Found this site with good info on winding output transformers:
http://www.geofex.com/Article_Folders/xformer_des/xformer.htm
Noticed one error in a formula they have for computing turns:
N=(E*10**8)/(kfbA) A (area) should be in sq. cm not sq. inches.
Site doesn't mention toroidal xfrmr winding techniques though. But the key to high performing toroids is just using back and forth dithering of the core while winding so as to make winding layers chevron shaped and cover near 360 degrees in one sweep around core per winding. (don't overlap ends) (use good insulation between layers and core too) This essentially amounts to micro-sectioning a winding. Interleaving does not benefit in toroid case (since leakage inductance is already minimal) and only makes distributed capacitance worse. I have some cheap (surplus) Avel-Lindberg power xfmrs that appear to have been wound this way. They have a flat frequency response to 150KHz and resonance at 450KHz!!! This is far better than most "Audio" output transformers! I measured a Hammond 1650T (a low turns ratio model) trannie to compare and it is flat to 30KHz and resonant at 50KHz. Main drawback of toroid outputs is that the DC current balance must be strictly maintained to prevent saturation problems.
See my earlier posting on how to extend any transformer's phase linear bandwidth up to Mhz range.
http://www.geofex.com/Article_Folders/xformer_des/xformer.htm
Noticed one error in a formula they have for computing turns:
N=(E*10**8)/(kfbA) A (area) should be in sq. cm not sq. inches.
Site doesn't mention toroidal xfrmr winding techniques though. But the key to high performing toroids is just using back and forth dithering of the core while winding so as to make winding layers chevron shaped and cover near 360 degrees in one sweep around core per winding. (don't overlap ends) (use good insulation between layers and core too) This essentially amounts to micro-sectioning a winding. Interleaving does not benefit in toroid case (since leakage inductance is already minimal) and only makes distributed capacitance worse. I have some cheap (surplus) Avel-Lindberg power xfmrs that appear to have been wound this way. They have a flat frequency response to 150KHz and resonance at 450KHz!!! This is far better than most "Audio" output transformers! I measured a Hammond 1650T (a low turns ratio model) trannie to compare and it is flat to 30KHz and resonant at 50KHz. Main drawback of toroid outputs is that the DC current balance must be strictly maintained to prevent saturation problems.
See my earlier posting on how to extend any transformer's phase linear bandwidth up to Mhz range.
transformer laminations
Just a comment on E-I transformer laminations.
Most info I have seen on DIY output transformers have mentioned "scrapless" or "no waist" laminations. These are proportioned in such a way as to waist the least amount of steel sheet when punched out and are the most commonly available laminations. They are widely used in power transformers due to their economy.
They are NOT however the best laminations for audio transformers. The famous Acrosound and Dynaco transformers used long E laminations. These have longer length to the three legs of the E lam. The reasons they are better for audio are twofold: The grain oriented steel used can have a larger proportion of the E lamination magnetic field correctly oriented to the grain orientation in operation. The leakage inductance of the windings is reduced as the coil is made longer. (See page 97 of "High Fidelity Sound Engineering" by Norman Crowhurst, Newnes, 1961)
The reason being that the leakage pathes become long thin loops through the winding cross-section which have nearly the same path length as the desired flux thru the steel core. Only if the leakage path lengths are significantly shorter than the steel can they overcome the much higher Mu of the iron. The conventional "Scrapless" laminations give a near square winding cross-section with near circular leakage pathes, N. G.
It is a simple extrapolation to take this long E enhancement idea to the extreme and one ends up with the toroid configuration where all flux pathes have to go around the toroid and there is no path length advantage whatever for the leakage flux. (Assuming the windings cover near 360 degrees.) Toroids also maximaize the steel grain orientation in the direction of the magnetic field 100%. Not surprising that toroids have greatly improved bandwidth and better magnetic properties. Their lack of any lamination end gaps distributed around the path causes the effective permiability to be greater so close attention to DC balance is a must! (ie. have to use a DC bias servo on output tubes to zero out DC current imbalance or else adjust every day!)
