Forums

Bifilar Wound Balun Transformer

Started by rickman November 3, 2012
On 11/4/2012 3:49 PM, John Larkin wrote:
> On Sat, 03 Nov 2012 23:25:46 -0400, rickman<gnuarm@gmail.com> wrote: > >> On 11/3/2012 9:42 PM, John Larkin wrote: >>> On Sat, 3 Nov 2012 20:40:37 -0500, "Tim Williams" >>> <tmoranwms@charter.net> wrote: >>> >>>> "John Larkin"<jjlarkin@highNOTlandTHIStechnologyPART.com> wrote in >>>> message news:ffcb985qk94e0cf265odu97o95d90sp1bh@4ax.com... >>>>> We do exactly that in a bunch of products, namely use the shield as a >>>>> primary winding and the inner as the fully isolated secondary of a >>>>> transformer. We do 1:1 and 1:2 (voltage step up) at levels from 5 >>>>> volts to over 100. >>>>> >>>>> https://dl.dropbox.com/u/53724080/Circuits/Xfmrs.JPG >>>>> >>>>> This makes a transformer with very low leakage inductance, so we get >>>>> sub-ns rise times into a 50 ohm load. >>>> >>>> Except that, as I said, the leakage is not particularly low. One gets >>>> better performance in that regard from, say, copper foil pairs (which, >>>> ultimately, is still doing the same thing, but with a low impedance >>>> symmetrical stripline, not 50 ohm coax). Which is often done in power >>>> circuitry. But "very low leakage" is not what you're going for, so it's >>>> best not to claim that's what you're doing. >>>> >>>> Tim >>> >>> But it works. >> >> See, this is the sort of stuff that, if I were a potential customer, >> would turn me off to doing business with you. Geeze, if I am talking to >> someone about what is going on in a system and they say to me, "but it >> works", I would think they didn't understand it at all. > > I'm an engineer. I don't need to understand it, I only need to make it > work.
Amazing. There are times when a line is drawn and a designer says, "I understand this well enough", but the way you say it comes off like an amateur. I have spent a lot of time in my career fixing systems designed by people who obviously "only needed to make it work", but then it stopped working for some unknown reason.
> If a deep theoretical understanding of transmission-line > transformers is helpful, I might use it. But if an hour of > instinct-driven experimenting works, I'll go with that. My > mosfet-transmission-line output stage, which we've used thousands of > times, took about an hour of experimenting to design. > > Some of the stuff that we do is so complex that closed-form solutions > are impossible, and serious simulation would cost way too much time > and money. > > In the electronic design business, we seldom really understand what > we're doing, at the first-principles level.
That is scary. I find a lot of people like that though. I just thought they were posers. I've never heard any of them brag about it. > We usually work further up
> the abstraction stack. We usually buy parts, read data sheets, and > connect them up. It's actually unusual to *make* a part. [1]
Yeah, that's what everyone does, but when they connect those parts, typically they understand everything about them and how to connect them that they need to.
>> Do you not see how your posts make you look? > > I posted pics of actual isolating transformers made with micro-coax. > And some nice sub-ns-risetime 100 volt pulses that were pumped through > similar transformers. Why would a customer be turned off by something > that works? > > A sub-ns rise time into a 50 ohm load implies equivalent leakage > inductance in the 10s of nH.
