As I said bifilar winding is not needed for Class A and AB operation OT's. The classic McIntosh is very similar to the patent in your link if not the same in many aspects. It might also be useful, but not necessary a priori, if you want to make a transformer with tertiary winding that allows classic UL operation for those pentodes that cannot have too high g2 voltage. However there is much better solution to this that doesn't require special transformers and Crowhurst wrote on it about 60 years ago (i.e. about the same age of those patents!). This is called "modfied-ultralinear" where the g2 is just supplied like in a normal pentode operation. Problem solved, better performance.
The fact is that the unity coupling (or similar stuff) is good but not just as good as "simpler" class AB UL using some cathode fb. The latter is better at driving typical speakers where load drops below nominal value and is quite reactive.
So it is not about garbage getting undeserved success and good things getting dumped. Sometimes it can happen but it doesn't work like this every time......
Last edited:
I studied this article of his. Norman H.Crowhurst, James Moir, Alexander B. Bereskin, David Hafler, McIntosh are all masters from that era of golden audio. Here is who should study and not on forums! But those different schemes described by Crowhurst all had their drawbacks. You always talk about the load and have never said about the driver. Meanwhile, Crowhurst wrote about this that all these schemes, apart from the tetrod and UL, require a much greater control voltage. And the scheme McIntosh and Cyrclotron is even greater. If you trace the history of all models of McIntosh, then with each model they just did that improved and simultaneously complicated the driver. Still, the amplitude increase in the driver is given not for free. There are either distortions in the driver, or its price. And bifilar winding can be used in any scheme, it is not harmful if properly reeled, like Bereskin. The truth is that in class B and AB, it will have more effect. And by the way I like AB class much more than pure class A because of more efficiency, and more interesting task.
I have made the modified UL. It is extremely flexible and one doesn't need to use the 57%-43% ratio at any rate. The driver is not a problem at all if things are done right: a straight differential with bootstapped anode load that is RC coupled to a cathode follower with dual supply and directly coupled to the finals. The bootstrapping principle is quite simple and efficient: it is possible to swing huge voltages with low distiortion using the same voltage supply of the finals. The cathode follower also maximizes the value anode load because of its huge input impedance. It also eliminates any possibility of blocking. No other feedback apart from cathode fb for the finals and bootstrap for the voltage amp. All the other nested and overall feedback loops were needed for class B operation. They are totally un-necessary for class AB or class A operation.
Last edited:
It looks like it is a 2.4K OT rated 10W. From THD pictures, core type and number of turns I would say minimum inductance is 12-13H that settles around 17-18H as soon as one gets some 10-20 mW output. 10W might be true down to 35-40Hz. At lower frequency I think it won't make it because Bac becomes bigger than Bdc and this is not possible in SE operation. Needs more inductance....
No, it follows the normal triode behavior but it is very small distortion and good enough to get an amplifier that satifies with IEC standards (i.e. <0.2% THD from 20Hz to 20KHz at FULL rated power). Distortion at 1W is 0.01% or less.
I will try to answer.
I am waiting a new one, I hope to reach the L close to 20H with X core. Actual is around 10 H.
The ratio is 1:20
Regarding the wire is bought by a company in Milan.
Regarding the performance with this trafo, I don't know if Mr. Pieter can understand the diagram but at 8 watt out the BW is - 3dB 15 Hz and 65 kHz at - 3dB.
Do you think id not a good performance?
If not please let me know other better ones.
I wrote we hope to have better results on low frequency.
to lisoformio
You can use the test set in post nr. 4 also without power supply for bias current.
As you can see in the same post the shape of THD vs freq and THD vs. freq aren't so different with bias and not bias.
We are trying to use another smart method to have the current.
Magnetic in Italy is the supplier of magnetic core.
I am waiting a new one, I hope to reach the L close to 20H with X core. Actual is around 10 H.
The ratio is 1:20
Regarding the wire is bought by a company in Milan.
Regarding the performance with this trafo, I don't know if Mr. Pieter can understand the diagram but at 8 watt out the BW is - 3dB 15 Hz and 65 kHz at - 3dB.