Since toroids are so difficult to wind without automatic equipment, I suggest using four I laminations in a square configuration with windings on all four legs as a simple DIY alternative. Solves the DC balance sensitivity issue too.
Better yet, see my other posting on how to use a cheap 60 Hz power transformer and a ferrite toroid with low turns to get near MHz phase linear response bandwidth.
Just a comment on E-I transformer laminations.
Most info I have seen on DIY output transformers have mentioned "scrapless" or "no waist" laminations. These are proportioned in such a way as to waist the least amount of steel sheet when punched out and are the most commonly available laminations. They are widely used in power transformers due to their economy.
They are NOT however the best laminations for audio transformers. The famous Acrosound and Dynaco transformers used long E laminations. These have longer length to the three legs of the E lam. The reasons they are better for audio are twofold: The grain oriented steel used can have a larger proportion of the E lamination magnetic field correctly oriented to the grain orientation in operation. The leakage inductance of the windings is reduced as the coil is made longer. (See page 97 of "High Fidelity Sound Engineering" by Norman Crowhurst, Newnes, 1961)
The reason being that the leakage pathes become long thin loops through the winding cross-section which have nearly the same path length as the desired flux thru the steel core. Only if the leakage path lengths are significantly shorter than the steel can they overcome the much higher Mu of the iron. The conventional "Scrapless" laminations give a near square winding cross-section with near circular leakage pathes, N. G.
It is a simple extrapolation to take this long E enhancement idea to the extreme and one ends up with the toroid configuration where all flux pathes have to go around the toroid and there is no path length advantage whatever for the leakage flux. (Assuming the windings cover near 360 degrees.) Toroids also maximaize the steel grain orientation in the direction of the magnetic field 100%. Not surprising that toroids have greatly improved bandwidth and better magnetic properties. Their lack of any lamination end gaps distributed around the path causes the effective permiability to be greater so close attention to DC balance is a must! (ie. have to use a DC bias servo on output tubes to zero out DC current imbalance or else adjust every day!)
Since toroids are so difficult to wind without automatic equipment, I suggest using four I laminations in a square configuration with windings on all four legs as a simple DIY alternative. Solves the DC balance sensitivity issue too.
Better yet, see my other posting on how to use a cheap 60 Hz power transformer and a ferrite toroid with low turns to get near MHz phase linear response bandwidth.
smoking-amp said:
Could you explain on that, please? I don't know what you mean by 'dithering the core while winding'.
The toroid core gets a winding put on in one rotational pass around the core. Since the number of turns will usually be far greater than what would fit on a single layer, the core rotation during winding gets a small but frequent rotary back and forth (triangle wave) superimposed on the single rotation pass (ending up with something more like a sawtooth with leading slopes longer than trailing slopes). This makes the turns stack up in chevron or V shaped layers (looking at both sides of the toroid, slanted layers if looking at just one side), giving the winding vertical thickness from the core to accomodate the total turns.
Since the turns getting stacked up in chevrons are near to each other electrically, they have relatively little voltage difference between them. This reduces the effective distributed capacitance of the overall winding. Essentially producing a "micro sectioning" of the winding.
Other windings get put on top with the same method, with ample insulation between windings to minimize inter winding capacitance. No need to interleave windings, since all magnetic flux is constrained to the core by symmetry, hence minimal leakage inductance already. All windings should cover near 360 degrees around the toroid.
(Leakage flux involves flux pathes that are shorter than the core material length and symmetric with respect to the winding, and there are next to none with toroidal symmetry. A square-ish or round winding cross section is the worst for leakage, since the leakage pathes make loops in the winding cross section that are shorter than the magn. material path. A long thin winding cross section is best, toroids doing especially well, long E laminations giving a practical compromise. )
Don
Since the turns getting stacked up in chevrons are near to each other electrically, they have relatively little voltage difference between them. This reduces the effective distributed capacitance of the overall winding. Essentially producing a "micro sectioning" of the winding.