I'm talking about the statements you make that sound like they are from someone with no level of understanding. I shouldn't be posting about this. It is clear that you understand completely what you are saying and I expect you understand how it makes you appear. So sorry for bothering you with this. Rick
On Sun, 04 Nov 2012 16:37:19 -0500, the renowned Jamie
<jamie_ka1lpa_not_valid_after_ka1lpa_@charter.net> wrote:

>John Larkin wrote: > >> On Sat, 03 Nov 2012 23:25:46 -0400, rickman <gnuarm@gmail.com> wrote: >> >> >>>On 11/3/2012 9:42 PM, John Larkin wrote: >>> >>>>On Sat, 3 Nov 2012 20:40:37 -0500, "Tim Williams" >>>><tmoranwms@charter.net> wrote: >>>> >>>> >>>>>"John Larkin"<jjlarkin@highNOTlandTHIStechnologyPART.com> wrote in >>>>>message news:ffcb985qk94e0cf265odu97o95d90sp1bh@4ax.com... >>>>> >>>>>>We do exactly that in a bunch of products, namely use the shield as a >>>>>>primary winding and the inner as the fully isolated secondary of a >>>>>>transformer. We do 1:1 and 1:2 (voltage step up) at levels from 5 >>>>>>volts to over 100. >>>>>> >>>>>>https://dl.dropbox.com/u/53724080/Circuits/Xfmrs.JPG >>>>>> >>>>>>This makes a transformer with very low leakage inductance, so we get >>>>>>sub-ns rise times into a 50 ohm load. >>>>> >>>>>Except that, as I said, the leakage is not particularly low. One gets >>>>>better performance in that regard from, say, copper foil pairs (which, >>>>>ultimately, is still doing the same thing, but with a low impedance >>>>>symmetrical stripline, not 50 ohm coax). Which is often done in power >>>>>circuitry. But "very low leakage" is not what you're going for, so it's >>>>>best not to claim that's what you're doing. >>>>> >>>>>Tim >>>> >>>>But it works. >>> >>>See, this is the sort of stuff that, if I were a potential customer, >>>would turn me off to doing business with you. Geeze, if I am talking to >>>someone about what is going on in a system and they say to me, "but it >>>works", I would think they didn't understand it at all. >> >> >> I'm an engineer. I don't need to understand it, I only need to make it >> work. If a deep theoretical understanding of transmission-line >> transformers is helpful, I might use it. But if an hour of >> instinct-driven experimenting works, I'll go with that. My >> mosfet-transmission-line output stage, which we've used thousands of >> times, took about an hour of experimenting to design. >> >> Some of the stuff that we do is so complex that closed-form solutions >> are impossible, and serious simulation would cost way too much time >> and money. >> >> In the electronic design business, we seldom really understand what >> we're doing, at the first-principles level. We usually work further up >> the abstraction stack. We usually buy parts, read data sheets, and >> connect them up. It's actually unusual to *make* a part. [1] >> >> >>>Do you not see how your posts make you look? >> >> >> I posted pics of actual isolating transformers made with micro-coax. >> And some nice sub-ns-risetime 100 volt pulses that were pumped through >> similar transformers. Why would a customer be turned off by something >> that works? >> >> A sub-ns rise time into a 50 ohm load implies equivalent leakage >> inductance in the 10s of nH. >> >> [1] invite interesting tales of actually making components. >> >> > At least you actually do something, not like a good many here that >would like to make people think otherwise. > > I spend more time at actually experimenting with what works the best >instead of fighting with PC software that only gets it close but not >good enough. > > I just love those that talk shit and most likely hardly even touch a >piece of equipment. When they do I am sure they're all thumbs and >fingers with it and most likely end up getting some one else to do it >for them and take all the credit for it.
I had a guy tell me a cheap scope (he always buys the most expensive equipment) I got for him was broken-- turns out the brightness control (or whatever you call it on a digital scope) was turned down. Same guy claimed an expensive SRS bridge with 0.1% accuracy was giving 10% error on a reading-- turned out he was using 100Hz to measure a tens-of-pF cap and it was performing well within spec according to the manual. He has written peer-reviewed papers on these things.. sad.
> > Those guilty of this need not to step forward, I already know who >most of you are. > >Jamie
It's a rare person that can get all the theory right and the practice right- they deserve to be well-rewarded. Best regards, Spehro Pefhany -- "it's the network..." "The Journey is the reward" speff@interlog.com Info for manufacturers: http://www.trexon.com Embedded software/hardware/analog Info for designers: http://www.speff.com
Jamie wrote:
> > Those that don't know shit, should shit elsewhere ! Get it?
Then you must have a hell of a time getting to Pluto to shit.
Jamie wrote:
> > Those guilty of this need not to step forward, I already know who > most of you are.
