Here it goes...
In post #88 Papa says: "One of the charms about the transformer vs JLH/PLH phase splitter is that it delivers pure symmetry in the same way a circlotron does - each like polarity part is subjected to the same conditions, although in opposite phase..."
In order to achieve "pure symmetry" the effects of the parasitic currents due to the transformer windings to have must have a symmetric effect on the two output FETs.
There are numerous formulations for the optimal modulation voltage for the "spare" winding: an approximation that assumes the open-loop gain (OLG) is very high and more exact formulas that takes OLG into account. I present here the simpler approximation.
Assumptions:
Thus the AC current into the upper winding 3 is the sum of the currents from other other 3 windings which is Itotal[3] = I[3,1]+I[3,2]+I[3,4]
The AC current into the lower winding 4 is Itotal[4] = I[4,1]+I[4,2]+I[4,3]
The AC currents have a symmetric effect on the output FETs when Itotal[4] = -Itotal[3]
0 = Itotal[3] + Itotal[4] = I[3,1]+I[3,2]+I[3,4] + I[4,1]+I[4,2]+I[4,3]
= I[3,1]+I[3,2] + I[4,1]+I[4,2]
= 2*PI*f*C*(V[3]-V[1] + V[3]-X + V[4]-V[1] + V[4]-X)
= 2*PI*f*C*(2*Gain*Vin-2*Vin - 2*X + 2*0)
= 4*PI*f*C*(Gain*Vin-Vin-X)
X = (Gain-1)*Vin
In post #88 Papa says: "One of the charms about the transformer vs JLH/PLH phase splitter is that it delivers pure symmetry in the same way a circlotron does - each like polarity part is subjected to the same conditions, although in opposite phase..."
In order to achieve "pure symmetry" the effects of the parasitic currents due to the transformer windings to have must have a symmetric effect on the two output FETs.
There are numerous formulations for the optimal modulation voltage for the "spare" winding: an approximation that assumes the open-loop gain (OLG) is very high and more exact formulas that takes OLG into account. I present here the simpler approximation.
Assumptions:
- The interwinding capacitance is C and is the same between any 2 windings.
- Winding 1: The driven primary winding has an AC voltage of Vin.
- Winding 2: The "spare" primary winding has AC voltage X.
- Winding 3: The upper secondary winding has an AC voltage of Vout=Gain*Vin.
- Winding 4: The lower secondary winding has an AC voltage of 0.
Thus the AC current into the upper winding 3 is the sum of the currents from other other 3 windings which is Itotal[3] = I[3,1]+I[3,2]+I[3,4]
The AC current into the lower winding 4 is Itotal[4] = I[4,1]+I[4,2]+I[4,3]
The AC currents have a symmetric effect on the output FETs when Itotal[4] = -Itotal[3]
0 = Itotal[3] + Itotal[4] = I[3,1]+I[3,2]+I[3,4] + I[4,1]+I[4,2]+I[4,3]
= I[3,1]+I[3,2] + I[4,1]+I[4,2]
= 2*PI*f*C*(V[3]-V[1] + V[3]-X + V[4]-V[1] + V[4]-X)
= 2*PI*f*C*(2*Gain*Vin-2*Vin - 2*X + 2*0)
= 4*PI*f*C*(Gain*Vin-Vin-X)
X = (Gain-1)*Vin
have no objection 
there is still one dilemma which you probably need to solve - is it that , evidently measurable progress , beneficiary to sound ?
what I grasp till now from Pa's writings , it's desirable to have K2 above K3 (and greater), but their ratio is still Golden one ......
considering that I believe in his ear/brain combo more than to mine ( if nothing else , mileage is greater
) , I also believe that we all are waiting to see his choice regarding that ....
great work ,lhquam
there is still one dilemma which you probably need to solve - is it that , evidently measurable progress , beneficiary to sound ?
what I grasp till now from Pa's writings , it's desirable to have K2 above K3 (and greater), but their ratio is still Golden one
......considering that I believe in his ear/brain combo more than to mine ( if nothing else , mileage is greater
) , I also believe that we all are waiting to see his choice regarding that ....great work ,lhquam
I am not sure that the ratio of H2 to H3 can be held constant across frequency with H2>H3. It is certainly possible to hold THD nearly constant ans shown in post #2237.
have no objection
...
what I grasp till now from Pa's writings , it's desirable to have K2 above K3 (and greater), but their ratio is still Golden one
...
great work ,lhquam
Ihquam,
i can follow your calc but do not understand your statements 3 and 4
Winding 3: The upper secondary winding has an AC voltage of Vout=Gain*Vin.
