Leak TL 25 plus OPT

[Yves]

. . .
I could not open those references; display shows an error tab. Are they correct?
. . .

Apologies for that, should be right now 

Yves.

[Yves]

Shorter (and I hope better) links.

http://www.dissident-audio.com/OPT_da/Papers/4-8-16.pdf

http://www.dissident-audio.com/Transfos/Dissident/Fiches/DA172.pdf

Yves.

[Mars]

@Yves: Thanks for the for the secondary section arrangement ideas.

Ampdog suggested that I look closer at the intersection capacitances. This is the effect that limits the high frequency response of a transformer, and is shown in the blue block on the screenshot. In our example above, we get a roll off at 58kHz. Now we would like to get closer to 100kHz.

As can be seen from the figures on the bottom right (see right hand red circle)



The inter winding capacitances are not the limiting factor, sitting at 444kHz

Now to improve the high frequency response of the transformer a few things can be done:

• have fewer secondary sections
• have thicker paper between primary and secondary section
• reduce the length of copper needed for primary by choosing lower load impedance (try something lower than 5000 ohms)
• reduce the length of copper needed for primary by choosing lower primary impedance (stay above 100H)

Now all of these improvements leads to a simpler transformer, so they are worthwile exploring.

[Mars]

I decided to reduce the primary impedance to closer to the minimum of 100Henry's. Also the secondary wire thickness was reduced to 0.75mm. With only 3 sections of secondary with a nice wad of 1.4mm paper between secondary and primary, the 100kHz top limit could be realised:



Less windings means more pressure on the iron to transfer the power; so the Tesla's goes up. For the this design it goes to a 1.27 Tesla which is good for M6 laminations, but slightly over the border for standard NOSS.

So the lesson is: if you want high frequency response, a simpler transformer with good iron is preferable.

[Yves]

Double click on "M6x" in the list while looking at the inductance . . .

[Ampdog]

Hi Karel,

Something I do not see - kindly broaden my horizons! I notice the 444kHz h.f. spec in the blue section of your programme display, and 58kHz in the green part. Are these for leakage reactance and capacitive -3dB points respectively, or to what conditions do the figures apply?

In the green section it shows a shunt cap. of 1nF and an Fo of 58kHz ... how reached? Now the winding/section  capacitances are distributed, but for a normal arrangement one can roughly calculate an h.f. -3dB point by taking half of the capacitance shunting the primary impedance. If one takes a Zp of 6K.ohm and half of the 1nF i.e. 500pF, one gets a ball-park h.f. roll-off at 53kHz; quite close to the indicated 58kHz. That leads me to believe that the 58kHz is the roll-off f as a result of intersection/interwinding capacitance - this also corresponds with my general experience with a 4-section secondary OPT (the leakage reactance should cause roll-off at > 100kHz). Then whence the quite high Fh of 444kHz?

Then I cannot find reference anywhere in the data entered, to the tube rp being used - this will  determine the practical roll-off point. (Basics for newbies: The performance of a transformer is determined in conjunction with the source/generator impedance. It 'transforms' from one impedance to another, it is not a stand-alone entity - at least it should not be!) E.g., data shows that p.p. KT88s as pentodes have an rp of about 30K, with UL taps of 40% about 5,5K, and as triodes about 3K. Where do I find these values included in the programme to determine roll-off frequencies? If not, which impedances do the programme use? (No criticism of your good work done implied, just curious.)

Regards!

[Mars]

By clicking on the M6x option for the metal used, you get this:



The primary impedance jumps up to 298 Henry. This is going to give very good bass response to the amp. If you consider this OPT goes down to 20Hz, it might cause a small earth quake.

The question is: can you have too much Primary Impedance?

@Ampdog: I think the green/light blue section gives the inter section capacitances. It is not all that clear. Maybe Yves can add some clarity. I thought the purple/blue number of 444kHz reflects the inter winding capacitances.

The Rp of the tube being used is entered at the top left. I used 18 000 ohms for the EL34.

I agree about the 4 sections being limiting to the high frequency response of the OPT. I am now exploring 3 section secondary winding arrangements, and will report my findings.  Thanks to this discussion I have learned much, again.

[Ampdog]

The question is: can you have too much Primary Impedance?

Yes - in the sense of economy vs. necessity. Higher primary impedance means larger core, more windings (as in higher resistance, increased losses) or both - for what purpose? To go down from 12 Hz - 6 Hz? (One is ignoring the H.T. shenanningans for now; the thread did start off as looking for a replacement for a Leak TL-25.)

