Audio Transformer Inductance

[ricardo]

square wave is good for B-H loop testing

Got any references or description of how to do this?

zobels, well, never seen a really big change in sound quality due to adding a zobel ... most just squash one peak into two or three peaks of lesser amplitude

It would have to be a really badly designed Zobel to do that.

I've used a lot of Zobels in mikes and mike preamps in my Calrec days.

Good transformers so the HF resonance is always supersonic.  The Zobel is not to affect the audio band but to avoid a peak at low RF frequencies.  You have to do this with certain forms of RFI protection too.

BTW, I don't damp HF resonances to Q=0.5 -6dB@resonance .. which is what you get when you damp for "no overshoot" on square waves.  I aim for Q=0.7 -3dB@resonance and no response peaking.

"no overshoot" is just a convenient way to get near the correct damping if you can't measure supersonic frequency response.

 

[Marik]

For ribbon transformers, it should be obvious that the Samar Audio transformers represent a significant advance in the state of the art. 

8) 8) 8)

They have larger Primary L,

No they don't. There is no reason to have larger Pri. L than it should be for a given bass response for a given source resistance. In other words, each physical size of the ribbon presents a different source impedance, so we believe rather in optimized inductance for a given ribbon. That's why we are going into significant production complications of carrying many different ratios to perfectly match MOST of the ribbons out there. This is a much better idea as opposed to "one size fits it all" larger Pri. L and single ratio, which leads to improper impedance matching, unnecessary increase of DCR and noise increase (after all, for each single Pri turn there will be some 30-50 Sec turns).

Lower DCR

Indeed, it is. The usual values are some 0.002 Ohm for Pri, and some 3 Ohm for Sec, going up to only 10 Ohm for very high ratios.

and higher "undistorted" output in a smaller package.

Here it is important to define what is "undistorted", "higher than what", and "what kind of distortions" we are talking about?

Definitely, 1 mil tapewound toroidal cores we use have very different mechanism and exhibits much lower losses/distortions than say, DI, EI, or C cores, of the other transformers you mentioned.

As far as SPL handling concerned, yes, every transformer we offer comes in two sizes of the cores--bigger (about 27mm diameter for "no size restriction" applications), and smaller ones (to fit smaller 22mm barrels). The difference between those is  160dB vs. 135dB SPL handling, respectively.

Hope it helps.

Best, M

 

[CJ]

now  those copper buss bar 1 turn primaries,

do they have to be in a perfect circle?

or is square ok?

what about the breakout leads, you do not want 3/4 turn, right?

so copper must almost be touching itself for 360 turn, less distortion?

when does a turn become a turn?

will 3/4 turn still work, what is turns ratio with 3/4 turn primary jus curious?

 

[trobbins]

Leakage inductance is introduced when a turn doesn't complete 360 deg in a tight format where flux is constrained to core. 

 

[PRR]

> when does a turn become a turn?

For iron-core with windows: there's no turns. Only "passes through window".

Try it. Find a small PT where you can put some thin wire around the winding.

Put "exactly one turn" on. Energize. Watch the 1-turn with a good 'scope. You may find "0.1V per turn".

Now start taking wire off. 6/10th of a turn through two windows, you still have 0.1V. 6/10th of a turn but in just one window, you have 0.05V. 1/10th of a turn in one window, still 0.05V.

Toroid goes by passes through hole. You don't even need "a turn": a straight bar through the window will induce the 0.05V.

We count "turns" because it's easy and we typically have hundreds of turns so a +/-0.5 turn difference is un-noticed (also un-likely for most lead-out systems). But we should count "window passes".

Oh... as trobbins says, leakage inductance is greater if you don't go all the way around. For good iron this has 1/1000th effect on bass inductance. The 1/10th turn induction is not 0.05V but 0.04995V. Insignificant for power or bass. If you really need a "1 turn" with extended treble, it might matter. However such low-low impedances usually do not have real problem with leakage inductance. In a ribbon mike, the "long" leads from transformer to ribbon may have as much stray inductance as the transformer.

 

[trobbins]

For clarification, a turn always does 'go around' for a circuit which passes current - even the conceptual diagram of a single long straight wire passing through a toroid - the the wire carries current then the circuit 'turn' has to be closed somehow !

 

[CJ]

that turn can be very big,

hang a current transformer on a 10,000 amp bus bar,

that bus bar might be long, if it is a big building,

for current to flow the loop must be connected, yes,

but the loop can be weird lookin, like michael j near the end,

is bubbles still alive?

