Choosing an output transformer

[Sheldon]

download this program and play: OPT Calucator

Sheldon

[BudP]

A couple of notes on the formula Johan provided.

Lp = 3,2*A*µ*N²/l*10(power 8)

A needs to be modified by the stacking factor normally applied to commercial core. This number varies with usage, i.e. 1X1 interleave 0.92, 4X4 interleave 0.93, planar butt stack with out gap 0.94, with gap 0.92 and either 3 section or 4 section chunk stack 0.92. This stacking factor takes real world stamping, E/I commercial core burr and misaligned core closure into account, and unless you are an expert at soft touch core construction and the core has no burr, these numbers will apply. 33 gauge M3 will have a stacking factor of 0.95 almost regardless of type of construction.

u (i don't remember my dos keyboard commands) is taken from the core perm charts. If you need an at rest inductance, choose the perm from the far left hand edge of my graph. This is for nominal core permeability and will provide a very accurate lower limit to your inductance measured at 1 vac & 120 Hz on a typical bridge. Incidentally, US core (meaning Tempel Steel) is almost always nom to max, European core is almost always max as is Japanese core and Chinese mil grade is also nom to max. Chinese commercial core is almost never nom but is typically min to nom. Commercial core has a spread of + or - 25% on either side of nom (nominal). This is a reflection of the care in alloying, and heat treating that matches the specific alloy that is actually made, as opposed to theoretically needed. Chinese commercial core will attain US core standards within the next 5 years or so. That they are as good as they are shows how hard they have been working the problem. Not all US manufacturers were even this good, back when they were in business.

I like how Johan is providing the mathematical aspects of transformer design and so I intend to provide sidelights, details and some of the choices that need to be made when designing for power or audio. The two are basically antithetical applications of the same formulas.

Bud

[Johan Potgieter]

Sorry, Moonbird,

My post seems to have thrown off .htm following the calc., thus for the last term: .calc.htm. (For some darn reason my pc does not want to take the whole expression including the '.htm').

Your ref output-trans-theory have been written by the same guys, Turneraudio, differing just slightly. Hope you can come right.

The reference given by Sheldon is the same as what I gave for 'Dissident'.

[Johan Potgieter]

Moonbird,

There still remains the question of intersection capacitance. I hope I will be excused for not discussing that here, as all can be found in RDH, chapter 5.3(v); I will only be repeating. (I would hate to become like the proverbial film: Overexposed and underdeveloped.)

It will hopefully be clear that there will be capacitance between sections which will take part in attenuating high frequencies, and those will override interlayer capacitance effects between layers of the same winding. Thus a similar state of affairs will be found when calculating capacitance than for calculating leakage inductance. Only that which decreases the leakage (small spacing between sections), will increase capacitance. Thus, if somehow it is found that the figure for leakage reactance found may be increased, that is best done by increasing the isolation thickness 'c'. (RDH, fig. 5.13F). In addition, the di-electric factor of the insulation will play a role in the final capacitance - all of which will be clear from chapter 5.3(v) mentioned above.

The greatest problem with the calculated parameters, is consistency in construction. It should be understood that the process of winding a transformer is just simply, that exactitude is impossible because layers would not be perfectly flat over a flat surface. Still usefull transformers are made. My one quibble with winders is that they require some 20+ space factor, while it is possible to work to <10%. (Space factor is the difference between exact winding and that which is reached in practice.) For myself I calculate on up to 90% coverage, and then compress windings in a vice before fitting the core. (The space is there if the calculations were not faulty.) It furthermore improves efficiency to have a 'full window' instead of one through which the cat can still jump.

That must be enough from me for now. I have not dealt with the various types of core (Bud did that) - with which a further compromise appears: Does one want to pay twice as much for a small benefit i.e. diminishing returns? Etc. etc. - this brings the story dangerously close to boutique design. One should keep perspective on what exactly influences final performance of the whole, to what degree.

A number of finer points remains; I have tried to show the route successfully taken by me. It remains to point out that the graphs used and many of the procedures described in RDH, are the actual products of that titan of electronics of half-a-century ago, Norman H. Crowhurst.

[moonbird]

Thanks folks you have given me much to read and consider. This is truely a great resource for we noobs!!

[moonbird]

To focus this discussion a bit -- please consider the 6T4 - a low mu 7-pin tube with seeming possibilities as an output tube in a low wattage amp.

Here is a typical operating point from the NJ7P tube database:

Class A AmplifierPlate Voltage ................................. 80 V
Grid No. 1 Voltage from Cathode Bias Res ................ 150 Ω
Amplification Factor .......................... 13
Plate Resistance (approx) ..................... 1860 Ω
Transconductance .............................. 7000 µ
Plate Current ................................. 18 mA

Could someone comment on possible input impedence for the OT and why?

Might this be a candidate for an OTL design? thx.

[Johan Potgieter]

Moonbird,

Unfortunately not. A very rudimentary calculation shows that only about 100mW could be obtained that way, with a 4,400 ohm load (single tube). A quite higher anode voltage would increase that, but one runs out of spec. before 1W is reached. You mention a low wattage amp, but I cannot guess where someone would use such a low output. Do you have any particular reason for wanting this tube?

[BudP]

For those of you scratching your heads about Johan's posting, here are two load line curves to look over. The 6.6kz is derived from max current and voltage numbers. When brought down into the tubes operating range the halves of the AC wave form are not quite symmetrical, though quite close. The 4.4kz plot shows what Jacobs derivation produces, exactly symmetrical AC waveforms. This then will be the lowest distortion and greatest power this tube can provide. Please note where the ends of the load line stop, this is how you do one of these things, assuming you have good information, which the tube data sheet available here provides.

Frank's electron Tube Data sheets

Have fun!

Bud

[moonbird]

BudL -- thanks much for the gentle nudge - I need to study your last post (not good today). I would like to step through this process under the watchful eye. Thanks much for taking the time ... I'll be back after some more reading tx again.

[Johan Potgieter]

Mmmmm .... Rather too rudimentary; erroneous! (my reply #27).

Thanks to Bud for placing those graphs. The load I used (if I am Jacob!) was the leftmost red line, left graph. Staying with negative grid voltages only, it will be noticed that the lowest the anode swing can be, will be 36V, at 10mA (rounding). That would mean a possible anode swing of 44Vpp (80 - 36) at a current swing op 10mAp. But as an output stage uses a transformer, not resistor, as load, that voltage swing is actually 44Vp and not peak-peak. (The transformer inductance allows an anode swing to 124Vp on the negative grid excursions.)

Thus the maximum output according to these figures would be (44 x 0,01)/2 (for r.m.s) = 220mW and not a rounded (because of efficiency losses) 100mW as stated previously. Apologies.

But still rather low for any practical use.