I think the green/light blue section gives the inter section capacitances. It is not all that clear. Maybe Yves can add some clarity. I thought the purple/blue number of 444kHz reflects the inter winding capacitances.

Interwinding C as per adjacent turns perhaps, total intersection C not (see previous post). We have both already found that too much total intersection-capacitance can start giving us phase shift grief not too comfortably > 30kHz, long before leakage influences start. I also cannot see the relevance of 400+ kHz - perhaps later.
 

The Rp of the tube being used is entered at the top left. I used 18 000 ohms for the EL34.

AH! (Cleaner spectales required and all that.)

But this is then where the valve characterisitcs under use come in. According to my data R18K/EL34 is pretty close to straight pentode use under certain conditions. Specific UL characterisitcs are notoriously scarce; I have a usefull set vs. screen taps for only the KT88. A single reference for EL34s under 43% UL typical use as e.g. in the Mullard 5-20, (Ip=62mA) shows an rp of rather 3K/valve. Inserting that value might ease your solution.

But to be practicel: Our Mr Leak glibly quotes KT66/EL34 for his output tubes. The optimal conditions for these valves are not nearly the same! Thus, vice versa, the exact specs of the required output transformer appears far from critical for achieving his laid down 0,1% distortion. (Just as a sobering thought, as a result of common manufacturing tolerances, even the better class of commecial output transformers can have consistencies exceeding +/- 10%.)

I agree about the 4 sections being limiting to the high frequency response of the OPT. I am now exploring 3 section secondary winding arrangements, and will report my findings.  Thanks to this discussion I have learned much, again.

IN CONCLUSION: ALL THE BEST TO YOU FOR 2012 DOWN THERE IN WELLINGTON, KAREL!  (I was born and bred in neighbouring Malmesbury.)

[Yves]

Yes - in the sense of economy vs. necessity. Higher primary impedance means larger core, more windings (as in higher resistance, increased losses) or both - for what purpose? To go down from 12 Hz - 6 Hz? (One is ignoring the H.T. shenanningans for now; the thread did start off as looking for a replacement for a Leak TL-25.)

Indeed !
But now that you have choosen M6X you can reduce the number of turns or even use a smaller core, just void to go over 1.4 Tesla.
After that (and having "reoptimized" wire gauge to keep both primary and secondary layers full) you will notice some improvement on hi frequency side. But . . .

Interwinding C as per adjacent turns perhaps, total intersection C not (see previous post). We have both already found that too much total intersection-capacitance can start giving us phase shift grief not too comfortably > 30kHz, long before leakage influences start. I also cannot see the relevance of 400+ kHz - perhaps later.

. . . 400Khz is already very (and unnecessary) hi, this value being the theorical f-3 for the specified tube's rp and the obtained leakage inductance.
What we really need is to insure that there is no peaks or dips caused by parasitic caps.

Yes, in the blue window the value shown is the estimated "intersection C" wich usually resonate with Leak L at a frequency indicated by "Fo" in the same window wich is much lower (but not yet objectionable  )

Another cause of peaks and dips if the self resonnant frequency of each primary section.
You can see it by right clicking on each in the "Winding" window at the center.
They all should be at least equal the the "Fo" stated just above and preferably above 100Khz so that "in band" phase shift remains lo enough to allow some feed back without instability.

To acheive that you can reduce the self capacitance of each section by inserting some insulation between their layers.
Enter a value (in mm) in the green box at the right of the "copper" window and right click again on one of the primary section.

Be aware that this reduces the "Max allowed thickness" of the inter winding insulator and also increases the Leak L 
Usally you don't need any inter layer insulation in the secondary, remove it setting the corresponding "ILThick" to zero.

Here again some iterations are necessary.

Have fun and best wishes for 2012 to all !

Yves.

[Ampdog]

Thanks greatly Yves,
Perhaps time for a few exercises in my own design procedures to get a better feel for the roles stuff play.

I will not interefere further with Karel's thread.

[Mars]

Thanks again Yves for the knowledgeable input. I did the right click in the windings window, and noted the the primary sections resonant frequency is below 100kHz. Upping the interlayer interleaving in the primary to 0.4mm got me above the 100kHz minimum.

I also dropped the selected wattage back to 25watts for the Leak TL25 spec and selected a thinner gauge for the primary wire. This is the result: a smaller core and better high frequency response


Could this be a worthy TL25 OPT?

[Yves]

Should be !

[Mars]

Yeah!!