What you know would fit in a thimble, with a million lawyer hearts.
On Sun, 04 Nov 2012 16:36:32 -0500, rickman <gnuarm@gmail.com> wrote:

>On 11/4/2012 3:49 PM, John Larkin wrote: >> On Sat, 03 Nov 2012 23:25:46 -0400, rickman<gnuarm@gmail.com> wrote: >> >>> On 11/3/2012 9:42 PM, John Larkin wrote: >>>> On Sat, 3 Nov 2012 20:40:37 -0500, "Tim Williams" >>>> <tmoranwms@charter.net> wrote: >>>> >>>>> "John Larkin"<jjlarkin@highNOTlandTHIStechnologyPART.com> wrote in >>>>> message news:ffcb985qk94e0cf265odu97o95d90sp1bh@4ax.com... >>>>>> We do exactly that in a bunch of products, namely use the shield as a >>>>>> primary winding and the inner as the fully isolated secondary of a >>>>>> transformer. We do 1:1 and 1:2 (voltage step up) at levels from 5 >>>>>> volts to over 100. >>>>>> >>>>>> https://dl.dropbox.com/u/53724080/Circuits/Xfmrs.JPG >>>>>> >>>>>> This makes a transformer with very low leakage inductance, so we get >>>>>> sub-ns rise times into a 50 ohm load. >>>>> >>>>> Except that, as I said, the leakage is not particularly low. One gets >>>>> better performance in that regard from, say, copper foil pairs (which, >>>>> ultimately, is still doing the same thing, but with a low impedance >>>>> symmetrical stripline, not 50 ohm coax). Which is often done in power >>>>> circuitry. But "very low leakage" is not what you're going for, so it's >>>>> best not to claim that's what you're doing. >>>>> >>>>> Tim >>>> >>>> But it works. >>> >>> See, this is the sort of stuff that, if I were a potential customer, >>> would turn me off to doing business with you. Geeze, if I am talking to >>> someone about what is going on in a system and they say to me, "but it >>> works", I would think they didn't understand it at all. >> >> I'm an engineer. I don't need to understand it, I only need to make it >> work. > >Amazing. There are times when a line is drawn and a designer says, "I >understand this well enough", but the way you say it comes off like an >amateur. I have spent a lot of time in my career fixing systems >designed by people who obviously "only needed to make it work", but then >it stopped working for some unknown reason.
How deep does your understanding go? Quantum mechanics? String theory? Do you do closed-form Maxwell's Equations on every circuit? Or full EM simulation? I bet you don't.
> > >> If a deep theoretical understanding of transmission-line >> transformers is helpful, I might use it. But if an hour of >> instinct-driven experimenting works, I'll go with that. My >> mosfet-transmission-line output stage, which we've used thousands of >> times, took about an hour of experimenting to design. >> >> Some of the stuff that we do is so complex that closed-form solutions >> are impossible, and serious simulation would cost way too much time >> and money. >> >> In the electronic design business, we seldom really understand what >> we're doing, at the first-principles level. > >That is scary. I find a lot of people like that though. I just thought >they were posers. I've never heard any of them brag about it.
I'm not bragging. I wish I had the tools to fully understand or simulate everything we do, down to the physics. Sometimes you do whatever works.
> > > > We usually work further up >> the abstraction stack. We usually buy parts, read data sheets, and >> connect them up. It's actually unusual to *make* a part. [1] > >Yeah, that's what everyone does, but when they connect those parts, >typically they understand everything about them and how to connect them >that they need to.
"That they need to." Yes. Understanding *everything* about an IC would be great, but we're not privvy to that information.
> > >>> Do you not see how your posts make you look? >> >> I posted pics of actual isolating transformers made with micro-coax. >> And some nice sub-ns-risetime 100 volt pulses that were pumped through >> similar transformers. Why would a customer be turned off by something >> that works? >> >> A sub-ns rise time into a 50 ohm load implies equivalent leakage >> inductance in the 10s of nH. > >I'm talking about the statements you make that sound like they are from >someone with no level of understanding. > >I shouldn't be posting about this. It is clear that you understand >completely what you are saying and I expect you understand how it makes >you appear. So sorry for bothering you with this.
