Paralleling Schottky Rectifiers
in [Design]
|
From: dagmargoodboat on 23 Jul 2010 13:20 On Jul 23, 3:21 am, Uwe Hercksen <herck...(a)mew.uni-erlangen.de> wrote: > dagmargoodb...(a)yahoo.com schrieb: > > > I never parallel rectifiers -- that's asking for thermal runaway. > > But, here, OnSemi says it's cool, even necessary to get full rating: > > >http://www.onsemi.com/PowerSolutions/product.do?id=MBR20200CTG > > > I guess those are matched with matched drops, so runaway won't be a > > problem. Hmm. > > Hello, > > they are within the same case, therefore they should have nearly the > same temperature. > > I saw a high current rectifier built with a lot of parallel diodes > mounted on an large heatsink. Series resistors for equal current > distribution might be necessary. > > Bye I've seen ballasting resistors in a few cases. I'm trying to save heat here though--everything's tiny and jammed together. In the old days I'd just throw a TO-220 on a heatsink and not give it a second thought. (An early PC I worked on circa 1978 had a 5V 35A linear supply, uA723 + 2n3055's (darlington'd, IIRC).) -- Cheers, James Arthur
From: Rich Grise on 23 Jul 2010 17:15 On Fri, 23 Jul 2010 08:35:53 +1000, Grant wrote: > On Thu, 22 Jul 2010 10:28:09 -0700 (PDT), dagmargoodboat(a)yahoo.com wrote: > >>I never parallel rectifiers -- that's asking for thermal runaway. But, >>here, OnSemi says it's cool, even necessary to get full rating: >> >>http://www.onsemi.com/PowerSolutions/product.do?id=MBR20200CTG >> >>I guess those are matched with matched drops, so runaway won't be a >>problem. Hmm. >> >>I need a >5A 40V diode (to carry 4A d.c.), preferably surface mount. > > Parallel some 3A diodes? I've often paralleled them, seems a common thing > to do, in that I see it often in stuff I dismantle. >> >>OTOH, that's gonna dissipate ~3W. Ouch. So it has to have a meaty >>package, and it needs some heat-spreading foil. Yuck. > > Another reason for paralleling is to get the smaller on voltage. >> >>Might just use a FET instead, if permitted (it's not my call). > I once spent about $60,000.00 of a client's money trying to develop a compact switcher for wheelchair chargers. The project died because I don't know how to design a switcher; I was going by app notes and telephone advice (this was in the early 1990's before 600W PC supplies came down to $29 bucks.) Anyway, one problem I _didn't_ have was with parallel diodes - they weren't Schottkys (Schottkies?) but "Ultrafast recovery" diodes, that came two in a package - I presume that being in the same package helped thermally stabilize them to each other. Thanks, Rich
From: dagmargoodboat on 23 Jul 2010 22:49 On Jul 23, 4:15 pm, Rich Grise <richgr...(a)example.net> wrote: > On Fri, 23 Jul 2010 08:35:53 +1000, Grant wrote: > > On Thu, 22 Jul 2010 10:28:09 -0700 (PDT), dagmargoodb...(a)yahoo.com wrote: > > >>I never parallel rectifiers -- that's asking for thermal runaway. But, > >>here, OnSemi says it's cool, even necessary to get full rating: > > >>http://www.onsemi.com/PowerSolutions/product.do?id=MBR20200CTG > > >>I guess those are matched with matched drops, so runaway won't be a > >>problem. Hmm. > > >>I need a >5A 40V diode (to carry 4A d.c.), preferably surface mount. > > > Parallel some 3A diodes? I've often paralleled them, seems a common thing > > to do, in that I see it often in stuff I dismantle. > > >>OTOH, that's gonna dissipate ~3W. Ouch. So it has to have a meaty > >>package, and it needs some heat-spreading foil. Yuck. > > > Another reason for paralleling is to get the smaller on voltage. > > >>Might just use a FET instead, if permitted (it's not my call). > > I once spent about $60,000.00 of a client's money trying to develop a > compact switcher for wheelchair chargers. The project died because I don't > know how to design a switcher; I was going by app notes and telephone > advice (this was in the early 1990's before 600W PC supplies came down to > $29 bucks.) Man, I wish I could've helped you. That would've been fun. I'd done several switchers by then. > Anyway, one problem I _didn't_ have was with parallel diodes - they > weren't Schottkys (Schottkies?) but "Ultrafast recovery" diodes, that came > two in a package - I presume that being in the same package helped > thermally stabilize them to each other. Same package helps, close matching helps too. If they're on the same chip you can do magic but that's Jim's trade, not ours. -- Cheers, James Arthur
From: Tim Williams on 23 Jul 2010 23:10 "Rich Grise" <richgrise(a)example.net> wrote in message news:pan.2010.07.23.21.15.27.870660(a)example.net... > I once spent about $60,000.00 of a client's money trying to develop a > compact switcher for wheelchair chargers. The project died because I don't > know how to design a switcher; I was going by app notes and telephone > advice (this was in the early 1990's before 600W PC supplies came down to > $29 bucks.) Well gee, you could've at least tried something. Burning $60,000 of transistors is better than "what'd you spend it on? Uh, nothing". I'm proud to say, in about five years of building my own switching supplies, I have yet to fill even a small bucket with all the transistors I've burned, and I've hardly spent $1000 doing so. Resources are definietly better today, though my equipment's the same. Actually you probably had better equipment, all I have is a rusty Tek 475. Tim -- Deep Friar: a very philosophical monk. Website: http://webpages.charter.net/dawill/tmoranwms
From: John on 23 Jul 2010 23:46
<dagmargoodboat(a)yahoo.com> wrote in message news:f0c2d629-45b7-4331-ae81-03abd4134eb3(a)w31g2000yqb.googlegroups.com... >I never parallel rectifiers -- that's asking for thermal runaway. > But, here, OnSemi says it's cool, even necessary to get full rating: > > http://www.onsemi.com/PowerSolutions/product.do?id=MBR20200CTG > > I guess those are matched with matched drops, so runaway won't be a > problem. Hmm. > > I need a >5A 40V diode (to carry 4A d.c.), preferably surface mount. > > OTOH, that's gonna dissipate ~3W. Ouch. So it has to have a meaty > package, and it needs some heat-spreading foil. Yuck. > > Might just use a FET instead, if permitted (it's not my call). > negative temperature coef, ? bad news as one will hog current. Same case helps a lot you can balance it out somewhat with a low series (each diode) resistance. SiC has a positive temp coef, which is great! |