 Briarcliff Manor NY 10510 hobbs at electrooptical dot net http://electrooptical.net
Old 600 ohm Post Office relays wired make 'n' break.
We use a lot of LTM8023 buck switchers, uModules. They seem to be a classic switcher, PNP switch and schottky catch diode. They behave very well and are really quiet.
-- John Larkin Highland Technology, Inc picosecond timing precision measurement jlarkin att highlandtechnology dott com http://www.highlandtechnology.com
Yeah, I've seen silicon power schottky diodes have nasty reverse recovery at high currents, when the guard ring diode starts to conduct. But wait! Schottkies don't have reverse recovery!
The New Age health thing is mostly over. Now people care about cocktails and craft beers.
-- John Larkin Highland Technology, Inc picosecond timing precision measurement jlarkin att highlandtechnology dott com http://www.highlandtechnology.com
Yeah, that sounds pretty normal.
I don't remember the number, but like I said, it was a "flying inductor" version, so there's an input sync-buck and an output sync-boost (which is also to say: there's an inductor in the middle of an H-bridge, but the '+V' legs of the H-bridge aren't common, but split into "in" and "out" sides). A quad of FETs, internal gate drive and control logic, the works. External inductor. Somewhere around an LGA-144 package.
Tim
-- Seven Transistor Labs, LLC Electrical Engineering Consultation and Contract Design Website: http://seventransistorlabs.com
I meant about fetishism, but maybe I misunderstand the purpose of said table.
Tim
-- Seven Transistor Labs, LLC Electrical Engineering Consultation and Contract Design Website: http://seventransistorlabs.com
Inverted sex? Sounds uncomfortable.
It was for people with back pain, somehow. I just got involved to help a guy who believed in this. He was driving the motor from a variac/rectifier and the motion was all jerky. He didn't believe that the motor drive impedance mattered, but he was an Earth type. So I made him a little motor controller.
-- John Larkin Highland Technology, Inc lunatic fringe electronics
Where did you pick off the HV?
Mark L. Fergerson
Flippin' 'eck! This was 50 years ago! To answer that I'd have to work out what I did back then, build it again and find the hot terminals. And I'm all out of those Post Office relays - by about 35 years. :-D
Thinking back that reminds me about something I'd rather forget. Once I'd got this thing working, I grounded one side with a copper rod into the earth and ran the hot side along a 10 foot wire 8" above ground level using insulated spacers. At intervals along this wire I hung small chunks of fresh offal and stood back to watch the 'fun' as the local half- starved moggies closed in for what they misfigured would be a free lunch. ISTR it was tremendously entertaining watching them gingerly trying to tease the meat off the line and being totally unable to work out WTF was going on. Ten year-old boys can be horrid little c*nts at times. :-/
If you set the relay to self oscillate you can take HV spikes off the switched end of the coil. Long time since I played with stuff like that.
NT
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.