When you have to have an air gap then leakage is inevitable. It needs to be contained and orientation rarely helps.
-- Regards, Joerg http://www.analogconsultants.com/
:-)
-- "For a successful technology, reality must take precedence over public relations, for nature cannot be fooled." (Richard Feynman)
Whoops :-)
What I meant is the song "Highway to Hell" by AC/DC.
-- Regards, Joerg http://www.analogconsultants.com/
As long as ferrites don't have infinite permeability, there will always be some fringing field. Manufacturers gap RM cores by grinding only the centre , so the outer shell is still an effective - if necessarily imperfect - scr een.
Non-progressively wound toroidal coils don't have any external magnetic fie ld - a simple progressive winding produces an effective single turn in the plane of the toroid. The neater non-progressive winding schemes can cancel most of the external electric field as well, at the expense of higher self- capacitance.
The pot core is the topological dual of the toroidal core. Leaving slots in it for the connection to the core makes it less than perfect, and the RM c ore has bigger slots.
Orientation can help in some cases, if you think about it far enough in adv ance.
-- Bill Sloman, Sydney
Not with such simple cores like the SRF series. The gap is always up top where the epoxy goes -> cheap. There is another series that's a tad better but I'd have to dig that out. Used them in a sensitive app but costs more. Tin-can style shields are so cheap, why the worry?
There is a trick to muffle that via a loop-back half-turn. But toroids are mostly out of league for switchers, on account of the cost.
Pot cores are nice but mechanically a pain. They cost a lot, too.
Orientation only goes so far. Instead of disturbing the electronics over here you now disturb the electronics over yonder instead.
-- Regards, Joerg http://www.analogconsultants.com/
Hey, I have an app where a gapped pot core might be the thing to use. It's a little speculative right now, but it might be worth a little research. Do you know of anyone who could research available cores and do some math, to see if what we need is even feasible? We need, basically, a couple of uH at high peak current and low duty cycle (ie, low longterm RMS current), space constrained. We have more height available than pcb footprint, and most commercial inductors tend to be short.
We could even stack two pot cores on a common mounting bolt, to get more height for available footprint. Wire in parallel to the PC board. Phase for leakage field cancellation?
-- 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
We
Off the top of my head I don't recall anyone specializing in this sort of research. But "a couple of uH" and pot core are a somewhat odd combination. Pot cores are nowadays (usually) only available in the very high permeability ferrites. I guess that is because all the old telco filter apps have become obsolete or gone DSP.
Of course, they will still bake you a custom core but that's going to cost.
height
If the air gap is in the middle as usual then the field cancellation effect won't be that great.
-- Regards, Joerg http://www.analogconsultants.com/
Le Thu, 20 Jun 2013 08:44:36 -0700, John Larkin a écrit:
How high is high?
I once worked on an IGBT bridge where the load was a mH order of magnitude/50kA (yep) air coil. Wires did jump on the floor when doing the short circuit test. Intimidating the first times.
-- Thanks, Fred.
The current is 130 amps maybe. Two inductors, each 2 uH at 65 amps, is in the balpark. So each inductor has to store about 35 millijoules.
Height needs to be about 1" or less for the stack. Max diameter is
0.75", but preferably less.I hate magnetics design. The part data sheets are all in gauss and mean path lengths and cross-section areas and permeabilities. Why can't they spec cores in electrical units like amp-turns and joules and ohms of copper? You're lucky to get A-sub-l.
-- 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
You could talk to these guys. They just designed a custom ferrite part for one of my projects that has a highly unusual form factor and power density, and numerous big players in the field had thrown in the towel on that one:
The first video is pretty cool, especially at 1:37 min :-)
This series already seems to come quite close to what you need, and it's mil-spec if that matters:
They are extremely conservative with their Isat spec, I'd talk to them to see how small you can go for 130 amps peak. But I don't know if it's a pot core inside (could be more open).
Don't know the exact requirements in your application, they also have toroidal swinging chokes that could possibly be stacked. Those lose inductance in a controlled fashion when the big pulse comes, and afterwards the nominal inductance returns.
