5k:8 ohm vs. 10k:16 ohm output transformer. What is the difference?

[Austin]

What is the differance between an output transformer 10k into 16 ohms vs. a 5k into 8 ohms? The dc resistance of the primary?

http://www.edcorusa.com/p/544/gxse15-16-10k

http://www.edcorusa.com/p/537/gxse15-8-5k

[Enzo]

A transformer works on turns ratios. That primary impedance ONLY exists if the specified load impedance is across the output. The transformer haqs no inherent impedance. In other words, you cannot hook a meter up to it and measure the impedance. Think of it as a set of gears. There is a ratio between input and output, so if a set of gears is rated 2000rpm in and 20 rpm out, and you instead spin the input at 4000rpm, then the output will turn at 40rpm. The gear has no inherent speed, only that relation to the output.

So the difference between two transformers like that can be nothing more than the load connected to them.

Transformers do not operate on resistance. One needs to consider winding resistance for power losses and stuff, but it has nothing to do with the impedance.

The thiner the wire used to wind it, the higher the resistance will be, regardless of the number of turns. On the other hand, a 50:1 turns ratio could be 50 turns and one turn, or 5000 turns to 100 turns. If the same wire is used, then the one with the higher turns count will have higher resistance, even though the turns RATIO is the same.

[R.G.]

The primary inductance, the leakage inductance, and the wire resistance. Probably also the frequency response.

A quip I think I originated and am fond of is that transformers don't have impedances, they have ratios. A transformer having a 5K:8 transformation ratio will make 8 ohms on the secondary look like 5K on the primary. It will also make a 10 ohm load look like (5K/8)*10 = 6250 ohms and 16 ohms look like 10K.

So how's that different from one specified at 10k:16?

Frequency response. Transformers are designed for a certain minimum primary inductance. The primary inductance "steals" signal current from the primary to operate the transformer. Otherwise, they would not work at all. Transformers are designed so that the primary inductance eats only a certain amount of current at the lowest freuqency the transformer's designed for. So if the transformer is designed to pass for example 40Hz at a loss of half the primary current when loaded with, say, 5K transformed impedance from the secondary, then the inductance has a impedance of 5K at 40Hz too, because it eats half the available current. This lets one calculate the inductance as L = 5K/(2*pi*40Hz). And indeed, this is one of the first design calculations a transformer designer does. It sets the low frequency response. All the rest of what they do as far as windings, turns, wire size, etc. have to get enough turns into it to make this inductance or more come true.

If you have a transformer designed for, say, 5K primary impedance, and you put twice as big a load on the secondary, then the load seen at the primary is 10K as far as the transformed load goes. But it has the same old primary inductance, which eats the same current at the same signal voltage, even if the secondary acts like 10K. So it works fine at "mid band" - frequencies far away from the low frequency or high frequency rolloffs. But the primary inductance will eat more of the available drive current so the low frequency response is worse.

If you designed for 10K:16 and load it at 8 ohms, so the primary sees 5K, then the low frequency response gets better because you had to design in a bigger primary inductance in the first place to reach the desired low frequency response for 10K.

So why not use the 10K:16 even when you only needed 5K:8? First, you had to put more turns in, use finer wire and possibly a bigger core to get the right inductance for 10K, so you may be paying more for it. And you certainly needed more turns, so the leakage inductance, which is what limits the high frequency response, went up at the same time, so you get better low end but worse top end. As well as likely paying more. And probably getting lower power handling at the same time.

These things are the gotchas at high and low frequencies. In the middle, they don't hurt you much. And sure enough, using a slightly wrong tap/impedance/load on a transformer does make it act like a tone control, changing the balance of highs and lows. As well as changing the amount of power those power tubes can get through it.

Transformer design is not terribly complicated, but there are a lot of factors to think about, and substantially everything is a variable. There aren't many fixed places to stand. So to do a design, you simply pick some things based on experience or intuition, and charge ahead. Then you check back to see if you made good guesses, and iterate.

[R.G.]

Darnit Enzo, you and I have to quit posting at the same time.

8-)

[Enzo]

Yeah, but yours was better.