Other windings get put on top with the same method, with ample insulation between windings to minimize inter winding capacitance. No need to interleave windings, since all magnetic flux is constrained to the core by symmetry, hence minimal leakage inductance already. All windings should cover near 360 degrees around the toroid.
(Leakage flux involves flux pathes that are shorter than the core material length and symmetric with respect to the winding, and there are next to none with toroidal symmetry. A square-ish or round winding cross section is the worst for leakage, since the leakage pathes make loops in the winding cross section that are shorter than the magn. material path. A long thin winding cross section is best, toroids doing especially well, long E laminations giving a practical compromise. )
Don
OK, I give up, please give me another clue, the search engine could not find it.
I did some experimenting with series connecting a conventional OPT with an air core transformer from an old HP wide range audio oscillator. I could never get the response flat.
Hi George,
Yup, these threads were from way back. And it IS just the series transformer connection scheme with low pass and hi pass filters shown in the RDH4 (page 888). Here is the thread:
http://www.diyaudio.com/forums/showthread.php?postid=12534#post12534
Problem may be that the scheme requires a good broadband inductor for the LP filter. Maybe have to make the LowPass inductor a series connection of an E-I core inductor and a ferrite inductor too.
Don
Yup, these threads were from way back. And it IS just the series transformer connection scheme with low pass and hi pass filters shown in the RDH4 (page 888). Here is the thread:
http://www.diyaudio.com/forums/showthread.php?postid=12534#post12534
Problem may be that the scheme requires a good broadband inductor for the LP filter. Maybe have to make the LowPass inductor a series connection of an E-I core inductor and a ferrite inductor too.
Don
Check out this thread too. Won't work for SE very easily though.
http://www.diyaudio.com/forums/showthread.php?postid=626852#post626852
I just read thru this thread again myself, forgot all about it in the trauma of moving this last Fall. Peter M.'s suggestion of using permalloy wire for wrapping should work well for a SE style xfmr since it gives a distributed air gap.
I also had worked out the kinks from the scheme to get rid of the distributed capacitance from bifilar windings used in it. Uses the same technique of distributed common mode chokes used in transmission line xfmrs. I forgot all about it! I was going to post on this. The trauma from moving was like a nuclear blast that wiped out my memory.
Don
http://www.diyaudio.com/forums/showthread.php?postid=626852#post626852
I just read thru this thread again myself, forgot all about it in the trauma of moving this last Fall. Peter M.'s suggestion of using permalloy wire for wrapping should work well for a SE style xfmr since it gives a distributed air gap.
I also had worked out the kinks from the scheme to get rid of the distributed capacitance from bifilar windings used in it. Uses the same technique of distributed common mode chokes used in transmission line xfmrs. I forgot all about it! I was going to post on this. The trauma from moving was like a nuclear blast that wiped out my memory.
Don
Hi,
Nothing 'bout toroîds here, but it may save hours of tidy hand computing.
http://www.dissident-audio.com/OPT_da/Page.html
Yves.
Nothing 'bout toroîds here, but it may save hours of tidy hand computing.
http://www.dissident-audio.com/OPT_da/Page.html
Yves.
Hi,
Does anyone have any design info (such as # of interleavings etc) for HK Citation II , Peerless S-271-A, Acrosound TO-330 OT's
I just want to consider trying to wind my own high end OT. I've read some on this site wind there own from top designs from the old types.
Might make it much more affordable but even though it looks very hard to do to do right!
Still might sound better than low cost OT.
Thanks Very Much in advance!
Sincerely,
Randy
Does anyone have any design info (such as # of interleavings etc) for HK Citation II , Peerless S-271-A, Acrosound TO-330 OT's
I just want to consider trying to wind my own high end OT. I've read some on this site wind there own from top designs from the old types.
Might make it much more affordable but even though it looks very hard to do to do right!
Still might sound better than low cost OT.
Thanks Very Much in advance!
Sincerely,
Randy
Last edited:
- Status
- Not open for further replies.