-- Regards, Joerg http://www.analogconsultants.com/
Le Thu, 20 Jun 2013 13:19:19 -0700, John Larkin a écrit:
Well,
L.I = n.B.Ae => n=L.I/(B.Ae)
Lets try the fatest RM core that roughly fit your needs (RM12) From ferroxcube catalog:
Ae=146mm^2 Say Bpeak = 200mT, then n=8.9 turns. Then you aim for A_l=2u/9^2 = 25nH/T^2
OK, lets see if that's practical: the A_l=160nH/T^2 already has a 1.57mm gap. To get to 25nH/T^2 you'll need to gap it to 1.57*160/25=10mm I guess you get the picture.
Another way to see that is: almost all energy is stored in the gap. W=1/2 L.I^2 = 1/2 v.B.H = 1/2 v.B^2/u0 with v = gap volume.
You want 1uH@130A so W=8.5mJ which gives v=2.W.u0/B^2=534mm^3
which for 1mm thickness is 26mm dia. 2mm gap and your central leg is 20mm dia.
Pbs are the low Bsat, and the bad gap placement, which limit its volume.
You'd better use some micrometal toroid.
Let see: T72-2 => A_l=12.8 => n=9 for 1.03uH It has an Ae=34.9mm^2 which gives B = L.I/(n.Ae) = 414mT
from
Design done. Well almost...
-- Thanks, Fred.
I ran the Micrometals software and they want a T130 or bigger. One solution is T130-8 with 6 turns. T130 is too big.
-- 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
Le Thu, 20 Jun 2013 16:00:22 -0700, John Larkin a écrit:
Check my calcs, I may have made a mistake, but I don't think so and the T72-2 should go. Maybe they derate Bsat, thanks to the unfriendly aging behavior of iron powdered cores, but in your case, if I understood well, this should not be a pb. Maybe contact them to have some figures, but the graph tells, it well I think.
And it's easy to give it a try...
-- Thanks, Fred.
Le Thu, 20 Jun 2013 23:15:48 +0000, Fred Bartoli a écrit:
Just checked with their tool.
Either 1x T72-2 with 9 turns => 1u and inductance is still 90% at 130A or 2 paralleled T72-2 with 13 turns => 1.97uH @ 65A or 2 stacked T72-2 with 7 turns => 1.13uH @ 130A
-- Thanks, Fred.
The SRF series from Bourns
may use ferrite cores, but it doesn't look as if they incorporate anything as sophisticated as a two-part RM core.
They may be cheap, but they are certainly nasty. My point throughout this t hread has been that they aren't actually all that much cheaper than a prope rly designed coil wound onto an off-the-shelf former by a cottage coil wind er near you, and since you get to choose the the off-the-shelf ferrite core that you wrap around the coil, you've got a lot more control over the indu ctor you use.
Because you can tailor the coil and the ferrite to you application, you wil l normally end up using less copper and less ferrite than you would if you settled for the nearest available off-the-shelf core, and there's a real ch ance that you will end up with a cheaper product, as well as a better one.
Machines for winding toroidal cores exist, and are cute, but since you've g ot to wind the copper on to the - gappable - dispensing drum before you win d it off again onto the toroid, they are relatively slow. This hasn't stopp ed iron-cored toroidal mains transformers from pretty much taking over the market for consumer electronics - the flat transformers give you nice flat pieces of kit to stick on your shelves.
They've been replaced by RM cores, which di seem to be produced in serious volumes.
So? Every improvement you make to electromagnetic compatibility only goes s o far. The trick is to make enough of them to make your problem small enoug h to live with. Lots of small improvements don't impress like one big impro vement, but they do get the job done. That's engineering versus showmanship .
-- Bill Sloman, Sydney
I think you haven't talked to cottage coil winders in a few decades. The last I knew was in the German town I used to live in as a teenager. That closed in the early 80's and became ... ... a BP gas station :-(
The current custom magnetics houses are larger places with CNC machines and the whole nine yards. There, you either have to cough up a substantial chunk of money for NRE or place a large minimum order.
Unless you are talking thousands per month, this is only wishful thinking.
That problem has been licked decades ago. But as I said earlier, it's now done by sizeable custom manufacturers and if you order only 2-3 pieces you have to first put a big check on the table.
Those are more leaky, not suitable for some applications.
Well, in my cases the most important metric is usually time, due dates, first customer shipments. If the success of a whole big machine depends on my electronics then I don't dilly-dally with inductor orientations.
