Choosing an output transformer

[moonbird]

Hello all --

Searched around a bit on this subject and found several specific discussions but nothing more general in nature ... so ... I would like to pose this question to the gurus (engineers) ... I are not ah engineer , but I am ready to try to understand the nuts and bolts of this thing .

What are the various scientific and 'shade tree' (if you will) methods for selecting an OT for a particular output section? I am interested in the various impedence, wattage, inductance, and frequency response considerations.

I ask because I have read various methods . For example, one can consult tube datasheets for the 'load resistance', but this is not the same thing as the OT impedence ... or is it? This 'load resistance' is also (obviously) not the same thing as the DCR of the OT ... right? What is the 'translation' method for these ''load resistance' specs?

What if I want to use some oddball tube that has no load resistance specified in the datasheet. I have heard various 'thumbnail' approaches like take 10% of the plate resistance. I am interested in all these "design for dummies' approaches people have successfully used.

I have also read of people finding 'optimal' impedences which must involve some math - I am interested in these methods too.

There also seems to be some tribal knowledge about improving high and low end response that involves the inductance or weight of the OT.

Please feel free to respond to all of any part of this subject. Thanks much for your time.

[Johan Potgieter]

I notice you have just posted. If I may kick off with a brief overview; others can quote the formulae or internet tutorial references.

1. The tube data recommendation for 'load resistance' is based on the optimum figure for the other conditions stated, which normally represent the popular use. Thus other conditions can exist where one or other parameter is optimised.

2. As for the transformer characteristics: One wants the transformer to 'transform', without its own characteristics being in the way. Now the impedance of an inductor decreases with frequency, thus there will be a low frequency where its own impedance will become small compared to the load. That is where it will start to attenuate those and lower frequencies. One wants the inductance high enough at the lowest frequency of interest not to start messing matters up. When one uses the ever-present -3dB concept, that is where the inductive impedance will be equal to the load resistance in parallel with the tube internal resistance (simply put).

Example (neglecting transformer wire resistance): One uses a 6,6K impedance (common for p.p. EL34s) and one desires a lowest frequency of 20 Hz. The tubes' internal resistance (say for UL) will be of the order of 8K.ohm. From the formula
Rl = 2.Pi.f.L

one finds the figure of 28,6 Hy. for primary inductance.

3. High frequency attenuation comes where the leakage inductance (i.e. the magnetism 'lost' because the primary and secondary do not couple perfectly), is sufficient to cause a 3dB drop in power output. This acts in series with the load and tube internal resistance (the latter being 14,6K.ohm = Re say). Now the highest frequency of interest is say 20kHz. With the same formula, one finds a value of 116 mHy.

Practical examples often extend these frequency limits because of other design factors. It is typical to find a p.p. EL34 output transformer of 70 Hy inductance and 16mHy leakage inductance.

4. High freuqency attenuation also occurs as a result of inter-winding capacitance - very often ignored! To calculate this accurately is difficult and depends on the winding tecnique, etc. Enough to say at this juncture that it does play a role; often the final result is best found by establishing these freuqencies in a practical hook-up and design from there.

5. These are the salient parameters. I did not include primary/secondary winding ratio; I am sure you know that. As important is naturally also to have a large enough transformer. Flux density for transformer cores for optimal use as power transformers are usually available from the core manufacturers. But for an output transformer of low distortion one might use only half of that figure, or whatever other guidance is apparent from the characteristics of the particular steel.

6. Types of core: The classic E-I type of core is well-known. But since its development grain-orientated steels are useful for giving lower distortion. In this case the transformer core is strip-wound to the particular dimensions and then sliced in two halves, later to be strapped together over the windings. Such cores are called C-cores, and provide superior transformers, also for power transformers, but at a cost penalty.

Lastly you would have seen toriod transformers. They are similar to C-cores, but not cut in half; the windings are put on by a special machine.

This very briefly and somewhat approximated. A complete guide to output transformers comes in "Radio Designer's Handbook" by F Langford-Smith (p. 206 - 233) (it is unfortunately out of print, but available on the internet). There are also PC design programmes. I am sure others will join in with more particulars and reading references.

[Johan Potgieter]

Further:

I have not answered all of your questions. Many of those are involved with the specific circuit design, and requires background that you might best read up on the internet - any successful (tube) design is intimately involved with the output transformer inter alia.

One thing you asked about: No, the transformer 'impedances' have little to do with dc winding (wire) resistance. In fact one keeps the latter as low as possible because it constitutes a loss. To better understand what 'impedance' is is a story on its own - I am sure there internet references out there. Sorry not to quote; I do not have any titles at this time.

Hope all this helps.

[moonbird]

Thx much Johan. Great kickoff and very helpful. Where can I find out more about the 'ever-present -3dB concept'. thx again.

[BudP]

To add to the transformer basic requirements information.

