Need a refresher on transformer power handling when Primary Z is halved or doubled

[imaradiostar]

I'm toying with using a 100 watt 6k6 primary output transformer as a 3k3 primary transformer so I get 2, 4 and 8 ohm taps rather than 4, 8 and 16. It'll be used with a pair of KT90's or KT120's. I forget if halving primary impedance like this tends to extend or reduce low frequency response. I recall a chart and an equation in RDH4 but I don't have time to look it up.

Please forgive my laziness.

jamie

[imaradiostar]

I did a little reading. This is how I understand it, please correct me if I'm wrong. This is an oversimplification that ignores DC resistance and copper losses and all of that.

A transformer can only take so much voltage on any given winding before it saturates and leaks magnetic flux. It takes x amount of voltage to make y amount of power so if the reflected impedances are lower there won't have to be as much voltage to achieve a given power level.

So...a transformer that has a -3dB point of 80Hz at 50 watts into 8 ohms will have 20 volts (rms sine) on the secondary and 450 on the primary. This assumes that the transformer starts to saturate around those voltages. By using the transformer as a 2k:4ohm transformer (and assuming I have tubes capable of this) the maximum voltages remain the same but I'm now putting 20 volts (rms sine) into a 4 ohm load, so (20*20)/4=100 watts of output power.

By extension- when used as 2k:4 ohms, this same theoretical transformer should now have a -3dB point of 40 Hz.

Do I understand this correctly?

jamie

[Merlinb]

Jamie is partly correct. If the only thing limiting the transformer is saturation, then halving the impedance will allow you to put twice the power through it for the same bandwidth, or the same power but with twice the bandwidth.

However, transformer power is also limited by copper losses, which is related to the current in the windings. Whatever impedance you choose the wire is still the same. If the transformer is designed to operate at its current limit then halving the impedance means you can't increase the power, but you can put the same power in at frequencies one octave lower than before...

Whether the transformer reaches its saturation limit or current limit first depends on its design, so you're in the dark really, unless you go to the trouble of actually testing it. The only safe assumption is that you can have the same power but twice the bandwidth.

[R.G.]

Put another way:

The primary inductance is unchanged when it is loaded with half the impedance. So the low frequencies are extended compared to the higher load, because the primary inductance's impedance level which eats half the available power by matching the transformed load resistance is lowered. Low frequency response is almost totally determined by the primary inductance and the loading level reflected from the secondary.

You did note that the tubes may not be able to deliver more low end power, as they have power limitations of their own in the form of the available voltage/current they can supply.

A transformer can only take so much voltage-time integral on any winding before it saturates. It's not the voltage - it's the volt-time integral. If it saturates at 300V at 80Hz, it will saturate at 150V at 40Hz and 600V at 160Hz. So saturation is very, very much a bass frequency phenomena. Resistance of a transformer to saturation goes down linearly as signal frequency goes up.

The saturation point and the low frequency response -6db point are only tangentially related, not directly. I'd have to go back and do a ton of math to see if they are any more tightly related. Maybe, but I don't remember ever running onto that. A transformer that has a 50W power specification and a -3db low end rolloff of 80Hz doesn't really tell you much at all about where it starts to saturate. All you know is that any sane designer would put saturation significantly lower than the bottom of the specified bandwidth. They don't relate directly that I know of. I need to go dig that out.

However, they still do lurk together. By running a 4K:8 at 2K:4, the low frequency response point is moved down by about 2:1. This may or may not run you into the saturation point if you use that low frequency extension. Because you have the same voltage and are trying to send in half the frequency, you're doubling the volt-time integral on the primary and might start running into saturation *if you feed it signal at the bottom frequencies and maximum voltage".

Same voltage into half the impedance is double the current, so the copper losses go up 4X. This may or may not run the transformer into overheating. It depends on how close to the edge of its insulation heat rating it was at its specified power level. No good way to tell this if the maker doesn't tell you and you don't want to tear it down.

[J M Fahey]

Saying about the same in other words.
1) since the inductance stays, you *should* have a lowered point where inductive impedance of the transformer itself (unrelated to secondary *reflected* impedance) , which is in parallel with the reflected one, starts robbing you of current and power, meaning you should have somewhat better low frequency response.
That should hold true for *low level* signals, but for high power ones, even if the parallel inductance is less of an enemy now, I don't think the core will like the increased current plus lower frequencies *at all*.
That combination alone should negate any power handling increase, and then some.
2) parasitic/series inductance will kill your highs.
3) all the above takes a second seat behind doubled resistive losses, which will eat you alive.
To boot, copper has a thermal positive (and quite high) resistance coefficient, meaning losses will heat it up more than normal, *increasing* losses even more.

