On Wednesday, July 17, 2013 11:25:53 AM UTC-4, Jeroen Belleman wrote:> On 2013-07-17 15:42, George Herold wrote: > > > > > > Sorry for the diverting question, but help me out here. (I find > > > magnet material a bit magical.) So I thought the energy in an > > > inductor was 1/2*L*i^2. > > > > > > But your and Joerg's comments seem to imply the energy is mostly in > > > the gap. > > > > > > Now my very limited understanding of making a gap in an inductor was > > > that the gap sorta sets the inductance... the actual value isn't so > > > dependent on the magnetic material. And doesn't the gap reduce the > > > inductance? > > > > > > I seem to be missing something fundamental. > > > > > > Thanks, George H. > > > > Hi George, > > > > The magnetic energy in a given volume is E=1/2 VB^2 / (u_0 u_r), > > so for a given B, the energy in a given volume of ferrite is u_r > > times less than in the same volume of air. > > > > For a ferrite magnetic circuit with an air gap, the total > > magnetic energy works out as E=1/2 B^2 (A l_f/(u_0 u_r) + A l_a/u0), > > where A is the cross section area of the magnetic circuit, and l_f > > and l_a are the path length in ferrite and air, respectively. > > (I neglect fringing. That's good enough, usually.) > > > > You can work this out from just two of Maxwell's four laws. > > > > Cheers, > > Jeroen BellemanThanks Jeroen, Reading Phil's reply reminded me of the discussion in Feynman's book. I maybe need a bit of hand holding as I 'walk' across the gap, and Dick will be the perfect person for that. :^) George H.
Flyback vs half-bridge
Started by ●July 16, 2013
Reply by ●July 17, 20132013-07-17
Reply by ●July 17, 20132013-07-17
Phil Hobbs wrote:> On 07/17/2013 10:00 AM, Joerg wrote: >> Phil Hobbs wrote: >>> On 7/16/2013 8:54 PM, Joerg wrote: >>>> Phil Hobbs wrote: >>>>> On 7/16/2013 3:53 PM, Joerg wrote:[...]>>> 1 uF. I can probably post a scope photo tomorrow, if I have time--a >>> bunch of lawyering has just descended on my head. (Not that I mind very >>> much--expert witness work uses a different part of my brain and >>> generally pays somewhat better than design work.) >>> >> >> That's one of the more serious problems we as a country have. Legal >> skirmishes usually have zero productivity from a national perspective >> but result in the fattest "payouts". It is the core problem why our >> health care system is a mess. Of course, that is never brought up by >> body politicus, for obvious reasons :-( > > You could say much the same thing about the army, but once the shooting > starts, you're glad they're there. The case I'm working on today > concerns an alleged misappropriation of trade secrets by a big > semiconductor company. There's a lot of money at stake, so I'm pretty > much in the noise. Designing stuff for start-ups is a bit different. >Designing for a start-up is what I am doing right now. Plus some regular consulting work. It's a lot more fun than legal stuff. [...]>> >>> ... Of course the ungapped ferrite has a pretty >>> frightening tempco of mu (it goes down by half when you hit it with cold >>> spray), so some sort of current control is going to be required, for >>> sure. >>> >> >> Yes, needs current control but that's standard even in sub-Dollar chips. > > I may just make a boost with a two-winding toroid, and return the other > end of the second winding to the raw -15V supply. (A lot of the Murata > toroids are actually dual-winding, but they expect you to wire them in > parallel to get the current handling capacity.) >But you only need 20mA. Even some common-mode chokes might be able to handle that. Also, there are some E-E core dual inductors that have a performance that could rival toroids. Plus there's always ye old shield can.> There are some PoE things that look like they could usefully be run > backwards, but the coupling coefficients are low enough that they must > have gaps. >Most PoE transformers are for flybacks, probably because that is the cheapest converter architecture. -- Regards, Joerg http://www.analogconsultants.com/