Then the cynic in me wonders: they CAN design them differently, but DID they?

Looking at the two products, the materials are the same, the impedances are listed, but only one of the two transformers includes primary inductance or even DC resistance. Hard to tell then.

[R.G.]

Yeah, but yours was better.

I don't know about better. It -was- longer anyway.

Then the cynic in me wonders: they CAN design them differently, but DID they?

Looking at the two products, the materials are the same, the impedances are listed, but only one of the two transformers includes primary inductance or even DC resistance. Hard to tell then.

Your cynic and mine agree. Transformer designers get lazy too. And their MBA-diseased management sometimes helps them "synergize" on designs.

Sometimes a lot.

[imaradiostar]

Sorry to dredge up an old topic but I figured you guys would appreciate this since it seems a departure from big business. I've spoken with the Edcor designer on the phone at length and he's an older fella and making transformers, for him, is a labor of love. His son is running the place and keeping it profitable while he and his wife order and build all of the one-offs and custom designs. I thought that was pretty neat. He's a fascinating guy to talk to if you ever need to order something from them.

Jamie

[Xrunner9316]

So by how much does the high frequency response decrease with 10K:16 transformer?

[R.G.]

So by how much does the high frequency response decrease with 10K:16 transformer?

In the words of the bard, aye, there's the rub. It - wait for it - varies.

First some generalities. Transformers in general deliver their best frequency responses with low impedances and low-ratios.

There are a number of reasons for this hidden back in the transformer modelling and design process. The fundamental bottom line on that is that the primary inductance must be bigger for the same low-frequency response if the input impedance is higher. That means more turns or a bigger core, or both. More turns increases the leakage inductance at the same rate that it increases the primary inductance. So increasing the primary impedance to get the same low frequency response decreases the high frequency response by increasing leakage inductance at the same time.

One decreases leakage inductance by clever winding tricks of interleaving and sectionalizing, and sometimes tossing the whole design out and changing from E-I cores to C cores to double C cores to toroids, to pot cores, etc. And this last is why how much the frequency response changes varies.

For a real design change (not the cynical one Enzo and I referred to), doubling the primary impedance level and keeping the same low frequency response requires a 40% increase in turns - 1.414 times, actually - to double the primary inductance. This doubles the leakage inductance and at the same time more than doubles the primary and secondary wire resistances. If the original transformer had simple windings without interleaving, the leakage can be reduced by interleaving. If it was already interleaved, it can be interleaved into more sections if that's practical. But if all the tricks have already been used in winding, only radical redesign for a different core remains as possible.

So it's not easy to say by how much the high frequencies get dropped.

[Tony Bones]

Very loosely, two transformers with similar design (core, interleaving, etc.) will have about the same number of octaves of bandwidth. So, suppose we take a transformer designed for 5k:8 and rewind it to have twice the primary inductance to be suitable for 10k:16. The entire usable bandwidth will be shifted down one octave.

[tedmich]

the windings I should suspect!


(already been used?)

[R.G.]

Very loosely, two transformers with similar design (core, interleaving, etc.) will have about the same number of octaves of bandwidth. So, suppose we take a transformer designed for 5k:8 and rewind it to have twice the primary inductance to be suitable for 10k:16. The entire usable bandwidth will be shifted down one octave.

That is quite accurate - including the "very loosely'. I can't tell from the OP's post whether he has one and wants to sub for the other, or is just doing theoretical comparisons. If one started out to make two transformers, one high and one low, and carefully wound them so the wires occupied nearly exactly the same spaces inside the winding window, and had the same interleaving, they would be close.

However, two trannies from even the same line of transformers are not necessarily going to have those requirements paid any attention to.

As Yogi Berra is said to have said, in theory, there's no difference between theory and practice; in practice, there is.

[loudthud]

I looked at the two transformers linked in the OP's post. First of all they are for single ended operation both rated at 15W and weigh 2.75 pounds. From there the specs are incomplete. Both seem to be rated at 100mA primary current. You would expect the 10K primary to be wound with more turns and smaller wire. This should equate to a lower current for the same flux at idle. The frequency response of both units is rated "40~18K Hz., <1dBu" whatever that means.