 

[ricardo]

Morgan Jones in "Valve Amplifiers" 3rd edition shows the HF equivalent circuit with the winding capacitances in 2 places such that the HF filter is 3rd order, 18dB/8ve.  Fig 4.21

My experience, all with mikes & mike preamps, is that it is 2nd order, 12dB/8ve.  ie Radio Designers Handbook 4th ed Fig 5.10D

Comments?  Experience?

 

[ricardo]

from Radio Designers Handbook 4th ed.  Chapter 5

 

[trobbins]

3rd order may only kick in well above SRF, depending on equivalent values.

Capacitance re winding is likely to be really a distributed capacitance, rather than lumped, so pidgeon-holeing the characteristic is probably not a good way to go. 

 

[ricardo]

3rd order may only kick in well above SRF, depending on equivalent values.

Capacitance re winding is likely to be really a distributed capacitance, rather than lumped, so pidgeon-holeing the characteristic is probably not a good way to go.

Yes.  The stray (winding) capacitance is distributed.  RDH chapter 5 goes into great detail on this.

But the equivalent circuit allows us to treat all these as lumped components.

My own experience suggests the RDH equivalent circuit is very representative and useful both for design as well as analysis.  Mr. Jones obviously has different experiences.

Just wondering if anyone here has looked at transformers and found signs of 3rd order HF roll-off.

 

[trobbins]

There are response differences between a distributed circuit and an equivalent lumped circuit which blur the region around the corner/peak frequency - which is I guess the region of most interest?

 

[ricardo]

Some important caveats about measuring HF performance of transformer ..

http://www.tubecad.com/october2000/  High Frequency Anomalies in Output Transformer Tests - John Atwood

They don't affect measuring Leakage L by the common shorted winding method but confuse Self Resonant Frequency.

Biggest effect in large Valve Output Transformers.  Small line & mike level transformers, IME & HO will show the simple 12dB/8ve behaviour as the 2 examples I posted.

 

[abbey road d enfer]

I question the MJ model.
My understanding is that DCR, leakage inductance and parasitic C are all distributed.
Then how can one make a distinction between C1 and C2? I believe it all comes lumped in one LRC equivalent.

 

[bockaudio]

is bubbles still alive?

Turns out Bubbles, and the ferris wheel, were rentals.

 

[emrr]

Just wondering if anyone here has looked at transformers and found signs of 3rd order HF roll-off.

http://www.groupdiy.com/index.php?topic=31454.0

 

[CJ]

are we talkin about the 20 k hz and above region or 20 hz and below?

what do you mean by 3 rd order roll off?

you mean the fundamental wave rolls off at 20 k and there is also a dip at 60 k?

i had some weird stuff on the low end, turns out it was caps in the scope forming a res filter at 15 hz, which gave a bump in the bass region, i thought i was on to something but all i got was a slap down from PRR, 

caps in the scope problem was solved by going for a dc input which removes the series cap which passes ac  to the scope but not dc, just in case it were on the same line as the ac, this is called a dc offset,

now at ultra low freqs, you have a really  low impedance ,

take a 10 henry primary at 10 hz and you have 2 pi f L reactance which is

6.28 * 10 * 10 = 6.28 * 100 = 600 ohms,


so weird waves tend to get smoothed out, noise is not a problem,most of the weirdness is at 20 k hz and above, however, magnetizing current waves start to jump out of the woodwork at the low freqs, so watch out for those, best way to get rid of those is to slap a load on the sec and pass some current thru the trans, thus your ratio of mag current to signal current is much better,

at the high end, the transformer jig is lookin at 6.28 * 20,000 * 1 henry (core loss)

so you have a 6 * 20,000 = 120,000 circuit, so 600 to 120 k ohms, quite a range of operation with those transformers,

that is why you have to keep fine tuning the signal generator voltage when sweeping a transformer, the load variation on the generator can cause regulation problems,

 

[abbey road d enfer]

what do you mean by 3 rd order roll off?

3rd-order as in -18dB/8ve slope

 

[CJ]

standard Bessell curve?

 

[ricardo]

standard Bessell curve?

Whether it's Bessel, Butterworth, peaked or overdamping is all dependent on the damping.

It's difficult to see if the asymptotic roll-off is 12dB/8ve or 18dB/8ve unless you can accurately plot the response well above the Self Resonant Frequency.

But the Phase Response at the resonant frequency is conclusive regardless of the damping.

12dB/8ve gives 90 degrees at resonance.  18dB/8ve gives 135 degrees.  All you need is a dual trace scope and an oscillator which goes just beyond the resonant frequency.