Reply by ●July 17, 20132013-07-17
George Herold wrote: [...]> Say are there any good books about transformers/ inductors/ magnetic > materials? The subject seems to go from the trivial freshman physics > transformer. To the "full hair ball" real world situation, with > nothing in between. >Wuerth has the "Trilogy of Magnetics". It's probably good for beginners, around $50. I've got it here because a sales guy dropped it off, but will give it back because I outgrew that a long time ago. All I had in the early 90's when I had to design my 1st mass production switcher was the "Unitrode IC Data Handbook". That got me going in this field. Probably TI still has it online somewhere because they swallowed Unitrode. -- Regards, Joerg http://www.analogconsultants.com/
Reply by ●July 17, 20132013-07-17
On Wednesday, July 17, 2013 5:18:25 PM UTC+2, Phil Hobbs wrote:> On 07/17/2013 10:00 AM, Joerg wrote: > > > Phil Hobbs wrote: > > >> On 7/16/2013 8:54 PM, Joerg wrote: > > >>> Phil Hobbs wrote: > > >>>> On 7/16/2013 3:53 PM, Joerg wrote: > > >>>>> Phil Hobbs wrote: > > >>>>>> So, I'm goofing off playing with small switchers. (Well, not exactly > > >>>>>> goofing off, but there are somewhat more pressing tasks waiting....) > > >>>>>> > > >>>>>> As I said in George's thread, "Opamp w/ Vsupply > 36V", I built a > > >>>>>> small > > >>>>>> half-bridge supply with positive and negative voltage doublers. > > >>>>>> I'm not > > >>>>>> that keen on it, because (a) it uses a fair number of parts for > > >>>>>> what it > > >>>>>> does, and (b) it has a nasty instability. > > >>>>>> > > >>>>> > > >>>>> (a) ... yes. (b) ... why? What happens? Changing to another > > >>>>> architecture > > >>>>> while using the same kind of loop usually doesn't do much to improve > > >>>>> stability. > > >>>>> > > >>>>> > > >>>>>> With a capacitively-coupled half bridge, if you let the transformer > > >>>>>> saturate, it instantly discharges the coupling cap, which about > > >>>>>> doubles > > >>>>>> the volt-seconds on the next half-cycle and guarantee that it will > > >>>>>> keep > > >>>>>> saturating until the FETs cook themselves. > > >>>>>> > > >>>>> > > >>>>> Not sure what you mean here, but usually current mode control is the > > >>>>> way > > >>>>> to avoid asymmetrical runaway. > > >>>> > > >>>> The IRS2153D puts a square wave out of a half-bridge--no feedback, no > > >>>> current limit, nada. If you put that into a transformer via a cap, all > > >>>> is well until you saturate the transformer. > > >>>> > > >>> > > >>> Ok, the IRS2153D is just a glorified gate driver with an oscillator in > > >>> there. I'd use a real switcher controller chip, some are in the same > > >>> price category. > > >>> > > >>> > > >>>> Say that happens on the positive half-cycle. > > >>>> At that moment, the voltage on the cap rapidly goes from V_DD/2 to V_DD. > > >>>> At the next edge, the voltage across the transformer is suddenly not > > >>>> V_DD/2 as expected, but V_DD. The transformer saturates in half the > > >>>> time it took previously, and the voltage on the cap goes from V_DD to 0. > > >>>> Then the cycle repeats. It's really obvious on a scope when this > > >>>> happens, and it's far from pretty. > > >>>> > > >>> > > >>> Did you use a really big cap? I've never had that happen. Usually > > >>> ferrite saturates softly enough to just "nudge" the cap. > > >> > > >> 1 uF. I can probably post a scope photo tomorrow, if I have time--a > > >> bunch of lawyering has just descended on my head. (Not that I mind very > > >> much--expert witness work uses a different part of my brain and > > >> generally pays somewhat better than design work.) > > >> > > > > > > That's one of the more serious