Forums

LTC Switcher CAD3 ?

Started by Robert Baer June 2, 2012
"John Larkin" <jjlarkin@highNOTlandTHIStechnologyPART.com> wrote in message 
news:2r7ns7t70t2icf6qut3ug6btce4jqdu7dd@4ax.com...
> On Sun, 3 Jun 2012 12:20:06 -0500, "Tim Williams" > <tmoranwms@charter.net> wrote: > >>"John Larkin" <jjlarkin@highNOTlandTHIStechnologyPART.com> wrote in >>message >>news:3r5ns7hvt6apkqu9l8023jh7gclchnq1mm@4ax.com... >>>>So make your own macro model and implement them, sheesh. >>> >>> Or buy a bunch of parts and test+breadboard. Works for me. >>> >>> http://dl.dropbox.com/u/53724080/Circuits/ESM/DCBB_2.JPG >> >>Last project I needed a model, I overestimated leakage and DCR >>intentionally, and made it work. Boards came in: much better performance, >>and no problems with ringing (note that lower leakage means higher ringing >>frequency and lower impedances, making them harder to dampen). >> >>That ISDN transformer looks a lot bulkier than Ethernet transformers. How >>much flux goes into one of those? >> >>Tim > > That's a 4-winding unit, 1:1:2:2, wired as a 5:1 step-up. I'm applying > 24 volts p-p, around 150 KHz, to a "1" winding and getting 120 p-p > out, into a diode doubler, to get 120 DC to drive a photodiode. I > don't know the flux density. It doesn't get hot. > > It's often easier to drop parts into a breadboard than to characterize > parts, create a model, and simulate. Measure temperatures with an IR > gadget instead of simulating core losses. > > Besides, a boy needs to solder once in a while. > > > -- > > 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
If you don't need isolation, you could do a 1:4 and add it to the top of the +24, autotransformer style. Might be slightly better efficiency. tm
On Sun, 3 Jun 2012 13:53:13 -0400, "tm" <No_one_home@white-house.gov>
wrote:

> >"John Larkin" <jjlarkin@highNOTlandTHIStechnologyPART.com> wrote in message >news:2r7ns7t70t2icf6qut3ug6btce4jqdu7dd@4ax.com... >> On Sun, 3 Jun 2012 12:20:06 -0500, "Tim Williams" >> <tmoranwms@charter.net> wrote: >> >>>"John Larkin" <jjlarkin@highNOTlandTHIStechnologyPART.com> wrote in >>>message >>>news:3r5ns7hvt6apkqu9l8023jh7gclchnq1mm@4ax.com... >>>>>So make your own macro model and implement them, sheesh. >>>> >>>> Or buy a bunch of parts and test+breadboard. Works for me. >>>> >>>> http://dl.dropbox.com/u/53724080/Circuits/ESM/DCBB_2.JPG >>> >>>Last project I needed a model, I overestimated leakage and DCR >>>intentionally, and made it work. Boards came in: much better performance, >>>and no problems with ringing (note that lower leakage means higher ringing >>>frequency and lower impedances, making them harder to dampen). >>> >>>That ISDN transformer looks a lot bulkier than Ethernet transformers. How >>>much flux goes into one of those? >>> >>>Tim >> >> That's a 4-winding unit, 1:1:2:2, wired as a 5:1 step-up. I'm applying >> 24 volts p-p, around 150 KHz, to a "1" winding and getting 120 p-p >> out, into a diode doubler, to get 120 DC to drive a photodiode. I >> don't know the flux density. It doesn't get hot. >> >> It's often easier to drop parts into a breadboard than to characterize >> parts, create a model, and simulate. Measure temperatures with an IR >> gadget instead of simulating core losses. >> >> Besides, a boy needs to solder once in a while. >> >> >> -- >> >> 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 > > >If you don't need isolation, you could do a 1:4 and add it to the top of the >+24, autotransformer style. Might be slightly better efficiency. > > >tm >
It needs isolation, or rather the customer thinks it does. Here's the circuit: http://dl.dropbox.com/u/53724080/Circuits/ESM/ESM_power_B.pdf and here's the board. http://dl.dropbox.com/u/53724080/PCBs/ESM_rev_B.jpg The dual-winding Coiltronics DRQ inductor/transformer is cute and cheap, but turned out to be fairly inefficient and marginal for the loads we might see. The Coilcraft toroid is a beast. That IR inverter driver chip is great. -- John Larkin Highland Technology, Inc jlarkin at highlandtechnology dot com http://www.highlandtechnology.com Precision electronic instrumentation Picosecond-resolution Digital Delay and Pulse generators Custom laser drivers and controllers Photonics and fiberoptic TTL data links VME thermocouple, LVDT, synchro acquisition and simulation
On Sun, 03 Jun 2012 05:11:52 -0700, Bill Sloman wrote:

> We've just measured a coupling of 0.98 in a gapped RM14 core - EPCOS > gapped it down from 6000uH/root turn for an ungapped pair to 630nH per > root turn by grinding back the centre pillar by 0.4mm,
I think that should be "per turn squared". -- "For a successful technology, reality must take precedence over public relations, for nature cannot be fooled." (Richard Feynman)
"John Larkin" <jjlarkin@highNOTlandTHIStechnologyPART.com> wrote in message 
news:2r7ns7t70t2icf6qut3ug6btce4jqdu7dd@4ax.com...
>> How >>much flux goes into one of those? > > That's a 4-winding unit, 1:1:2:2, wired as a 5:1 step-up. I'm applying > 24 volts p-p, around 150 KHz, to a "1" winding and getting 120 p-p > out, into a diode doubler, to get 120 DC to drive a photodiode. I > don't know the flux density. It doesn't get hot.
^ ^ ^ Yabbut, I didn't ask about flux density. That would require knowing what the core is, and you didn't buy the core. I asked about flux, which is what waveform you put on it. So you don't know what waveform you're applying? Did you test if it saturates and if so, at what frequency? If you don't have a clue, there's no telling when the power supply is going to poop out! That's bad news for your customers. Tim -- Deep Friar: a very philosophical monk. Website: http://webpages.charter.net/dawill/tmoranwms
On Jun 3, 9:02=A0pm, Fred Abse <excretatau...@invalid.invalid> wrote:
> On Sun, 03 Jun 2012 05:11:52 -0700,BillSlomanwrote: > > We've just measured a coupling of 0.98 in a gapped RM14 core - EPCOS > > gapped it down from 6000uH/root turn for an ungapped pair to 630nH per > > root turn by grinding back the centre pillar by 0.4mm, > > I think that should be "per turn squared".
Wrong. Inductance is proportional to the square of the number of turns, as long as they are closely coupled. http://en.wikipedia.org/wiki/Inductor -- Bill Sloman, Nijmegen
Bill Sloman wrote:
> > On Jun 3, 9:02 pm, Fred Abse <excretatau...@invalid.invalid> wrote: > > On Sun, 03 Jun 2012 05:11:52 -0700,BillSlomanwrote: > > > We've just measured a coupling of 0.98 in a gapped RM14 core - EPCOS > > > gapped it down from 6000uH/root turn for an ungapped pair to 630nH per > > > root turn by grinding back the centre pillar by 0.4mm, > > > > I think that should be "per turn squared". > > Wrong. Inductance is proportional to the square of the number of > turns, as long as they are closely coupled. > > http://en.wikipedia.org/wiki/Inductor
If L = alpha * N**2, then alpha has units of henries per turn squared. (Of course a turn isn't really a unit, just as a radian and a steradian aren't, but it sure isn't per square root turn.) 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 845-480-2058 hobbs at electrooptical dot net http://electrooptical.net
On Sun, 3 Jun 2012 14:40:27 -0500, "Tim Williams"
<tmoranwms@charter.net> wrote:

>"John Larkin" <jjlarkin@highNOTlandTHIStechnologyPART.com> wrote in message >news:2r7ns7t70t2icf6qut3ug6btce4jqdu7dd@4ax.com... >>> How >>>much flux goes into one of those? >> >> That's a 4-winding unit, 1:1:2:2, wired as a 5:1 step-up. I'm applying >> 24 volts p-p, around 150 KHz, to a "1" winding and getting 120 p-p >> out, into a diode doubler, to get 120 DC to drive a photodiode. I >> don't know the flux density. It doesn't get hot. > ^ ^ ^ >Yabbut, I didn't ask about flux density. That would require knowing what >the core is, and you didn't buy the core. I asked about flux, which is what >waveform you put on it. So you don't know what waveform you're applying? >Did you test if it saturates and if so, at what frequency? If you don't >have a clue, there's no telling when the power supply is going to poop out! >That's bad news for your customers. > >Tim
Don't be silly. Of course I test for basic stuff like saturation, and efficiency vs frequency vs load, and I thermal image everything to make sure nothing is unreasonably warm. I'm not used to referring to volt-seconds as "flux". That is, to me, flux density x core cross-sectional area, still a magnetic thing. I'd call, what I think you mean, excitation in volt-seconds. My breadboard has a 0.1 ohm resistor in series with the main transformer primary so I can see the excitation current. Added after the pic I posted. If I design a driver, how could I not know the waveform I'm making? You can see the scope probe in my picture. This part is all simple stuff; hours of work, not days. What's more interesting is the diffamp and the 12-bit, 250 MHz ADC just a few inches away from all these switchers. We're running just about 1 LSB RMS noise. -- John Larkin Highland Technology, Inc jlarkin at highlandtechnology dot com http://www.highlandtechnology.com Precision electronic instrumentation Picosecond-resolution Digital Delay and Pulse generators Custom laser drivers and controllers Photonics and fiberoptic TTL data links VME thermocouple, LVDT, synchro acquisition and simulation
"John Larkin" <jlarkin@highlandtechnology.com> wrote in message 
news:hhjns7l313vvbl367ais7l2cogqfq4t57v@4ax.com...
> I'm not used to referring to volt-seconds as "flux". That is, to me, > flux density x core cross-sectional area, still a magnetic thing. I'd > call, what I think you mean, excitation in volt-seconds.
Well that's what it is. The SI defines flux in webers == V.s (a derived unit), which is B * A_e for a core of cross section A_e, give or take a factor of turns N.
> If I design a driver, how could I not know the waveform I'm making? > You can see the scope probe in my picture.
Could've fooled me...
> This part is all simple stuff; hours of work, not days. What's more > interesting is the diffamp and the 12-bit, 250 MHz ADC just a few > inches away from all these switchers. We're running just about 1 LSB > RMS noise.
Well, I wouldn't care about the sample rate, as long as it, or the analog input bandwidth, includes the switching fundamental and maybe a couple harmonics. Once you get past 100 megs, you're going to see almost all the harmonics anyway. If ordinary noise abatement can't do it (only localized currents on the ground plane, heavy bypassing, ferrite beads, etc.), harmonics can be controlled (ala LT's AN70). I get the impression that you'd only go to quite that extreme if you need >16 bits accuracy; microvolts get tedious, but LT's good at it. Of course, if you needed 16 bits at 250MSa, you'll be paying serious money for it, and putting it on a red carpet at every turn. I'd claim in return that we're getting a few LSBs accuracy with our similar (albeit slower) ADC inside an induction heater power supply, but since we're using it to measure the active waveforms, crosstalk is in phase and doesn't interfere with the measurements. We have lots of board area, so the meat and potatoes are comfortably on the opposite side of the board. Induction is kind of a weird business, not quite as glacial as power transmission, but top designs in the field are still the ones with boards measured in square feet, loaded with analog and discrete digital circuitry. The newer designs additionally have microcontrollers, and only a few have even touched FPGAs. Tim -- Deep Friar: a very philosophical monk. Website: http://webpages.charter.net/dawill/tmoranwms
On Sun, 03 Jun 2012 03:24:10 -0800, Robert Baer
<robertbaer@localnet.com> wrote:

>Charles wrote: >> >> >> "Robert Baer" wrote in message >> news:xqydnWyQzY_6HlTSnZ2dnUVZ_g-dnZ2d@posted.localnet... >> >> Switcher cad?? >> NO transformers. >> NO saturable inductors. >> NO selection of ferrite core materials. >> NO selection of ferrite core sizes or shapes. >> WTF?? >> >> http://pkropik.com/storage/1202227692_sb_switchercad_1.pdf >> >> WTF don't you understand? > Yawn..NOT one word concerning what i mentioned. > 2 megs wasted, "covering" a quickie SPICE "review".
This link has a bit of info to help you maybe. Also don't forget to use google and the vast amount of information on the web about LTspice and how to model circuits with core models. It's still spice and isn't perfect and you are going to still tweak things to get it to run right. http://ltwiki.org/index.php5?title=Transformers
On Jun 3, 10:09=A0pm, Phil Hobbs
<pcdhSpamMeSensel...@electrooptical.net> wrote:
> Bill Sloman wrote: > > > On Jun 3, 9:02 pm, Fred Abse <excretatau...@invalid.invalid> wrote: > > > On Sun, 03 Jun 2012 05:11:52 -0700,BillSlomanwrote: > > > > We've just measured a coupling of 0.98 in a gapped RM14 core - EPCO=
S
> > > > gapped it down from 6000uH/root turn for an ungapped pair to 630nH =
per
> > > > root turn by grinding back the centre pillar by 0.4mm, > > > > I think that should be "per turn squared". > > > Wrong. Inductance is proportional to the square of the number of > > turns, as long as they are closely coupled. > > >http://en.wikipedia.org/wiki/Inductor > > If L =3D alpha * N**2, then alpha has units of henries per turn squared. > (Of course a turn isn't really a unit, just as a radian and a steradian > aren't, but it sure isn't per square root turn.)
Applying dimensional analysis to a manufacturer's design fudge factor is the sort of thing that physicists do. It's not a particularly useful activity. Reminding junior engineers at every possible opportunity that inductance is proportional to the square of the number of turns in a winding is an extremely useful activity, and saves loads of time at design reviews. -- Bill Sloman, Nijmegen