The pragamatism issue is real, and there are serious risks from not fully understanding a system. But there we are, working with what we have. I can't see a lot of risk in designing a transmission-line transformer by experiment. It should be very reproducible, and it is. A little math can check for things like breakdown voltage, if that matters. -- 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
"John Larkin" <jjlarkin@highNOTlandTHIStechnologyPART.com> wrote in 
message news:u2rd98l8ie7ucnfqk2em03grjthvdmoo1q@4ax.com...
> How deep does your understanding go? Quantum mechanics?
As a physicist, I can affirm that. Others may vary.
> String theory?
Doesn't count -- even the string theorists don't understand the stuff. ;-)
> Do you do closed-form Maxwell's Equations on every circuit? Or full EM > simulation?
And you don't? I do on every single board I make. Not closed-form, but open-form approximation, qualitative accuracy. Implemented in wetware, too. Works very well.
> I'm not bragging. I wish I had the tools to fully understand or > simulate everything we do, down to the physics. Sometimes you do > whatever works.
Well, if you really wished, you'd buy the entire Ansoft suite and *do* it -- but I'm guessing that wish isn't as unconditional as it was phrased. In actuality, you don't care at all, and are more than happy enough guessing. Which again illustrates your inconsistent self-representation. <snip> <quoted from proceeding post>
> I posted pics of actual isolating transformers made with micro-coax. > And some nice sub-ns-risetime 100 volt pulses that were pumped through > similar transformers. Why would a customer be turned off by something >that works? > > A sub-ns rise time into a 50 ohm load implies equivalent leakage > inductance in the 10s of nH.
Are you aware that ~20nH is ~86mm of 50 ohm, 0.67c coax? Assuming the headers pictured are 0.1" centers, the cores are roughly T37 size ferrites, a bit thicker than usual. I get 14mm for the length of a single turn on a regular T37, so it might be closer to 18mm per turn, maybe 20mm with coax thickness. That's 60mm total length, or 14nH. The soldered connections and board traces have almost as much, depending on if there's a ground plane just out of sight or not. But by then it's not mutual, which is all the more reason it's not LL you're supposing about. Actual performance will show helical resonator action starting around 1GHz, which is what the under-hump on your leading edge comes from. And probably other nasties if you tested it with a ps generator rather than the "sub-ns" this particular device produces. Tim -- Deep Friar: a very philosophical monk. Website: http://webpages.charter.net/dawill/tmoranwms
On 11/4/2012 8:43 PM, Tim Williams wrote:
> "John Larkin"<jjlarkin@highNOTlandTHIStechnologyPART.com> wrote in > message news:u2rd98l8ie7ucnfqk2em03grjthvdmoo1q@4ax.com... >> How deep does your understanding go? Quantum mechanics? > > As a physicist, I can affirm that. Others may vary.
I learned a little in chemistry classes... very little.
>> String theory? > > Doesn't count -- even the string theorists don't understand the stuff. ;-)
My understanding is not that they don't understand it, it just doesn't predict anything different from existing quantum theory.
>> Do you do closed-form Maxwell's Equations on every circuit? Or full EM >> simulation? > > And you don't? > > I do on every single board I make. Not closed-form, but open-form > approximation, qualitative accuracy. Implemented in wetware, too. Works > very well.
Have you solved the Schr&#2013266166;dinger wave equation for any of your systems. Only then will I call you a real engineer ;)
> Are you aware that ~20nH is ~86mm of 50 ohm, 0.67c coax? > > Assuming the headers pictured are 0.1" centers, the cores are roughly T37 > size ferrites, a bit thicker than usual. I get 14mm for the length of a > single turn on a regular T37, so it might be closer to 18mm per turn, > maybe 20mm with coax thickness. That's 60mm total length, or 14nH. The > soldered connections and board traces have almost as much, depending on if > there's a ground plane just out of sight or not. But by then it's not > mutual, which is all the more reason it's not LL you're supposing about. > > Actual performance will show helical resonator action starting around > 1GHz, which is what the under-hump on your leading edge comes from. And > probably other nasties if you tested it with a ps generator rather than > the "sub-ns" this particular device produces.