problems we as a country have. Legal > > > skirmishes usually have zero productivity from a national perspective > > > but result in the fattest "payouts". It is the core problem why our > > > health care system is a mess. Of course, that is never brought up by > > > body politicus, for obvious reasons :-( > > > > You could say much the same thing about the army, but once the shooting > > starts, you're glad they're there. The case I'm working on today > > concerns an alleged misappropriation of trade secrets by a big > > semiconductor company. There's a lot of money at stake, so I'm pretty > > much in the noise. Designing stuff for start-ups is a bit different. > > > > > > > > > > >> <snip> > > >>>>> John Larkin has recently used these tiny flyback transformers. Nothing > > >>>>> wrong with hanging two in parallel on the primary and using the 2nd one > > >>>>> with the secondary flipped around, for the -45V. > > >>>>> > > >>>>> Do you need this isolated? Else you could consider just inverting and > > >>>>> boost. > > >>>>> > > >>>> > > >>>> Doesn't have to be isolated, but it does need to be quiet, hence the > > >>>> toroids. > > >>>> > > >>> > > >>> That usually rules out flybacks. They spew a lot of noise. Then I'd use > > >>> a half-bridge with CM control and series inductors on the output. > > >> > > >> Fortunately the half-bridge's worst-case condition is zero load, so it > > >> isn't too hard to control. ... > > > > > > > > > They work ok with zero load if you can tolerate the voltage increase. > > > That would be determined by the turns ratio. If it has to be 45.000V > > > that's a different story. Then you'll have to let it go into a skipping > > > mode. > > > > > > > > >> ... Of course the ungapped ferrite has a pretty > > >> frightening tempco of mu (it goes down by half when you hit it with cold > > >> spray), so some sort of current control is going to be required, for sure. > > >> > > > > > > Yes, needs current control but that's standard even in sub-Dollar chips. > > > > I may just make a boost with a two-winding toroid, and return the other > > end of the second winding to the raw -15V supply. (A lot of the Murata > > toroids are actually dual-winding, but they expect you to wire them in > > parallel to get the current handling capacity.) >they generally spec them both for series and parallel ;) how about a capacitive coupled boost? something like this: http://obsoletetechnology.wordpress.com/projects/studio-electronics/dc-dc-bipolar-power-supply-for-effect-pedals/ -Lasse
Reply by ●July 17, 20132013-07-17
Klaus Kragelund wrote:> On Wednesday, July 17, 2013 4:06:25 PM UTC+2, Joerg wrote: >> Klaus Kragelund wrote: >> >>> On Tuesday, July 16, 2013 9:53:11 PM UTC+2, Joerg wrote: >>>> Phil Hobbs wrote: >> >> >> [...] >> >> >> >>>>> With a capacitively-coupled half bridge, if you let the >>>>> transformer saturate, it instantly discharges the coupling >>>>> cap, which about doubles the volt-seconds on the next >>>>> half-cycle and guarantee that it will keep saturating until >>>>> the FETs cook themselves. >>>> Not sure what you mean here, but usually current mode control >>>> is the way to avoid asymmetrical runaway. >>> You need to be careful about that. Even if you have primary >>> current mode peak limit, then inequalities in the secondary >>> diodes forward drops can let the xformer go into staircasing, >>> saturating the xformer. >> >> >> If the current limit works properly it'll curb in time, it normally >> >> >> won't let it staircase into a danger zone. Of course, all this >> assumes >> >> regular ferrite where the core will not saturate super-hard. >> >> >> >> I am just wondering whether all that isn't overkill for generating >> two >> >> 45V rails at 20mA. That's just a couple of watts. A decent gate >> driver >> >> should almost be able to capacitively drive that without any FETs. >> > > In this case I think the xformer RDC and MOSFET RDC would be so high > that a solution with just a forward converter capacitive coupled > primary would be fine. >Yup, done a few of those myself, they work fine. Only 2W are needed here.> For the flyback it needs to gapped. Non-gapped flybacks is asking for > trouble >It's going to result in *KABLAM* :-) But since Phil's stuff is all noise-critical he probably should not use a flyback architecture. Gapped cores spew a lot of EMI around. -- Regards, Joerg http://www.analogconsultants.com/