The inductance required is a complex number, combining both the "Load Line", or AC impedance with the DC Plate resistance. This parallel addition (1/load line + 1/ Rp) must then be added to the direct DC resistance of the transformer primary and the reflected DC resistance of the secondary. All of the tube information can be derived from the tube curves and application data. Just which application data to use is not always clear. Once a total tube and transformer impedance number has been developed, you should use 100 ohms as a primary DCR and 10% of the secondary impedance to derive a number for secondary DCR, as transformer modeling numbers, you derive the required inductance as shown above. This - 3dB down number is your lowest distortion point in the bass and does not reflect an inductance that you might find by using an LCR bridge measuring at 120 Hz and 1 volt as this condition is for a minimum excitation of the core. Instead this is a point on the attached PDF that corresponds to the total AC and DC flux load being applied to the primary of the transformer. To obtain a -0.5 dB frequency response you must multiply the -3 dB amount by 2.76, for -1 dB down multiply by 1.93. The formula for derived performance prediction of inductance can be found in the RDH vol. 4 at the top of page 245. You will find an M6 core perm chart on the preceding page, note that it is for maximum core performance and my curve is for nominal core performance, for all M grades.

In a general sense, if you have 450 volts DC and 30 ma DC across the primary, your maximum AC swing voltage will be 48% of that number X 0.707 for AC RMS voltage applied to the transformer primary, at maximum power. In an ideal situation this swing voltage would cause your peak point on the core permeability curve to be about 8 kilogauss for push pull (PP) and the same for single ended (SE). The problem with SE is that the DC flowing in the primary creates flux that is added to the AC flux for a total permeability and the DC flux provides the starting point. So, for SE you want the ampere turns to equal about 4 kilogauss and the AC signal flux to equal another 4 kilogauss. The reason for the 8 kilogauss peak point is the distortion caused within the window in each of the planar magnetic fields, created within the window for each core layer. Above 8 kilogauss (for all commercial M series iron core material) peak signal saturation, as opposed to RMS saturation, will begin to pinch the linear phase transform between windings and this will cause mid and high frequency distortion. Most commercial audio outputs do drive the peak point beyond this 8 kilogauss point, often out to 12 kilogauss or more based on the assumption that very few audio signals are going to cause a full power signal and that the speakers being driven are going to be distorting far more than the amp.

The entire design of an output transformer is a balancing act. If you use extreme perm core, like amorphous C core, you cannot make use of capacitive coupling or low dielectric constant insulating materials because the result is a Q peak in frequency. Hence you find primaries and secondaries wound side by side in these designs. This does limit the amount of internal gradient information that the transformer can maintain. These types of transformers measure very well, when built properly, are very clear with great dynamics and transients.

Using commercial core requires you to utilize capacitive coupling for all frequencies above 400 Hz in the M series, 3500 Hz for 48% nickle and 7000 Hz for 80% nickle. When properly built these types have superior performance in internal gradient information, are also very clear with very rich transient structures. They do not measure as well as the amorphous core units and have an overall "softer" presentation due to the slightly higher distortion they produce.

Both types will provide excellent musical reproduction, but you need to think about which will suit your particular taste.

Bud

[TerryO]

Originally Posted by BudP

To add to the transformer basic requirements information.
~~~~~~~~~~~~~~~~~Big Snip~~~~~~~~~~~

Bud

Bud,
You practically took the words out of my mouth!

Well, that is, if I understood anything about Transformers.

Transformers: Where the "Black Arts" and Science contend.

Best Regards,
TerryO

[BudP]

Well Terry, as you delve into this smokey space, keep in mind that the requirements for good power transformers are antithetical to those for good audio transformers. Not everyone who starts out to design, build and measure these beasts knows this at the beginning.

Other than that, jump in, not a very deep pool of black magic, at least until you get to the pointy hat grotto.....

Bud

[moonbird]

For me Bud this info is very helpful -- had to read this about 10 times but it slowly is making more sense to me -- it will be a task to get on top of the math.

BTW - some kind soul posted the chapter of the RDH that you refer to here. For those following along, you can get it here (permeability chart and all):

http://www.gyraf.dk/schematics/Radio...ansformers.pdf

Once again thanks.

[Johan Potgieter]

Originally Posted by BudP

Other than that, jump in, not a very deep pool of black magic, at least until you get to the pointy hat grotto.....
Bud

...and the divergent results from different winders for the same design, and ...

[moonbird]

the ancient greeks had a wonderful word for this 'techne" (teknay) ... there is the scientific theory and technology -- and then there is the artful application of it ... making xformers seems to be a great example of this idea.

but comon you other gurus -- don't be bashful here - there is this heavy stuff -- very interesting -- but many great amp makers do their thing without using any of this stuff -- lets hear how you do your black magic too!!!