Hey, want to help me design a LF shielded loop antenna from coax? It sounds like it would be right down your alley! I don't know nothing about birthin' no babies, Ms. Scarlett! But it looks like I'm going to have to learn... Rick
On Sun, 4 Nov 2012 19:43:52 -0600, "Tim Williams"
<tmoranwms@charter.net> wrote:

>"John Larkin" <jjlarkin@highNOTlandTHIStechnologyPART.com> wrote in >message news:u2rd98l8ie7ucnfqk2em03grjthvdmoo1q@4ax.com... >> How deep does your understanding go? Quantum mechanics? > >As a physicist, I can affirm that. Others may vary. > >> String theory? > >Doesn't count -- even the string theorists don't understand the stuff. ;-) > >> Do you do closed-form Maxwell's Equations on every circuit? Or full EM >> simulation? > >And you don't?
Use Maxwell's equations? No, I don't. And I rarely do EM sims; I did post about one such, here a while back, using ATLC2 to design a PCB stackup for an edge-launch SMA connector. It was tedious, and it worked. It was sort of cool... https://dl.dropbox.com/u/53724080/ATLC/Edge_alone_4.jpg https://dl.dropbox.com/u/53724080/ATLC/E-field.jpg
> >I do on every single board I make. Not closed-form, but open-form >approximation, qualitative accuracy. Implemented in wetware, too. Works >very well.
"Qualitative accuracy?" What does that mean? I use a couple of PC apps to calculate trace impedances, Appcad and TXLine. But you don't have to understand a lot to drive those.
> >> I'm not bragging. I wish I had the tools to fully understand or >> simulate everything we do, down to the physics. Sometimes you do >> whatever works. > >Well, if you really wished, you'd buy the entire Ansoft suite and *do* >it -- but I'm guessing that wish isn't as unconditional as it was phrased.
It's a matter of cost, learning curve, and need. So far, the tools that I've mentioned, and a few home-made programs, do what we need.
>In actuality, you don't care at all, and are more than happy enough >guessing. Which again illustrates your inconsistent self-representation.
I don't guess about things that matter... see above. But I don't think there's anything wrong with designing a gate driver/mosfet/transmission line transformer circuit using instinct and experiment. The edges I'm getting are about 20x faster than the numbers on the mosfet data sheet. So much for theory. If I need a pullup resistor or a bypass cap value, or a trace width for small DC currents, I usually guess.
> ><snip> ><quoted from proceeding post> >> I posted pics of actual isolating transformers made with micro-coax. >> And some nice sub-ns-risetime 100 volt pulses that were pumped through >> similar transformers. Why would a customer be turned off by something >>that works? >> >> A sub-ns rise time into a 50 ohm load implies equivalent leakage >> inductance in the 10s of nH. > >Are you aware that ~20nH is ~86mm of 50 ohm, 0.67c coax?
I calculate 6 nH/inch, pretty close. What I'm really aware of is that it works and we sell lots of them.
> >Assuming the headers pictured are 0.1" centers, the cores are roughly T37 >size ferrites, a bit thicker than usual. I get 14mm for the length of a >single turn on a regular T37, so it might be closer to 18mm per turn, >maybe 20mm with coax thickness. That's 60mm total length, or 14nH. The >soldered connections and board traces have almost as much, depending on if >there's a ground plane just out of sight or not. But by then it's not >mutual, which is all the more reason it's not LL you're supposing about. > >Actual performance will show helical resonator action starting around >1GHz, which is what the under-hump on your leading edge comes from. And >probably other nasties if you tested it with a ps generator rather than >the "sub-ns" this particular device produces.
That particular transformer in the photo was driven by a gaasfet. It worked fine, made really pretty isolated pulse outputs, about 500 ps edges. About 2000 of those transformers trigger most all the various gadgets at NIF. -- 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
"rickman" <gnuarm@gmail.com> wrote in message 
news:k777h0$knd$1@dont-email.me...