Reply by ●July 17, 20132013-07-17
On Wednesday, July 17, 2013 12:25:06 PM UTC-4, Joerg wrote:> Klaus Kragelund wrote: > > > On Wednesday, July 17, 2013 4:06:25 PM UTC+2, Joerg wrote: > > >> Klaus Kragelund wrote: > > >> > > >>> On Tuesday, July 16, 2013 9:53:11 PM UTC+2, Joerg wrote: > > >>>> Phil Hobbs wrote: > > >> > > >> > > >> [...] > > >> > > >> > > >> > > >>>>> With a capacitively-coupled half bridge, if you let the > > >>>>> transformer saturate, it instantly discharges the coupling > > >>>>> cap, which about doubles the volt-seconds on the next > > >>>>> half-cycle and guarantee that it will keep saturating until > > >>>>> the FETs cook themselves. > > >>>> Not sure what you mean here, but usually current mode control > > >>>> is the way to avoid asymmetrical runaway. > > >>> You need to be careful about that. Even if you have primary > > >>> current mode peak limit, then inequalities in the secondary > > >>> diodes forward drops can let the xformer go into staircasing, > > >>> saturating the xformer. > > >> > > >> > > >> If the current limit works properly it'll curb in time, it normally > > >> > > >> > > >> won't let it staircase into a danger zone. Of course, all this > > >> assumes > > >> > > >> regular ferrite where the core will not saturate super-hard. > > >> > > >> > > >> > > >> I am just wondering whether all that isn't overkill for generating > > >> two > > >> > > >> 45V rails at 20mA. That's just a couple of watts. A decent gate > > >> driver > > >> > > >> should almost be able to capacitively drive that without any FETs. > > >> > > > > > > In this case I think the xformer RDC and MOSFET RDC would be so high > > > that a solution with just a forward converter capacitive coupled > > > primary would be fine. > > > > > > > Yup, done a few of those myself, they work fine. Only 2W are needed here. > > > > > > > For the flyback it needs to gapped. Non-gapped flybacks is asking for > > > trouble > > > > > > > It's going to result in *KABLAM* :-) > > > > But since Phil's stuff is all noise-critical he probably should not use > > a flyback architecture. Gapped cores spew a lot of EMI around.Do they make gapped toroids? You could make something with symmetrical gaps one on each side, or maybe a bunch of stripes...(I guess that's hard to make.) George H.> > > > -- > > Regards, Joerg > > > > http://www.analogconsultants.com/
Reply by ●July 17, 20132013-07-17
On 07/17/2013 12:16 PM, Lasse Langwadt Christensen wrote:> On Wednesday, July 17, 2013 5:18:25 PM UTC+2, Phil Hobbs wrote: >> On 07/17/2013 10:00 AM, Joerg wrote: >> >>> Phil Hobbs wrote: >> >>>> On 7/16/2013 8:54 PM, Joerg wrote: >> >>>>> Phil Hobbs wrote: >> >>>>>> On 7/16/2013 3:53 PM, Joerg wrote: >> >>>>>>> Phil Hobbs wrote: >> >>>>>>>> So, I'm goofing off playing with small switchers. (Well, not exactly >> >>>>>>>> goofing off, but there are somewhat more pressing tasks waiting....) >> >>>>>>>> >> >>>>>>>> As I said in George's thread, "Opamp w/ Vsupply > 36V", I built a >> >>>>>>>> small >> >>>>>>>> half-bridge supply with positive and negative voltage doublers. >> >>>>>>>> I'm not >> >>>>>>>> that keen on it, because (a) it uses a fair number of parts for >> >>>>>>>> what it >> >>>>>>>> does, and (b) it has a nasty instability. >> >>>>>>>> >> >>>>>>> >> >>>>>>> (a) ... yes. (b) ... why? What happens? Changing