I've created all kinds of HV circuits in the last 5 years,** and each one needs an external HV supply. So I really need some simple HV-supply circuits to plop onto these PCB's to make them self-sufficient. Today's circuit is a basic flyback, to convert a 12V,1A wallwort source to 500V max, at 5-10W load. I used a UC3843B running at 100kHz driving a MOSFET switching a 47uH inductor (dI = dt V/L = 2A in 8us). This can provide up to 100uJ each cycle, to charge a say 2000uF storage cap to 400V in 16 seconds. I settled on an ES1J fast-recovery output diode (after rejecting an MRA4007), After rejecting an STP7NB60 as too lossy, I experimented and picked an Infineon IPP65R225C7 superjunction MOSFET: Rds(on) = 0.2 ohms (0.5 ohms max at Tj = 150C), and Coss = 14pF at 400V (20pF at 100V). Nice. But the flyback inductor was an issue. See photo https://www.dropbox.com/s/cwo10wm5upehvrw/PWR-741_flyback_350V-inductors_small.jpg?dl=0 I started with an older 1A part from the drawer: a Mouser 47uH, 0.1 ohms, rated 1.3A, see photo, lower right. With this part my circuit had 2.2W of wasted power at 5W out. Heating was bad, the inductor melted a soldered lead and disconnected itself from the circit! Whew, ugly! I needed a beefier part. Enter a Coilcraft PCV-0-473-05L, photo lower left. Those fat wires, this had to be the answer. Rated 0.035 ohms and 6A. But it increased my wasted power to 3.7W, double ouch! Giving up on my commercial inventory, I made an inductor using an RM8 bobbin and core. (See photo, upper part, mounted on the PCB.) I selected a large gap core, A_L = 100nH/t^2. N = sqrt(L / A_L) = 22 turns. With #20 wire, Rdc=0.036 ohms. I measured Q=39 at 100kHz, esr = 2pi f L/Q = 0.75 ohms. Q=80 at 1MHz. Looking pretty good, ignoring coreloss at full power (the ckt uses 40mW at no load). Yeah! With the RM8 inductor, total losses dropped to 1.1 watts. Note, the MOSFET, sense resistor, diode and capacitor share in the loss generation, so I don't know the inductor loss. Everything runs cool, which wasn't the case at the beginning. ** more on that later ... -- Thanks, - Win
inductor for HV flyback supply
Started by ●August 5, 2016
Reply by ●August 5, 20162016-08-05
On Friday, August 5, 2016 at 1:07:26 PM UTC+10, Winfield Hill wrote:> I've created all kinds of HV circuits in the last > 5 years,** and each one needs an external HV supply. > So I really need some simple HV-supply circuits to > plop onto these PCB's to make them self-sufficient. > > Today's circuit is a basic flyback, to convert a > 12V,1A wallwort source to 500V max, at 5-10W load. > > I used a UC3843B running at 100kHz driving a MOSFET > switching a 47uH inductor (dI = dt V/L = 2A in 8us). > This can provide up to 100uJ each cycle, to charge > a say 2000uF storage cap to 400V in 16 seconds. > > I settled on an ES1J fast-recovery output diode > (after rejecting an MRA4007), After rejecting > an STP7NB60 as too lossy, I experimented and > picked an Infineon IPP65R225C7 superjunction > MOSFET: Rds(on) = 0.2 ohms (0.5 ohms max at Tj > = 150C), and Coss = 14pF at 400V (20pF at 100V). > Nice. > > But the flyback inductor was an issue. See photo > https://www.dropbox.com/s/cwo10wm5upehvrw/PWR-741_flyback_350V-inductors_small.jpg?dl=0 > > I started with an older 1A part from the drawer: > a Mouser 47uH, 0.1 ohms, rated 1.3A, see photo, > lower right. With this part my circuit had 2.2W > of wasted power at 5W out. Heating was bad, the > inductor melted a soldered lead and disconnected > itself from the circit! Whew, ugly! I needed a > beefier part. Enter a Coilcraft PCV-0-473-05L, > photo lower left. Those fat wires, this had to > be the answer. Rated 0.035 ohms and 6A. But