Anyway test it, if it was tightly specified, all gremlins will raise their heads; if it was oversized or overspecified to begin with, you might have very acceptable results.

Although I would personally use it at or around rated specs, OTs are not very forgiving beasts.

[imaradiostar]

So...I might be losing enough to copper losses that it isn't worth it anyway. Still, as things tend to go with tube amps it's worth trying it to see how it sounds.

Specifically I'm looking at Edcor transformers. They seem to be a real bargain and the specs look really good. Their 100 watt transformers weigh in around ten pounds and are spec'd from 20Hz-20kHz.

Suppose I were to forgo the impedance halving mess. Is it reasonable to say that a transformer capable of 20 Hz will hit other walls (copper, saturation, etc) trying to deliver 150 watts at 40 Hz?

Jamie

[J M Fahey]

Well, the 20Hz rating is certainly a -3dB spec , so all they claim is it's delivering half power there.
So its "100W" rating really means "50W" down there.
So I'd forget about 150W at 40Hz; I'd be happy if it at least could get close to te official 100W rating.
Now, if you talk, say, 1 KHz, yes, you should be able to stretch the power a bit.
I'd use those Edcor transformers at their rated specs.
6600 ohms too high for 100W? Raise the +B accordingly and you're there.
Just keep your screens around safe 400V and you will do very well.
JM2C

[imaradiostar]

It's -1dB at 20 Hz so I've got a little more wiggle room.

Wouldn't it have to be -6db to be half since we're talking power and not voltage?

This is a little silly anyway- who really needs an extra 50 watts above a 100 watt amp? I was only trying to buy a cheap component so I could experiment with KT90's and KT120's.

jamie

[R.G.]

If all you are after is experimenting with KT90s and KT120s, go ahead. Just limit the low frequency you feed to it and you're fine. Magnetic power handling goes up with frequency in transformers. If it's -1db at 20 and 100W midband, it's 90W at 20, so it's *magnetically* about 180W at 40(roughly). *Magnetic* power handling and saturation are very much low end issues.

The next limit on power handling is thermal, so the combined copper losses and core losses are what heat the windings above the insulation breakdown temperature. Insulation temps are completely independent of the magnetic or thermal losses, so where it breaks down depends entirely on what insulation is used. The iron works OK up to the Curie point of the iron (550-770C depending on alloy),and the copper works until it melts at over 1000C; transformers are happy to keep working at temperature which will fry eggs and give blisters to fingers as long as the insulation doesn't break down. Although the copper losses do go up as the resistivity of the copper rises with temperature.

So copper losses will dominate at midband for increased-current/power operation. At the high end, you run into eddy current core losses going up dramatically. this can be addressed by limiting bandwidth to lower frequencies. Copper losses are (disregarding skin effect) frequency independent.

[imaradiostar]

(also replied elsewhere)

OK, wow!!! I just spent 30 or 40 minutes on the phone with Larry at Edcor. He said I'm welcome to share what we talked about so I'll share some of the things I learned!

Larry does all of the design work on their audio transformers. He's a 50 year veteran of the industry. He's worked with Peerless and Altec and has a wealth of information.

As for his transformer designs, he really creates the best product he can all the time. He strives to completely fill the window on any EI lam transformer so his designs generally have oversized copper and can be taken well above the ratings on the website. I asked specifically about using the CXPP100-MS-6.6K as 3.3k to 8,4, and 2 ohms and he says it's designed to take more than 150ma per primary half and should easily produce well over a hundred watts from 30Hz up- so I'll be ordering one!

He also said they're limited to 100 watts for production ouput transformers because of end bell sizes. With the windows more or less completely filled he can't simply increase the stack size to increase the power and bandwidth. I asked if he's considered a Partridge style end bell and he's open to it so I'm sending him some pictures and information. He does have the larger lams for certain products but they're limited by mounting and endbells and it's important to them that they maintain a certain look and professional style.

He says that their audio/tube business is growing every day and they're open to new products and ideas. He loves his job and is happy to try custom and one-off designs, though they generally are built on weekends while most major production items for big customers are built during the week.

I also asked specifically about a transformer for experimentation that would have 3, 4 and 5 k primaries and multiple secondaries of my choosing. He says it's really not a problem at all and he welcomes doing custom work such as that.

I'm looking forward to trying some big Edcor iron!

jamie

[trem]

I have bought 5 Edcor OT's during 2011 (for repairs/new builds of other peoples amps) and I have been impressed with their service. They seem to make a good product. Not saying they are any "better" than Heyboer, Mag Comp, etc......just that Edcor seems to be another option on par with other quality tranny makers we are familiar with.....
Good Luck