to another >> >>>>>>> architecture >> >>>>>>> while using the same kind of loop usually doesn't do much to improve >> >>>>>>> stability. >> >>>>>>> >> >>>>>>> >> >>>>>>>> With a capacitively-coupled half bridge, if you let the transformer >> >>>>>>>> saturate, it instantly discharges the coupling cap, which about >> >>>>>>>> doubles >> >>>>>>>> the volt-seconds on the next half-cycle and guarantee that it will >> >>>>>>>> keep >> >>>>>>>> saturating until the FETs cook themselves. >> >>>>>>>> >> >>>>>>> >> >>>>>>> Not sure what you mean here, but usually current mode control is the >> >>>>>>> way >> >>>>>>> to avoid asymmetrical runaway. >> >>>>>> >> >>>>>> The IRS2153D puts a square wave out of a half-bridge--no feedback, no >> >>>>>> current limit, nada. If you put that into a transformer via a cap, all >> >>>>>> is well until you saturate the transformer. >> >>>>>> >> >>>>> >> >>>>> Ok, the IRS2153D is just a glorified gate driver with an oscillator in >> >>>>> there. I'd use a real switcher controller chip, some are in the same >> >>>>> price category. >> >>>>> >> >>>>> >> >>>>>> Say that happens on the positive half-cycle. >> >>>>>> At that moment, the voltage on the cap rapidly goes from V_DD/2 to V_DD. >> >>>>>> At the next edge, the voltage across the transformer is suddenly not >> >>>>>> V_DD/2 as expected, but V_DD. The transformer saturates in half the >> >>>>>> time it took previously, and the voltage on the cap goes from V_DD to 0. >> >>>>>> Then the cycle repeats. It's really obvious on a scope when this >> >>>>>> happens, and it's far from pretty. >> >>>>>> >> >>>>> >> >>>>> Did you use a really big cap? I've never had that happen. Usually >> >>>>> ferrite saturates softly enough to just "nudge" the cap. >> >>>> >> >>>> 1 uF. I can probably post a scope photo tomorrow, if I have time--a >> >>>> bunch of lawyering has just descended on my head. (Not that I mind very >> >>>> much--expert witness work uses a different part of my brain and >> >>>> generally pays somewhat better than design work.) >> >>>> >> >>> >> >>> That's one of the more serious problems we as a country have. Legal >> >>> skirmishes usually have zero productivity from a national perspective >> >>> but result in the fattest "payouts". It is the core problem why our >> >>> health care system is a mess. Of course, that is never brought up by >> >>> body politicus, for obvious reasons :-( >> >> >> >> You could say much the same thing about the army, but once the shooting >> >> starts, you're glad they're there. The case I'm working on today >> >> concerns an alleged misappropriation of trade secrets by a big >> >> semiconductor company. There's a lot of money at stake, so I'm pretty >> >> much in the noise. Designing stuff for start-ups is a bit different. >> >> >> >>> >> >>> >> >>>> <snip> >> >>>>>>> John Larkin has recently used these tiny flyback transformers. Nothing >> >>>>>>> wrong with hanging two in parallel on the primary and using the 2nd one >> >>>>>>> with the secondary flipped around, for the -45V. >> >>>>>>> >> >>>>>>> Do you need this isolated? Else you could consider just inverting and >> >>>>>>> boost. >> >>>>>>> >> >>>>>> >> >>>>>> Doesn't have to be isolated, but it does need to be quiet, hence the >> >>>>>> toroids. >> >>>>>> >> >>>>> >> >>>>> That usually rules out flybacks. They spew a lot of noise. Then I'd use >> >>>>> a half-bridge with CM control and series inductors on the output. >> >>>> >> >>>> Fortunately the half-bridge's worst-case condition is zero load, so it >> >>>> isn't too hard to control. ... >> >>> >> >>> >> >>> They work ok with zero load if you can tolerate the voltage increase. >> >>> That would be determined by the turns ratio. If it has to be 45.000V >> >>> that's a different story. Then you'll have to let it go into a skipping >> >>> mode. >> >>> >> >>> >> >>>> ... Of course the