-- Regards, Joerg http://www.analogconsultants.com/
You may not have talked to a cottage coil-winder in a few decades, but I go t some prototype coils wound at
in Horst in the Netherlands last year. They wouldn't appreciate being calle d a cottage coil-winder but it didn't look like a huge place when I dropped in to deliver my formers and winding specifications. Horst is no metropoli s.
I got a couple of pairs of coils wound for about 200 euro - not exactly a h uge NRE, or a huge minimum order. I save a bit by putting on the double scr eening myself - it's a bit fiddly, and doesn't exactly need a coil winding machine.
.I beg to differ.
That's the simplest of the astatic - non-progressive - schemes. Rayner and Kibble list a few more in section 4.2.1 of their "Coaxial AC Bridges" ISBN
0-85274-389-0, if you can find a copy.As far as I know, it's out of print. For a while the British National Physi cal Laboratory at Teddington would print copies on demand (under license fr om Adam Hilger) but they'd stopped that when I last looked, and their curre nt web-site is perversely opaque.
"Ayrton-Perry" is better
I didn't - see above.
Granting the broad-line distributor stocking patterns, RM core would seem t o be suitable for a lot more applications than pot cores.
Unless it's obvious and easy.
-- Bill Sloman, Sydney
This example is most certainly not a cottage place. They have laser alignment and even engineering capabilities, something you won't find at Uncle Leroy's Coilwerks.
Then they were fairly easy coils and did not need a custom bobbin and a custom core.
Well, when have you ordered your last coil and how small was it?
Example from my current project: The coil has to be over 100uH, under
2mm long, 200um in diameter and magnetically completely shielded. Yes, that's micrometer. Try that at a cottage place.
I don't use literature or even math for that. I prototype it and measure the H-field, minimize it, then document the geometries and procedure for production.
That's for resistors, you can't wind a coil like that.
What year was that and what kind of coil? I am talking real cutting edge stuff, where the average cottage coil winder says it's impossible. For the rest, I can wind 2-3 pieces myself.
Yes, they are, but not for the very noise-critical ones.
It never is, else they would not need a consultant to design it :-)
-- Regards, Joerg http://www.analogconsultants.com/
kkeling
Not as "easy" as the Bourns SRF parts that you can buy off the shelf, but a whole lot more suitable for what I wanted to do.
Or try to buy it off the shelf.
This is a long way from the Bourns SRF coils you are touting.
Re-inventing the wheel?
Actually, you can. The other name for it - according to Rayner and Kibble - is the "bootlace technique" where you wind on half the turns progressively , so that they are evenly spread around the toroid, then wind on the other half in the same sense, but coming back around the toroid, so that the end of the coil comes off at the same point as the start came in.
Since the wires cross from time to time, I suppose it might look like a boo tlace ...
But you claim that then it is cheaper to go with the Bourns SRF coils, desp ite the very limited range available - compared with what you can get wound onto off-the-shelf formers and cores.
Depends on how they are used.
What's obvious and easy for a consultant isn't necessarily obvious and easy for a wet-behind-the-ears customer who has never heard of an Ayrton-Perry wound inductor.
-- Bill Sloman, Sydney
[...]
The Bourns is not what I meant, that's a bone-simple inductor. "Not easy" to me means something where many coil manufacturers have bowed out with words like "that is impossible".
So when did you?
That cannot be bought off-the-shelf.
I am not touting that, the SRF series is merely what David is using and in his case it is the right choice. Unless he absolutely needs another percent or two in efficiency, in which case bigger is usually better.
No, I know how the wheel is built and I am designing one that's nicely balanced. I do not need books for most of the stuff I design. Although I must confess that when I had to do some heavy complex X+jY math a couple weeks ago it produced some sweat beads. It's been so long ago that I did it last time. However, a brief excursion onto the Internet was sufficient to get back into that stuff.
That's an old trick. I know it under the name backwards interleave but there are probably many other names for it.
It is cheaper. If you have the space and the efficiency is ok, why go custom when you don't have to?
RM cores are more noisy, there is no way around that. The more open a magnetic path is the more stuff leaks.
So where is your example of an Ayrton-Perry inductor?
-- Regards, Joerg http://www.analogconsultants.com/
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.