it > increased my wasted power to 3.7W, double ouch! > > Giving up on my commercial inventory, I made > an inductor using an RM8 bobbin and core. > (See photo, upper part, mounted on the PCB.) > > I selected a large gap core, A_L = 100nH/t^2. > N = sqrt(L / A_L) = 22 turns. With #20 wire, > Rdc=0.036 ohms. I measured Q=39 at 100kHz, > esr = 2pi f L/Q = 0.75 ohms. Q=80 at 1MHz. > Looking pretty good, ignoring coreloss at > full power (the ckt uses 40mW at no load). > > Yeah! With the RM8 inductor, total losses > dropped to 1.1 watts. Note, the MOSFET, > sense resistor, diode and capacitor share > in the loss generation, so I don't know > the inductor loss. Everything runs cool, > which wasn't the case at the beginning. > > ** more on that later ...Why futz around with flyback supplies when you can use a Baxandall Class-D oscillator? Peter Baxandall invented it to deal with the problem of getting a kV or so for photomultiplier supplies, and Jim Williams used it to generate even higher voltages for the cold cathode back-lights that used to be popular in lap-top computers. http://sophia-elektronica.com/Baxandall_parallel-resonant_Class-D_oscillator1.htm http://sophia-elektronica.com/0344_001_Baxandal.pdf It's two pot cores - transformer and inductor - rather than just one, but it's a lot tidier, and you should be able to get away with smaller pot cores. -- Bill Sloman, Sydney
Reply by ●August 5, 20162016-08-05
Winfield Hill wrote:> I've created all kinds of HV circuits in the last > 5 years,** and each one needs an external HV supply. > So I really need some simple HV-supply circuits to > plop onto these PCB's to make them self-sufficient. > > Today's circuit is a basic flyback, to convert a > 12V,1A wallwort source to 500V max, at 5-10W load. > > I used a UC3843B running at 100kHz driving a MOSFET > switching a 47uH inductor (dI = dt V/L = 2A in 8us). > This can provide up to 100uJ each cycle, to charge > a say 2000uF storage cap to 400V in 16 seconds. > > I settled on an ES1J fast-recovery output diode > (after rejecting an MRA4007), After rejecting > an STP7NB60 as too lossy, I experimented and > picked an Infineon IPP65R225C7 superjunction > MOSFET: Rds(on) = 0.2 ohms (0.5 ohms max at Tj > = 150C), and Coss = 14pF at 400V (20pF at 100V). > Nice. > > But the flyback inductor was an issue. See photo > https://www.dropbox.com/s/cwo10wm5upehvrw/PWR-741_flyback_350V-inductors_small.jpg?dl=0 > > I started with an older 1A part from the drawer: > a Mouser 47uH, 0.1 ohms, rated 1.3A, see photo, > lower right. With this part my circuit had 2.2W > of wasted power at 5W out. Heating was bad, the > inductor melted a soldered lead and disconnected > itself from the circit! Whew, ugly! I needed a > beefier part. Enter a Coilcraft PCV-0-473-05L, > photo lower left. Those fat wires, this had to > be the answer. Rated 0.035 ohms and 6A. But it > increased my wasted power to 3.7W, double ouch! > > Giving up on my commercial inventory, I made > an inductor using an RM8 bobbin and core. > (See photo, upper part, mounted on the PCB.) > > I selected a large gap core, A_L = 100nH/t^2. > N = sqrt(L / A_L) = 22 turns. With #20 wire, > Rdc=0.036 ohms. I measured Q=39 at 100kHz, > esr = 2pi f L/Q = 0.75 ohms. Q=80 at 1MHz. > Looking pretty good, ignoring coreloss at > full power (the ckt uses 40mW at no load). > > Yeah! With the RM8 inductor, total losses > dropped to 1.1 watts. Note, the MOSFET, > sense resistor, diode and capacitor share > in the loss generation, so I don't know > the inductor loss. Everything runs cool, > which wasn't the case at the beginning. > > ** more on that later ... > >No photo; took a long time to "load"...