ungapped ferrite has a pretty >> >>>> frightening tempco of mu (it goes down by half when you hit it with cold >> >>>> spray), so some sort of current control is going to be required, for sure. >> >>>> >> >>> >> >>> Yes, needs current control but that's standard even in sub-Dollar chips. >> >> >> >> I may just make a boost with a two-winding toroid, and return the other >> >> end of the second winding to the raw -15V supply. (A lot of the Murata >> >> toroids are actually dual-winding, but they expect you to wire them in >> >> parallel to get the current handling capacity.) >> > > they generally spec them both for series and parallel ;)You'd think so, but no. The ones I'm using (Murata 32331C) aren't called out in the datasheet as anything except a single winding between pins 1 and 3, whereas in reality there's another independent one between pins 2 and 4 (the datasheet is at http://tinyurl.com/oyxguho ).> > how about a capacitive coupled boost? > > something like this: > http://obsoletetechnology.wordpress.com/projects/studio-electronics/dc-dc-bipolar-power-supply-for-effect-pedals/ > > -Lasse >It's a possibility, for sure. I normally don't think of capacitive boosters and hundreds of milliwatts at the same time, but it could work OK. It's slightly inconvenient due to having to run off +-15 to get the required output, but not awful. I already have a buck-boost generating -15V from a +16-20V laptop supply, so if I voltage-doubled that and stacked it on top of the raw +-15 rails, I'd get there that way as well. I wouldn't have as much control over the supply sequencing that way, but you can't beat the price. Fun stuff. 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
Reply by ●July 17, 20132013-07-17
On 07/17/2013 12:35 PM, George Herold wrote:> On Wednesday, July 17, 2013 12:25:06 PM UTC-4, Joerg wrote: >> Klaus Kragelund wrote: >> >>> On Wednesday, July 17, 2013 4:06:25 PM UTC+2, Joerg wrote: >> >>>> Klaus Kragelund wrote: >> >>>> >> >>>>> On Tuesday, July 16, 2013 9:53:11 PM UTC+2, Joerg wrote: >> >>>>>> Phil Hobbs wrote: >> >>>> >> >>>> >> >>>> [...] >> >>>> >> >>>> >> >>>> >> >>>>>>> With a capacitively-coupled half bridge, if you let the >> >>>>>>> transformer saturate, it instantly discharges the coupling >> >>>>>>> cap, which about doubles the volt-seconds on the next >> >>>>>>> half-cycle and guarantee that it will keep saturating until >> >>>>>>> the FETs cook themselves. >> >>>>>> Not sure what you mean here, but usually current mode control >> >>>>>> is the way to avoid asymmetrical runaway. >> >>>>> You need to be careful about that. Even if you have primary >> >>>>> current mode peak limit, then inequalities in the secondary >> >>>>> diodes forward drops can let the xformer go into staircasing, >> >>>>> saturating the xformer. >> >>>> >> >>>> >> >>>> If the current limit works properly it'll curb in time, it normally >> >>>> >> >>>> >> >>>> won't let it staircase into a danger zone. Of course, all this >> >>>> assumes >> >>>> >> >>>> regular ferrite where the core will not saturate super-hard. >> >>>> >> >>>> >> >>>> >> >>>> I am just wondering whether all that isn't overkill for generating >> >>>> two >> >>>> >> >>>> 45V rails at 20mA. That's just a couple of watts. A decent gate >> >>>> driver >> >>>> >> >>>> should almost be able to capacitively drive that without any FETs. >> >>>> >> >>> >> >>> In this case I think the xformer RDC and MOSFET RDC would be so high >> >>> that a solution with just a forward converter capacitive coupled >> >>> primary would be fine. >> >>> >> >> >> >> Yup, done a few of those myself, they work fine. Only 2W are needed here. >> >> >> >> >> >>> For the flyback it needs to gapped. Non-gapped flybacks is asking for >> >>> trouble >> >>> >> >> >> >> It's going to result in *KABLAM* :-) >> >> >> >> But since Phil's stuff is all noise-critical he probably should not use >> >> a flyback architecture. Gapped cores spew a lot of EMI around. > > Do they make gapped toroids? You could make something with symmetrical gaps one on each side, or maybe a bunch of stripes...(I guess that's hard to make.) > > George H.You can get pot cores with gaps in the centre post, which are pretty good though not as good as a toroid. 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