>>> String theory? >> >> Doesn't count -- even the string theorists don't understand the stuff. >> ;-) > > My understanding is not that they don't understand it, it just doesn't > predict anything different from existing quantum theory.
Notwithstanding Feynman's quote, "nobody really understands QM", that's more accurately the problem, as I also understand it.
> > >>> Do you do closed-form Maxwell's Equations on every circuit? Or full EM >>> simulation? >> >> And you don't? >> >> I do on every single board I make. Not closed-form, but open-form >> approximation, qualitative accuracy. Implemented in wetware, too. >> Works >> very well. > > Have you solved the Schr&#2013266166;dinger wave equation for any of your systems. > Only then will I call you a real engineer ;) > > >> Are you aware that ~20nH is ~86mm of 50 ohm, 0.67c coax? >> >> Assuming the headers pictured are 0.1" centers, the cores are roughly >> T37 >> size ferrites, a bit thicker than usual. I get 14mm for the length of >> a >> single turn on a regular T37, so it might be closer to 18mm per turn, >> maybe 20mm with coax thickness. That's 60mm total length, or 14nH. >> The >> soldered connections and board traces have almost as much, depending on >> if >> there's a ground plane just out of sight or not. But by then it's not >> mutual, which is all the more reason it's not LL you're supposing >> about. >> >> Actual performance will show helical resonator action starting around >> 1GHz, which is what the under-hump on your leading edge comes from. >> And >> probably other nasties if you tested it with a ps generator rather than >> the "sub-ns" this particular device produces. > > Hey, want to help me design a LF shielded loop antenna from coax?
Sure! My work is done: http://vk1od.net/antenna/shieldedloop/ Well, maybe not *my* work, but... helpful nonetheless. Lots of excellent analysis on his website. Tim
On 11/4/2012 11:52 PM, Tim Williams wrote:
> "rickman"<gnuarm@gmail.com> wrote in message > news:k777h0$knd$1@dont-email.me... >>>> String theory? >>> >>> Doesn't count -- even the string theorists don't understand the stuff. >>> ;-) >> >> My understanding is not that they don't understand it, it just doesn't >> predict anything different from existing quantum theory. > > Notwithstanding Feynman's quote, "nobody really understands QM", that's > more accurately the problem, as I also understand it. > >> >> >>>> Do you do closed-form Maxwell's Equations on every circuit? Or full EM >>>> simulation? >>> >>> And you don't? >>> >>> I do on every single board I make. Not closed-form, but open-form >>> approximation, qualitative accuracy. Implemented in wetware, too. >>> Works >>> very well. >> >> Have you solved the Schr&#2013266166;dinger wave equation for any of your systems. >> Only then will I call you a real engineer ;) >> >> >>> Are you aware that ~20nH is ~86mm of 50 ohm, 0.67c coax? >>> >>> Assuming the headers pictured are 0.1" centers, the cores are roughly >>> T37 >>> size ferrites, a bit thicker than usual. I get 14mm for the length of >>> a >>> single turn on a regular T37, so it might be closer to 18mm per turn, >>> maybe 20mm with coax thickness. That's 60mm total length, or 14nH. >>> The >>> soldered connections and board traces have almost as much, depending on >>> if >>> there's a ground plane just out of sight or not. But by then it's not >>> mutual, which is all the more reason it's not LL you're supposing >>> about. >>> >>> Actual performance will show helical resonator action starting around >>> 1GHz, which is what the under-hump on your leading edge comes from. >>> And >>> probably other nasties if you tested it with a ps generator rather than >>> the "sub-ns" this particular device produces. >> >> Hey, want to help me design a LF shielded loop antenna from coax? > > Sure! My work is done: > http://vk1od.net/antenna/shieldedloop/ > Well, maybe not *my* work, but... helpful nonetheless. Lots of excellent > analysis on his website. > > Tim
Certainly this is a good page, thanks. But that's not the same as designing one... or helping. Still, thanks. Rick