Reply by ●August 5, 20162016-08-05
Single stage..? Yeah, that's gonna cost you dearly if you want any kind of efficiency, or compact size. And cost, but I'm guessing most of your circuits are lab use, so that one's not too important, at least. Have you considered a mains inverter transformer? They're largely around $5, and about as big as the RM core, or the multilayer rod inductor, but... they are much easier to use, at least. Figure that, since they're made for 2.5kV reinforced isolation, and 320VDC primary supplies, there should be no problem getting 500V or more (without even adding a doubler). The one I made is resonant (sort of, but not really sinusoidal.. so bug off, Bill :^) ), and delivers up to 2kV and 10mA. Here's the circuit: http://seventransistorlabs.com/Images/HVPower1.png and a picture of the build: http://seventransistorlabs.com/Images/HVPower2.jpg There's very little winding area used on that ~200W capacity transformer (EE33 size). I wound it poorly on purpose, but it still had less leakage than I was wanting/expecting. (As a result, the operating frequency swings over quite a range, between unloaded and short-circuit conditions.) It would be quite reasonable to get 50, maybe 100W from a transformer this size. Of course, UC384x are quite capable, and a good choice. Just beware of leakage inductance and parasitic capacitance, if you look into transformers at all. :) Tim -- Seven Transistor Labs, LLC Electrical Engineering Consultation and Contract Design Website: http://seventransistorlabs.com "Winfield Hill" <hill@rowland.harvard.edu> wrote in message news:no0vsp0235e@drn.newsguy.com...> > I've created all kinds of HV circuits in the last > 5 years,** and each one needs an external HV supply. > So I really need some simple HV-supply circuits to > plop onto these PCB's to make them self-sufficient. > > Today's circuit is a basic flyback, to convert a > 12V,1A wallwort source to 500V max, at 5-10W load. > > I used a UC3843B running at 100kHz driving a MOSFET > switching a 47uH inductor (dI = dt V/L = 2A in 8us). > This can provide up to 100uJ each cycle, to charge > a say 2000uF storage cap to 400V in 16 seconds. > > I settled on an ES1J fast-recovery output diode > (after rejecting an MRA4007), After rejecting > an STP7NB60 as too lossy, I experimented and > picked an Infineon IPP65R225C7 superjunction > MOSFET: Rds(on) = 0.2 ohms (0.5 ohms max at Tj > = 150C), and Coss = 14pF at 400V (20pF at 100V). > Nice. > > But the flyback inductor was an issue. See photo > https://www.dropbox.com/s/cwo10wm5upehvrw/PWR-741_flyback_350V-inductors_small.jpg?dl=0 > > I started with an older 1A part from the drawer: > a Mouser 47uH, 0.1 ohms, rated 1.3A, see photo, > lower right. With this part my circuit had 2.2W > of wasted power at 5W out. Heating was bad, the > inductor melted a soldered lead and disconnected > itself from the circit! Whew, ugly! I needed a > beefier part. Enter a Coilcraft PCV-0-473-05L, > photo lower left. Those fat wires, this had to > be the answer. Rated 0.035 ohms and 6A. But it > increased my wasted power to 3.7W, double ouch! > > Giving up on my commercial inventory, I made > an inductor using an RM8 bobbin and core. > (See photo, upper part, mounted on the PCB.) > > I selected a large gap core, A_L = 100nH/t^2. > N = sqrt(L / A_L) = 22 turns. With #20 wire, > Rdc=0.036 ohms. I measured Q=39 at 100kHz, > esr = 2pi f L/Q = 0.75 ohms. Q=80 at 1MHz. > Looking pretty good, ignoring coreloss at > full power (the ckt uses 40mW at no load). > > Yeah! With the RM8 inductor, total losses > dropped to 1.1 watts. Note, the MOSFET, > sense resistor, diode and capacitor share > in the loss generation, so I don't know > the inductor loss. Everything runs cool, > which wasn't the case at the beginning. > > ** more on that later ... > > > -- > Thanks, > - Win
Reply by ●August 5, 20162016-08-05
On Friday, August 5, 2016 at 5:21:10 PM UTC+10, Tim Williams wrote:> Single stage..? > > Yeah, that's gonna cost you dearly if you want any kind of efficiency, or > compact size. And cost, but I'm guessing most of your circuits are lab use, > so that one's not too important, at least. > > Have you considered a mains inverter transformer? They're largely around > $5, and about as big as the RM core, or the multilayer rod inductor, but... > they are much easier to use, at least. > > Figure that, since they're made for 2.5kV reinforced isolation, and 320VDC > primary supplies, there should be no problem getting 500V or more (without > even adding a doubler). > > > The one I made is resonant (sort of, but not really sinusoidal.. so bug off, > Bill :^) ), and delivers up to 2kV and 10mA. Here's the circuit: > http://seventransistorlabs.com/Images/HVPower1.png > and a picture of the build: > http://seventransistorlabs.com/Images/HVPower2.jpg > > There's very little winding area used on that ~200W capacity transformer > (EE33 size). I wound it poorly on purpose, but it still had less leakage > than I was wanting/expecting. (As a result, the operating frequency swings > over quite a range, between unloaded and short-circuit conditions.) It > would be quite reasonable to get 50, maybe 100W from a transformer this > size. > > Of course, UC384x are quite capable, and a good choice. Just beware of > leakage inductance and parasitic capacitance, if you look into transformers > at all. :)Parastic - parallel - capacitance in high turns ratio transformers can be a real problem. If you can bank-wind the secondary, you can cut it down quite a bit, but multi-section formers for even big RM cores seem to have become hard to get hold of. The charm of the Baxandal class-D oscillator is that handles the parallel capacitance more elegantly than the Royer inverter, but it does set an upper limit on the oscillation frequency. Leakage inductance is also less of a problem in the Baxandall set-up - you aren't trying to change the currents through the transformer as fast as you need to in a Royer oscillator. -- Bill Sloman, Sydney
Reply by ●August 5, 20162016-08-05
Robert Baer wrote...> > Winfield Hill wrote: >> I've created all kinds of HV circuits in the last >> 5 years,** and each one needs an external HV supply. >> So I really need some simple HV-supply circuits to >> plop onto these PCB's to make them self-sufficient. >> >> Today's circuit is a basic flyback, to convert a >> 12V,1A wallwort source to 500V max, at 5-10W load. >> [ snip ] >> >> But the flyback inductor was an issue. See photo >>https://www.dropbox.com/s/cwo10wm5upehvrw/PWR-741_flyback_350V-inductors_small.jpg?dl=0 > > No photo; took a long time to "load"...Hmm, I clicked the dropbox link in a second browser window, and it popped right up. The link is spread across two lines, maybe your newsreader / browser only paid attention to the first line, not good. Try this one, with a shortened filename: https://www.dropbox.com/s/cwo10wm5upehvrw/PWR_s.jpg -- Thanks, - Win
Reply by ●August 5, 20162016-08-05
bill.sloman@ieee.org wrote...> >On Friday, August 5, 2016 at 1:07:26 PM UTC+10, Winfield Hill wrote: >> I've created all kinds of HV circuits in the last >> 5 years,** and each one needs an external HV supply. >> So I really need some simple HV-supply circuits to >> plop onto these PCB's to make them self-sufficient. >> >> Today's circuit is a basic flyback, to convert a >> 12V,1A wallwort source to 500V max, at 5-10W load. >> >> I used a UC3843B running at 100kHz ...https://www.dropbox.com/s/cwo10wm5upehvrw/PWR_s.jpg> Why futz around with flyback supplies when you can > use a Baxandall Class-D oscillator? > > Peter Baxandall invented it to deal with the problem > of getting a kV or so for photomultiplier supplies, > and Jim Williams used it to generate even higher > voltages for the cold cathode back-lights that used > to be popular in lap-top computers. > >http://sophia-elektronica.com/Baxandall_parallel-resonant_Class-D_oscillator1.htm > > http://sophia-elektronica.com/0344_001_Baxandal.pdf > > It's two pot cores - transformer and inductor - > rather than just one, but it's a lot tidier, you > should be able to get away with smaller pot cores.Thanks for the suggestion Bill, I know you love that circuit, and resonance is good. Indeed for higher power and voltages it'd be the right idea. But the uc3843 and other parts are small, SO-8, etc., mounted on the underside of the PCB, and in the beginning I had hoped to use a single simple commercial off-the-shelf inductor, and avoid any custom magnetics. May still do so. Anyway, even with having to wind 22 turns on an RM8, the result is still simple, small and cheap. BTW, some of your web article links are broken. -- Thanks, - Win
Reply by ●August 5, 20162016-08-05
On Friday, August 5, 2016 at 9:05:33 PM UTC+10, Winfield Hill wrote:> bill.sloman@ieee.org wrote... > > > >On Friday, August 5, 2016 at 1:07:26 PM UTC+10, Winfield Hill wrote: > >> I've created all kinds of HV circuits in the last > >> 5 years,** and each one needs an external HV supply. > >> So I really need some simple HV-supply circuits to > >> plop onto these PCB's to make them self-sufficient. > >> > >> Today's circuit is a basic flyback, to convert a > >> 12V,1A wallwort source to 500V max, at 5-10W load. > >> > >> I used a UC3843B running at 100kHz ... > > https://www.dropbox.com/s/cwo10wm5upehvrw/PWR_s.jpg > > > Why futz around with flyback supplies when you can > > use a Baxandall Class-D oscillator? > > > > Peter Baxandall invented it to deal with the problem > > of getting a kV or so for photomultiplier supplies, > > and Jim Williams used it to generate even higher > > voltages