Reply by ●July 17, 20132013-07-17
On Wed, 17 Jul 2013 09:16:55 -0700 (PDT), Lasse Langwadt Christensen <langwadt@fonz.dk> wrote:>On Wednesday, July 17, 2013 5:18:25 PM UTC+2, Phil Hobbs wrote: >> On 07/17/2013 10:00 AM, Joerg wrote: >> >> > Phil Hobbs wrote: >> >> >> On 7/16/2013 8:54 PM, Joerg wrote: >> >> >>> Phil Hobbs wrote: >> >> >>>> On 7/16/2013 3:53 PM, Joerg wrote: >> >> >>>>> Phil Hobbs wrote: >> >> >>>>>> So, I'm goofing off playing with small switchers. (Well, not exactly >> >> >>>>>> goofing off, but there are somewhat more pressing tasks waiting....) >> >> >>>>>> >> >> >>>>>> As I said in George's thread, "Opamp w/ Vsupply > 36V", I built a >> >> >>>>>> small >> >> >>>>>> half-bridge supply with positive and negative voltage doublers. >> >> >>>>>> I'm not >> >> >>>>>> that keen on it, because (a) it uses a fair number of parts for >> >> >>>>>> what it >> >> >>>>>> does, and (b) it has a nasty instability. >> >> >>>>>> >> >> >>>>> >> >> >>>>> (a) ... yes. (b) ... why? What happens? Changing to another >> >> >>>>> architecture >> >> >>>>> while using the same kind of loop usually doesn't do much to improve >> >> >>>>> stability. >> >> >>>>> >> >> >>>>> >> >> >>>>>> With a capacitively-coupled half bridge, if you let the transformer >> >> >>>>>> saturate, it instantly discharges the coupling cap, which about >> >> >>>>>> doubles >> >> >>>>>> the volt-seconds on the next half-cycle and guarantee that it will >> >> >>>>>> keep >> >> >>>>>> saturating until the FETs cook themselves. >> >> >>>>>> >> >> >>>>> >> >> >>>>> Not sure what you mean here, but usually current mode control is the >> >> >>>>> way >> >> >>>>> to avoid asymmetrical runaway. >> >> >>>> >> >> >>>> The IRS2153D puts a square wave out of a half-bridge--no feedback, no >> >> >>>> current limit, nada. If you put that into a transformer via a cap, all >> >> >>>> is well until you saturate the transformer. >> >> >>>> >> >> >>> >> >> >>> Ok, the IRS2153D is just a glorified gate driver with an oscillator in >> >> >>> there. I'd use a real switcher controller chip, some are in the same >> >> >>> price category. >> >> >>> >> >> >>> >> >> >>>> Say that happens on the positive half-cycle. >> >> >>>> At that moment, the voltage on the cap rapidly goes from V_DD/2 to V_DD. >> >> >>>> At the next edge, the voltage across the transformer is suddenly not >> >> >>>> V_DD/2 as expected, but V_DD. The transformer saturates in half the >> >> >>>> time it took previously, and the voltage on the cap goes from V_DD to 0. >> >> >>>> Then the cycle repeats. It's really obvious on a scope when this >> >> >>>> happens, and it's far from pretty. >> >> >>>> >> >> >>> >> >> >>> Did you use a really big cap? I've never had that happen. Usually >> >> >>> ferrite saturates softly enough to just "nudge" the cap. >> >> >> >> >> >> 1 uF. I can probably post a scope photo tomorrow, if I have time--a >> >> >> bunch of lawyering has just descended on my head. (Not that I mind very >> >> >> much--expert witness work uses a different part of my brain and >> >> >> generally pays somewhat better than design work.) >> >> >> >> >> > >> >> > That's one of the more serious problems we as a country have. Legal >> >> > skirmishes usually have zero productivity from a national perspective >> >> > but result in the fattest "payouts". It is the core problem why our >> >> > health care system is a mess. Of course, that is never brought up by >> >> > body politicus, for obvious reasons :-( >> >> >> >> You could say much the same thing about the army, but once the shooting >> >> starts, you're glad they're there. The case I'm working on today >> >> concerns an alleged misappropriation of trade secrets by a big >> >> semiconductor company. There's a lot of money at stake, so I'm pretty >> >> much in the noise. Designing stuff for start-ups is a bit different. >> >> >> >> > >> >> > >> >> >> <snip> >> >> >>>>> John Larkin has recently used these tiny flyback transformers. Nothing >> >> >>>>> wrong with hanging two in parallel on the primary and using the 2nd one >> >> >>>>> with the secondary flipped around, for the -45V. >> >> >>>>> >> >> >>>>> Do you need this isolated? Else you could consider just inverting and >> >> >>>>> boost. >> >> >>>>> >> >> >>>> >> >> >>>> Doesn't have to be isolated, but it does need to be quiet, hence the >> >> >>>> toroids. >> >> >>>> >> >> >>> >> >> >>> That usually rules out flybacks. They spew a lot of noise. Then I'd use >> >> >>> a half-bridge with CM control and series inductors on the output. >> >> >> >> >> >> Fortunately the half-bridge's worst-case condition is zero load, so it >> >> >> isn't too hard to control. ... >> >> > >> >> > >> >> > They work ok with zero load if you can tolerate the voltage increase. >> >> > That would be determined by the turns ratio. If it has to be 45.000V >> >> > that's a different story. Then you'll have to let it go into a skipping >> >> > mode. >> >> > >> >> > >> >> >> ... Of course the ungapped ferrite has a pretty >> >> >> frightening tempco of mu (it goes down by half when you hit it with cold >> >> >> spray), so some sort of current control is going to be required, for sure. >> >> >> >> >> > >> >> > Yes, needs current control but that's standard even in sub-Dollar chips. >> >> >> >> I may just make a boost with a two-winding toroid, and return the other >> >> end of the second winding to the raw -15V supply. (A lot of the Murata >> >> toroids are actually dual-winding, but they expect you to wire them in >> >> parallel to get the current handling capacity.) >> > >they generally spec them both for series and parallel ;) > >how about a capacitive coupled boost? > >something like this: >http://obsoletetechnology.wordpress.com/projects/studio-electronics/dc-dc-bipolar-power-supply-for-effect-pedals/ > >-LasseThat's like my "bubble" (boost-doubler) converter that I posted a few days ago, except that I didn't use an integrated switching controller and tweaked some things. Rob described it as "a sepic without the second winding." -- John Larkin Highland Technology Inc www.highlandtechnology.com jlarkin at highlandtechnology dot com Precision electronic instrumentation Picosecond-resolution Digital Delay and Pulse generators Custom timing and laser controllers Photonics and fiberoptic TTL data links VME analog, thermocouple, LVDT, synchro, tachometer Multichannel arbitrary waveform generators
Reply by ●July 17, 20132013-07-17
George Herold wrote:> On Wednesday, July 17, 2013 12:25:06 PM UTC-4, Joerg wrote: >> Klaus Kragelund wrote:[...]>> >>> For the flyback it needs to gapped. Non-gapped flybacks is asking >>> for trouble >> >> >> It's going to result in *KABLAM* :-) >> >> >> >> But since Phil's stuff is all noise-critical he probably should not >> use >> >> a flyback architecture. Gapped cores spew a lot of EMI around. > > Do they make gapped toroids? You could make something with > symmetrical gaps one on each side, or maybe a bunch of stripes...(I > guess that's hard to make.) >Yup: http://www.cmi-ferrite.com/news/Articles/microgap.pdf But most of the time it's done in powder fashion which results in a "distributed gap". Comes with its own sorts of issues, such as the fact that the resin used as a binder can deteriorate over time. Especially when running them a bit hot. -- Regards, Joerg http://www.analogconsultants.com/
