Transformer insulateing tape?
in [Design]
|
Prev: Capacitor discharge probes
Next: makes no sense to me
From: Archimedes' Lever on 19 Jun 2010 20:59 On Sat, 19 Jun 2010 20:01:03 -0400, Jamie <jamie_ka1lpa_not_valid_after_ka1lpa_(a)charter.net> wrote: >Years ago for non critical apps, we used wax paper for layer >insulation for winding multilayers.. It was referred to as 'fish paper'.
From: Archimedes' Lever on 19 Jun 2010 21:12 On Sat, 19 Jun 2010 20:03:54 -0400, Hammy <spam(a)spam.com> wrote: >Primary inductance 300uH, Dmax 0.4, switching frequency 60kHz +/- 10% >Operateing mode (if it isnt already obvious) DCM. You said AC fed. Jeez, you didn't even have to mention AC line if it was not a line transformer. If it is a pot core, you may find a nice operating freq at 56-59kHz. Generally about 57kHz. I also assume that you probably bought the same line of pot cores that the whole industry generally uses. YMMV. Shouldn't matter, I know... but if you get it going at 60kHz, I'd almost bet money that it gets more efficient at the numbers I gave you. That is from almost a decade of characterization. Could have been our circuits, but it was that way on several different designs, so I lean toward the core being 'happy'. Also, I rarely need a gap of more than a few mils, but you are doing a flyback IIRC. Will your primary be parallel wound (flat) or are you going to fashion an in-house Litz by twisting them into one stranded line? (YES, it IS Litz at that point and tests will prove it) Do not over twist. It should be no more than needed to keep them grouped. Use double strength mag wire insulation type. A twisted set will likely yield a bit more efficient operation than the flat wind will. Stacked bobbin breaks? or stacked winding schema where you stack by hand? Also are you saying "stacked" as in would one over the top of the other, or stacked as in one at the bottom of the bobbin space, and one above that with the primary underneath? The volts per turn choice looks great
From: Hammy on 19 Jun 2010 21:47 On Sat, 19 Jun 2010 18:12:34 -0700, Archimedes' Lever <OneBigLever(a)InfiniteSeries.Org> wrote: >On Sat, 19 Jun 2010 20:03:54 -0400, Hammy <spam(a)spam.com> wrote: > >>Primary inductance 300uH, Dmax 0.4, switching frequency 60kHz +/- 10% >>Operateing mode (if it isnt already obvious) DCM. > > > You said AC fed. I'll assume the misunderstanding is with my use of "DCM" I'm referring to Discontinuous Conduction Mode or complete energy transfer. The converter is fed from the AC outlet but its not for APFC, its rectified and filter the tranny will see between 110Vdc to 185Vdc with about 20% ripple.Technically I guess the switch turns this to AC at least as far as the flyback inductors are concerned. > Jeez, you didn't even have to mention AC line if it was not a line >transformer. > > If it is a pot core, you may find a nice operating freq at 56-59kHz. > > Generally about 57kHz. I also assume that you probably bought the same >line of pot cores that the whole industry generally uses. YMMV. I'm using an etd/29/16/10 ferrite 3C90 this one from newark. http://canada.newark.com/ferroxcube/etd29-16-10-3c90/ferrite-core-half-etd29-3c90/dp/68C5566 I'm using the 13 pin former that's shown in the data sheet. > > Shouldn't matter, I know... but if you get it going at 60kHz, I'd >almost bet money that it gets more efficient at the numbers I gave you. > That is from almost a decade of characterization. Could have been our >circuits, but it was that way on several different designs, so I lean >toward the core being 'happy'. > > Also, I rarely need a gap of more than a few mils, but you are doing a >flyback IIRC. > Will your primary be parallel wound (flat) or are you going to fashion >an in-house Litz by twisting them into one stranded line? (YES, it IS >Litz at that point and tests will prove it) Do not over twist. It should >be no more than needed to keep them grouped. Use double strength mag >wire insulation type. A twisted set will likely yield a bit more >efficient operation than the flat wind will. Stacked bobbin breaks? or >stacked winding schema where you stack by hand? Yes I'm hand twisting the wires and