Do you think id not a good performance?
If not please let me know other better ones.
I wrote we hope to have better results on low frequency.
to lisoformio
You can use the test set in post nr. 4 also without power supply for bias current.
As you can see in the same post the shape of THD vs freq and THD vs. freq aren't so different with bias and not bias.
We are trying to use another smart method to have the current.
Magnetic in Italy is the supplier of magnetic core.
to 45
Of course we have to understand also how much FB is involved.
Sometime I check on the lab the amps with and without FB, the results in some cases are terrible.
Walter
Of course we have to understand also how much FB is involved.
Sometime I check on the lab the amps with and without FB, the results in some cases are terrible.
Walter
Sorry, but you came up yourself with the inductance being too low.
When it is some 10H; yes it is too low, representing an AC impedance of around 1200 ohm at 20 Hz; not good for distortion.
Again,
I told you about L but you can't consider the performance so bad.
This is the first step and I think it is a good start point
or not?
Walter
I told you about L but you can't consider the performance so bad.
This is the first step and I think it is a good start point
or not?
Walter
Not so bad frequency response, but otherwise not good; you will agree.
But it's about litz wire, and you still have to reply on my questions....
But it's about litz wire, and you still have to reply on my questions....
Ok 🙂
I understood. I'm doing a new transformers test equipment for the lab and will made mods in order to use your test method. Thank for the idea.
Some info more? I googled but found nothing.
Check some older posts Walter.
There is still uncertainty about the primary litz wire.
Some OT posts have been moved to: http://www.diyaudio.com/forums/tubes-valves/308956-more-output-stage-local-feedback.html
I wrote about MY ERROR on reporting the right spec (mmq instead diam).
And I confirm other values.
Note:
I ordered a Hammond 16xx s.e. 20H , 20 watt declared, so I check this one also.
I am trying to have EDCOR but they ask 6 weeks for delivery an order!!!!!!!
So I will ask to Sowter, I am an old customer , to have a good s.e. to test.
This just for info
Walter
And I confirm other values.
Note:
I ordered a Hammond 16xx s.e. 20H , 20 watt declared, so I check this one also.
I am trying to have EDCOR but they ask 6 weeks for delivery an order!!!!!!!
So I will ask to Sowter, I am an old customer , to have a good s.e. to test.
This just for info
Walter
I buy an export grade double insulation wire that comes in reels of 25Kg, when they fraction it the tension on the wire is huge (old machine with bad bearings) and a 4% stretching is not rare in thin wires, winding by hand adds even more tension, anyway a 4% error (manufacturer tolerance is about 2.5%) is negligible compared with information provided by the OP. 😉
My calculation was with 0.31mm diameter by the way.
At this high, to believe on values given by the OP I need a picture where appear the transformer, the multimeter and today newspaper... 😛😀
One more try Walter:
When I calculate 1784 turns of normal magnet wire with diameter 0,315 mm on a V51 (SG89/55) core which has average length of 23,7 cm each turn, it is according to your measured DC resistance of 92 ohm.
When your primary litz wire is 0.31 mm² which is equivalent to 0,62 mm diameter normal copper wire, you would have much lower DCR with 1784 turns, and not the space on the bobbin because 0,31 mm² litz wire is at least 0,85 mm diameter.
So my conclusion is your primary litz wire must be equivalent to 0,31 mm diameter normal wire and will be some 0,45 mm outer diameter.
Can you confirm this??
When I calculate 1784 turns of normal magnet wire with diameter 0,315 mm on a V51 (SG89/55) core which has average length of 23,7 cm each turn, it is according to your measured DC resistance of 92 ohm.
When your primary litz wire is 0.31 mm² which is equivalent to 0,62 mm diameter normal copper wire, you would have much lower DCR with 1784 turns, and not the space on the bobbin because 0,31 mm² litz wire is at least 0,85 mm diameter.
So my conclusion is your primary litz wire must be equivalent to 0,31 mm diameter normal wire and will be some 0,45 mm outer diameter.
Can you confirm this??
Last edited by a moderator:
- Status
- Not open for further replies.