Tells me that John is full of shit, and that it was an agreement. Sure, woven is required to be called true litz and be sold as such, but all the configurations yield the effect, and as you perfectly stated, there is good better and best.
When one is dealing with a five turn primary and makes fifty different transformers, all classed over time for their operational characteristics in the same front end, one eventually compiles huge data sets of said characteristics.
A single, solid 18 ga primary was far less efficient than the final litz wire I ended up using. As I configured each ga and strand choice, I noted the gains and arrived at the optimal ga for the frequency I was running at. It matches the ga in a table mentioned by a respondent in this thread, though I was at a slightly larger ga. Imagine that. I probably could have made it run even better if I had continued into testing a few yet smaller strand choices.
Regardless, the unit went from dropping out at just below 6 volts, which would not meet the spec, to running all the way down to below 3 volts. The designed source is a 9 volt battery.
It also made the choke on the driver FET operate far better. It was 5 turns on a core not much bigger than a quarter inch in diameter. It took testing of several core types and sizes as well as turns count and wire choices.
Yes, optimally, A Litz configuration has to have the strands arranged so that they do not remain in the center of the bundle, BUT the difference is not that great, especially on miniature transformers meant for low power switchers and the like. The fact that a solid core is no longer being used means the effect WILL occur, even if a given strand never makes it out to the edge of the bundle.
I'm having trouble finding it, but a major university set up a high frequency switcher program targetted towards high efficiency inverters. They had a write up or two on litz wire. I thought it was Purdue, but that doesn't seem to be coming up in google.
We ran at about 57kHz, and found that to be an optimal frequency for the cores we were using and characteristics of the switching circuit design we were using.
It ended up being the equivalent for 18 or 20 gauge, and was using both
36 and 43 gauge. I think I ended up using the 36, since I mentioned using a few gauges larger size than the table described for that frequency. I cannot remember the strand count, and do not feel like extrapolating it for you. I posted the formula already.
A bit OT, but I remember seeing a video of high current experiments done at the Magnet Lab, (then at MIT circa 1960's) Where they were using several ~2-3" wide strips of copper.
Flat strip conductors negate skin effect in a similar manner as that of a Litz wire configuration. The main problem with using it as wire is the amount of space it takes up on a bobbin per turn. That makes it only practical used as a hook-up link, or inter-node connection link between points in a chassis.
Yes, low turn counts are possible, and it is good for that, but it is not good for any high turns count app. There is square and flat wire, but it is far narrower than the media in this thread refers to.
"Tim Williams" kirjoitti viestissä:hlqjpu$bi8$ snipped-for-privacy@news.eternal-september.org...
I have seen that kind of copper foil conductors used in some ATX power supplies. Don't remember how many turns as I was only interested in the ferrite core for other projects.
Both methods provide more surface area in which surface currents can flow, lowering Rac. Or, if you prefer, they've got more skin in the game.
Flat conductors have the advantage of greater cross sectional area in limited winding windows (vs litz, where the stranding's insulation winds up gobbling a lot of the cross section, crowding out copper). Flat conductors also transport heat out of the transformer much more effectively.
With litz the cross-sectional area lost to insulation increases both Rac and Rdc, so, for power applications with limited winding windows it can actually be counter-productive--for a given insulation thickness and skin depth there's an optimum strand size and number.
In practice, I braided my own 'litz' once to reduce copper loss in a
200-300KHz-ish converter. I think I used 9 strands of #29 solid copper magnet wire, braided to guide each strand through the bundle appropriately, then compared loss to actual super-fine litz. No difference detectable. So, not perfect, but good enough.
I wound up just using an even fatter gauge of solid copper magnet wire--easier, and nearly as good.
What frequency is this for. If you're under 1MHz, you're mainly fighting proximity effect, not skin effect. To deal with proximity effect, all you need is bunched conductors (twisted), not Litz. The Litz wire I have come across use bunched groups twisted into a larger bunched group. This closely approximates Litz.
Hey qrk, I haven't seen that information before, do you have anything to site that would make me believe it? To quote Dagmargoodboat, [At 290Khz] "Comparing the braid to the equivalent-cross-section solid wire:
(view table in Courier font)
Winding Rac (calculated)
---------- ------------------
7 x 0,23mm 1.46*Rdc
1 x 0,608 4.29*Rdc
So, the braid was ~ 3x better.
Here are a couple of skin effect calculators.
formatting link
formatting link
Ok, you need to explain what you mean by bunched conductors, Are they insulated bunched conductors? As stated before proximity effect is minimized by making every conductor find itself in the same position in the bundle an equal amount of time. Twisting may or may not do that, depends on the amount of conductors twisted. Mike
Yes, "closely approximates Litz" because it would not be as good regarding proximity effect. It doesn't have every conductor find itself in the same position in the bundle an equal amount of time. Mike
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.