thanks for the tip about being to tight I had forgotten about the optimum twist per inch. > > Also are you saying "stacked" as in would one over the top of the >other, or stacked as in one at the bottom of the bobbin space, and one >above that with the primary underneath? Here's is a picture of what I mean when I say AC stacked windings. http://i49.tinypic.com/wugfw5.jpg My minus output windings are configured the same as the positive ones shown in the pic. I'm using two pins for the return. But I am also layering the primary 2 layers 1 as the first layer then the secondaries and bias winding then the second layer of the primary. This is supposed to increase coupling and reduce leakage. > The volts per turn choice looks great
From: Bill Sloman on 19 Jun 2010 23:08 On Jun 20, 12:22 am, Archimedes' Lever <OneBigLe...(a)InfiniteSeries.Org> wrote: > On Sat, 19 Jun 2010 14:18:46 -0700 (PDT), Bill Sloman > > <bill.slo...(a)ieee.org> wrote: > >On Jun 19, 6:50 pm, Hammy <s...(a)spam.com> wrote: > >> What kind of tape can you use for insulation between primary to > >> secondary of flyback transformer. I know mass produced professional > >> designs use a 3M specialty tape. What I'm looking for is a tape I > >> could pick up at Staples or some office supply store or hardware > >> store. > > >> The vinyl electrical tape I can find is only rated for 600V and 80C > >> MAX TEMP. > > >> Is there a commonly available tape that is good for higher temp and > >> offers higher insulation? I've read of people using a mylar tape is > >> there a brand name ,type anyone could recommend? > > >> This is for a 90 - 140Vac input flyback. > > >As has been mentioned here, 3M's yellow polyester film is popular. > > >Farnell/Newark stocks it in three widths - 12mm, 19mm and 25mm. You > >can order a single 66 metre roll for a couple of dollars. The Farnell > >order codes are 726-977, 753-002 and 753-014. > > I already posted the Farnell links, oh, great idiot that never reads the > goddamned thread. So you did. If I'd read through the URLs that you gave, or clicked on the links, I'd have realised that you were directing the OP to Farnell. Sorry about that, but I usually treat URLs as meaningless character strings that point to real information, thus saving myself the time required to parse them. > >If I remember rightly, the tape is 60 micron thick. > > Folks around this industry use mils to describe thicknesses. Americans are particularly prone to do this. The polyester backing film is nominally 1 mil thick, but with the adhesive it builds up to 2.5 mils - nominally 63.5 microns. We used it between layers of round wire, so presumably some of the adhesive was squeezed sideways into the gaps between the wires. http://multimedia.3m.com/mws/mediawebserver?mwsId=66666UuZjcFSLXTt4XfEm8z6EVuQEcuZgVs6EVs6E666666-- > Especially on things like tapes and films. The electronics industry is infested with dimensions expressed in bizarre fractions of an inch and I don't think that tapes and films are any worse affected than - say - the lead pitches on ics and connectors. > Granted microns can be used, but it has a lot less familiarity among > folks, and also a lot less ability to look at something and declare a > number. I can do mils fairly accurately, but calling something "1 mil" > is a lot easier than trying to tag it with a micron numeric where the > margin for error is higher. We checked our tape and film thicknesses with micrometers, some of which were calibrated in microns, others in thous (thousandths of an inch or mills). Both gave the same margin of error. -- Bill Sloman, Nijmegen
From: Archimedes' Lever on 19 Jun 2010 23:34
On Sat, 19 Jun 2010 21:47:50 -0400, Hammy <spam(a)spam.com> wrote: >I'm using an etd/29/16/10 ferrite 3C90 this one from newark. > >http://canada.newark.com/ferroxcube/etd29-16-10-3c90/ferrite-core-half-etd29-3c90/dp/68C5566 > >I'm using the 13 pin former that's shown in the data sheet. Yeah... I think ours was that or perhaps 3c85. Anyway, we found that happy point over several product iterations we had that were fired in completely different manners, so we were pretty convinced it was the transformer. |