I second this. The beauty of this configuration is that you get great performance without having to sacrifice power. You will sacrifice a lot of potential swing for a slight improvement in distortion and Zout.
What is the leakage reactance of the twin coupled transformers,
Versus the leakage reactance of the McIntosh transformers.
How well do a pair of output transformers couple at high frequency?
High Frequency performance factors, anybody?
Versus the leakage reactance of the McIntosh transformers.
How well do a pair of output transformers couple at high frequency?
High Frequency performance factors, anybody?
The leakage L of the two OTs effectively are in parallel, halving the Leak reactance. However the distributed capacitance doubles (versus the single OT alone). But each OT has only half the voltage swing on it, so that should cancel the extra C effectively.
The bifilar windings in the Mac OT cut the distributed C by half, and couples the plate to opposite cathode closely.
The cross coupling caps work just as well as the bifilar for coupling the Plate to opposite Cathode.
So no real reason to wind bifilar windings. Twin coupled OTs should be comparable in bandwidth. Just like putting twice as many interleaves into a single OT.
The bifilar windings in the Mac OT cut the distributed C by half, and couples the plate to opposite cathode closely.
The cross coupling caps work just as well as the bifilar for coupling the Plate to opposite Cathode.
So no real reason to wind bifilar windings. Twin coupled OTs should be comparable in bandwidth. Just like putting twice as many interleaves into a single OT.
Last edited:
One dis-advantage of the Twin however. I don't think you get 4X the primary inductance with the Twin, just 2X. Even though there are 2X the net primary turns.
Connecting two inductors in series would only double the inductance. The secondary coupling in the Twin has equal voltages, so no effect to connecting those together.
The Twin would be best if wound on a single core. A long E core would be ideal, just put the two winding bobbins next to each other.
Connecting two inductors in series would only double the inductance. The secondary coupling in the Twin has equal voltages, so no effect to connecting those together.
The Twin would be best if wound on a single core. A long E core would be ideal, just put the two winding bobbins next to each other.
Last edited:
If you found two OPTs that were bobbin wound you could restack them with the open side of the E laminations of the two cores facing together. Would be a little air gap but the inductance would hopefully be a lot higher. Whether the gains would outweigh the losses would have to be a matter for experiment. Maybe a cheap pair of those PA line transformers with a 25% wattage primary centre tap would be a good starting point?
I am not talking about the leakage reactance to the loudspeaker, yes that is in parallel.
But what I am talking about is the Series leakage reactance from the plate windings to the cathode windings.
I just bet that the 2 transformer model, is not nearly as good as the McIntosh single transformer.
Yes?
No?
But what I am talking about is the Series leakage reactance from the plate windings to the cathode windings.
I just bet that the 2 transformer model, is not nearly as good as the McIntosh single transformer.
Yes?
No?
I would think that once the cross coupling capacitors are in place the leakage reactance would be completely bypassed, therefore practically zero.
Yes, the cross coupling caps can be set for whatever low impedance you want. Even Mac started using them.
But either technique only really matters for true class B operation. For an audio amp you want to eliminate crossover distortion (gm modulation ), so the plate windings already have low impedance across them at crossover to absorb any inductive cut-off spikes. The bifilar windings may be cheaper to wind (1/2 the mechanical turns) if your equipment is already designed to do bifilar winds.
But either technique only really matters for true class B operation. For an audio amp you want to eliminate crossover distortion (gm modulation ), so the plate windings already have low impedance across them at crossover to absorb any inductive cut-off spikes. The bifilar windings may be cheaper to wind (1/2 the mechanical turns) if your equipment is already designed to do bifilar winds.
Last edited:
For all the talk of McIntosh amps being "class B," I've analyzed the bias points of all of the KT88 amps and I've never seen one come out less than 50mA.
Yeah, that's leaner than most designers run them, but not so lean that you would have issues with notch distortion with a standard output transformer. I believe that McIntosh discovered that even if he could bias colder and avoid switching issues, it wasn't worth doing so. In my experience, Zout rises pretty rapidly (gm falling) with KT88s as you start biasing below 50mA.
I prefer 60mA myself. It is still pretty easy on the tubes and gm/Zout improvements are much smaller above that point.
The bifilar windings of course come with a downside of vulnerability to HV breakdown. (Gnd and B+ right next to each other) Although there are super wire insulations available. Polyimide for example.
With some OTs apparently wound random wise, I would be concerned about a wire vibrating in there unless the windings got vacuum impregnated. That has dielectric downsides though.
Every time someone comes up with a thread to clone the Mac OT, we go thru a futile effort to explain that it can be done cheaply and easily using Twin OTs.
By the way, long E laminations are available (at least were) for constant voltage xfmrs. Some conveniently have 2X the standard bobbin width. And then there is the straight Circlotron that avoids all the OT winding problems altogether.
But, hey, everyone to their own taste or quest.
Edcor lists a CXPP25-1.6K with 16 Ohm secondary (parallel secondaries to get 8 Ohm, 3.2K effective primary in Twin Coupled mode, 50 Watt)
And also CXPP60-2.4K, CXPP100-1.7K, CXPP100-2.2K, CXPP100-2.5K all available with 16 Ohm secondaries for paralleling.
With some OTs apparently wound random wise, I would be concerned about a wire vibrating in there unless the windings got vacuum impregnated. That has dielectric downsides though.
Every time someone comes up with a thread to clone the Mac OT, we go thru a futile effort to explain that it can be done cheaply and easily using Twin OTs.
By the way, long E laminations are available (at least were) for constant voltage xfmrs. Some conveniently have 2X the standard bobbin width. And then there is the straight Circlotron that avoids all the OT winding problems altogether.
But, hey, everyone to their own taste or quest.
Edcor lists a CXPP25-1.6K with 16 Ohm secondary (parallel secondaries to get 8 Ohm, 3.2K effective primary in Twin Coupled mode, 50 Watt)
And also CXPP60-2.4K, CXPP100-1.7K, CXPP100-2.2K, CXPP100-2.5K all available with 16 Ohm secondaries for paralleling.
Last edited:
Oh yeah, using Twin OTs with UL taps, one can also get an effective primary Z which is the -same- as the nominal Zprimary of one of the OTs. This gives a lower % CFB too, for easier drive. ( for .41 UL taps, CFB = 0.41/(1 + 0.41) = 0.29 for 29 % CFB)
Just do the cross couple caps for the UL taps in addition (to the primary ends and B+ taps). Then connect the tubes from primary ends of one OT (plates) to UL taps on the twin OT (cathodes).
A little more complex to solve for the primary Z. For 41% UL, it just comes out to 0.5 x (1 + 0.41)^2 times the nominal OT primary Z. I've been calling it the "Elliptron" configuration, but no one seems to have ever tried it. The 16 Ohm secondaries still get paralled to form an 8 Ohm secondary at twice the Watts.
Just do the cross couple caps for the UL taps in addition (to the primary ends and B+ taps). Then connect the tubes from primary ends of one OT (plates) to UL taps on the twin OT (cathodes).
A little more complex to solve for the primary Z. For 41% UL, it just comes out to 0.5 x (1 + 0.41)^2 times the nominal OT primary Z. I've been calling it the "Elliptron" configuration, but no one seems to have ever tried it. The 16 Ohm secondaries still get paralled to form an 8 Ohm secondary at twice the Watts.