Winding 4: The lower secondary winding has an AC voltage of 0.
Having four identical windings, i thought each were showing same AC voltage.
What "Gain" exactly define?
i can follow your calc but do not understand your statements 3 and 4
Winding 3: The upper secondary winding has an AC voltage of Vout=Gain*Vin.
Winding 4: The lower secondary winding has an AC voltage of 0.
Having four identical windings, i thought each were showing same AC voltage.
What "Gain" exactly define?
I am referring to AC voltage between the winding and ground (or other windings) not the AC voltage across the winding.
By Gain, I am referring to the closed-loop gain of the amplifier. Actually, the ratio of Vout to the voltage into the top of the driven primary winding.
By Gain, I am referring to the closed-loop gain of the amplifier. Actually, the ratio of Vout to the voltage into the top of the driven primary winding.
I have been listening to the flat 0.1% THD version with Gain=5, Rs=0R12.
I am not happy with it.
Not having 4 instrumented pots I have not done serious listening at THD levels other that 0.1%. I find that I cannot maintain a flat THD vs. Frequency spectrum without adjusting both P3 and P4, the upper and lower Zen pots.
Since Papa (and others) have been suggesting THD (particularly H2) up to 0.1%, I will shift away from high open-loop gain and feedback to more degeneration and possibly higher gain to reduce feedback. My next experiment will be with 0R47 source resistors.
I am not happy with it.
Not having 4 instrumented pots I have not done serious listening at THD levels other that 0.1%. I find that I cannot maintain a flat THD vs. Frequency spectrum without adjusting both P3 and P4, the upper and lower Zen pots.
Since Papa (and others) have been suggesting THD (particularly H2) up to 0.1%, I will shift away from high open-loop gain and feedback to more degeneration and possibly higher gain to reduce feedback. My next experiment will be with 0R47 source resistors.
After switching back to my F5 balanced with minimal THD, I am less happy with it now. High frequencies are too too sharp.
By first F5 build had badly mismatched FETs. But there was a satisfying sweetness to its sound that I have not been able to replicate. At that time I did not have equipment to measure the harmonics, so I do not know what the harmonic mix might have been. So far, my tweaking of the Teaser-6 does not come anywhere near to that sound.
By first F5 build had badly mismatched FETs. But there was a satisfying sweetness to its sound that I have not been able to replicate. At that time I did not have equipment to measure the harmonics, so I do not know what the harmonic mix might have been. So far, my tweaking of the Teaser-6 does not come anywhere near to that sound.
.....the problem could be that two perfect matched halves produce mostly k3 and no k2 and a slight mismatch will care for the k2 part. Of course absolute distortion is growing in this way, but the SemiSouth have so much less distortion compared to the IRF that you can risk some more k2 distortion to tailor the sound!
you must only find a way to get the offset of about 200mV back to zero but keep the distortion spectra....or take a cap!
This could also be achieved by adding a small extra parallel [variable] current source that imbalances the balanced outputs to get just the trifle of H2 you like.
It reduces open-loop gain by a factor of 2 and changes the transformer impedances by a factor of 4. I cannot see any advantage for doing it.
lhquam: A method to tackle what you are doing is to start with an amp which already has an accredited sweet spot and then deny it [and yourself]; but of course reversibly, and at will. For example; you seem happy with your F5. I gather [implicitely] from your reports that it operates at its sweet spot. So here is a proposed approach to unravel the role of the Jensen transformer and loop feedback as potential variables affecting the sweet spot.
The attached file ModF5.pdf shows a proposed modification of the F5 schematic which I copied from the attendant article by Mr. Pass. It is not F5; but a tool to study the suspect variables above. I cut the pristine schematic of F5 at selected spots by using a razor blade. I then inserted the Jensen transformer plus other passive components, and finally spliced everything back together. It may need further tweaking; but I hope that my surgery was least intrusive so as to "preserve' the parent sweet spot. The mod circuit has a Zen Pot [P3; is it the sweet spot generator?], and the bias of the output stage is essentially unperturbed. I am not sure about the components in the NFB path; but they can be adjusted to suit.
The schematics [ConceptualF6#0,1,2.pdf] which I proposed in an earlier post several days ago can also be used as tools to study the effects of above suspect variables. The parent Zen amps are already accredited with "sweet spot" musicality.
You also will have difficulty trying to keep the ratio constant with different
amplitudes. Probably best to select for your listening level.