for the cold cathode back-lights that used > > to be popular in lap-top computers. > > > >http://sophia-elektronica.com/Baxandall_parallel-resonant_Class-D_oscillator1.htm > > > > http://sophia-elektronica.com/0344_001_Baxandal.pdf > > > > It's two pot cores - transformer and inductor - > > rather than just one, but it's a lot tidier, you > > should be able to get away with smaller pot cores. > > Thanks for the suggestion Bill, I know you love > that circuit, and resonance is good. Indeed for > higher power and voltages it'd be the right idea. > > But the uc3843 and other parts are small, SO-8, > etc., mounted on the underside of the PCB, and > in the beginning I had hoped to use a single > simple commercial off-the-shelf inductor, and > avoid any custom magnetics. May still do so.Too many people make a habit of avoiding custom magnetics. The Rowland Institute ought to have a simple coil winder - there are cheap ones available from China if it doesn't - and you don't need much of a range of RM cores, formers, mylar tape and enamelled wire to do useful stuff. Printed windings are neater, but you've got to talk to a specialist printed circuit shop to get useful copper fill factors. I talked to one in the Netherlands, but didn't get to the point of having anything made - that would have been for a better electric piano keyboard, but I couldn't get the Linux mechanical drafting program to do anything useful, and tossed it all into the too-hard basket.> Anyway, even with having to wind 22 turns on an > RM8, the result is still simple, small and cheap. > > BTW, some of your web article links are broken.That's been on the to-do list for a while. I seem to be coming out of jet-lag today, and my wife won't be back until Monday, so I might get it done over the weekend. I've had a some help with the Baxandall article and it should look a little better in the next few hours. -- Bill Sloman, Sydney
Reply by ●August 5, 20162016-08-05
bill.sloman@ieee.org wrote...> > Too many people make a habit of avoiding custom magnetics. > The Rowland Institute ought to have a simple coil winder - > there are cheap ones available from China if it doesn't - > and you don't need much of a range of RM cores, formers, > mylar tape and enamelled wire to do useful stuff.We have a winder, but need an fixture for small bobbins. I have the full range of materials, including my favorite Kapton tape, in various widths. It's the painful magnet- wire enamel removal and soldering to tiny bobbin pins that slows me down. That'd be an even worse issue with extra windings. But I heard a rumor about a special magnet wire with disappearing enamel at soldering temps.>> BTW, some of your web article links are broken. > > That's been on the to-do list for a while. ... > I've had a some help with the Baxandall article and > it should look a little better in the next few hours.It'd be nice if you'd add a downloadable pdf version. -- Thanks, - Win
Reply by ●August 5, 20162016-08-05
On Friday, August 5, 2016 at 10:54:06 PM UTC+10, Winfield Hill wrote:> bill.sloman@ieee.org wrote... > > > > Too many people make a habit of avoiding custom magnetics. > > The Rowland Institute ought to have a simple coil winder - > > there are cheap ones available from China if it doesn't - > > and you don't need much of a range of RM cores, formers, > > mylar tape and enamelled wire to do useful stuff. > > We have a winder, but need an fixture for small bobbins. > I have the full range of materials, including my favorite > Kapton tape, in various widths. It's the painful magnet- > wire enamel removal and soldering to tiny bobbin pins > that slows me down. That'd be an even worse issue with > extra windings. But I heard a rumor about a special > magnet wire with disappearing enamel at soldering temps.The professional coil winders at Cambridge Instruments had "solder pots" - little iron pots of molten solder with a tight-fitting lid with a very small - wire-sized - hole in the top. You poked the end of the enamel wire into the hole and it came out tinned. I was decidedly envious, since I had to clean off the enamel with a sharp knife or a chunk of emery paper. If I'd been closer to their benches, I'd have walked my prototypes across to one of the pots, but scraping was quicker.> >> BTW, some of your web article links are broken. > > > > That's been on the to-do list for a while. ... > > I've had a some help with the Baxandall article and > > it should look a little better in the next few hours. > > It'd be nice if you'd add a downloadable pdf version.That's exactly what it is. On my machine Firebox calls up Adobe Acrobat to display it and that offers you a down-load option. When I checked just now, it picked up the new - tidier, if marginally bigger - version. -- Bill Sloman, Sydney
