Transfomer questions - impedance

[rob_gould]

I'm trying to find my way through the minefield of LA2A transformers, and have a couple of general questions.

I understand how the number of turns on pri and sec gives the turns ratio; that is easy.  But what I can't find in any of the stuff I've read is how turns ratio affects impedance.  Indeed, is impedance a product of number of turns, or the turns ratio, or is it related to neither of these things?

For example, The Sowter 4383 is a 1:4 tranny, with an impedance of 600:10k.

The Sowter 1009 is a 1:9 tranny and the impedance is 600:50k, so obviously here the relationship between turns ratio and impedance is not the same.

Would anyone mind just filling in the gaps for me here?

Cheers,

Rob

 

[ioaudio]

the missing gap is  - squared

so a transo with the turns ratio of 1:4 has a impedance ratio of 1:4² = 1:16
600 x 16 = 9600

1:9 turns equals 1:81 impedance
600 x 81 = 48600


-max

 

[rob_gould]

Haha right OK excellent - thanks Max  ;D

So here's the next question :

I want to have a 1:1 input transformer to try and lessen the huge gain of the LA2A.  From what I've read, this should be possible, but of course the impedances are going to be the same on the primary and secondary.

I have read a post by Silent Arts somewhere suggesting that 10k:10k would be the appropriate impedance.  The source for the LA2a will be my soundcard, so what kind of issue will there be having the primary impedance at 10k rather than 600 ohm?

EDIT : By the way the interface is an Echo Audiofire 12, and no output impedance value is quoted on the spec sheet...

 

[emrr]

this is a basic bridging versus matching question.

I'm building mine with the Jensen 10K:10K as per their mod plan, as I happen to have them already.  20K:20K fine, 30K:30K fine, 5K:50K fine.  All these options are out there. 

 

[rob_gould]

Thanks Doug.

Perhaps I'm biting off more than I can chew here because as I'm reading this

http://en.wikipedia.org/wiki/Impedance_bridging

and this

http://en.wikipedia.org/wiki/Impedance_matching

there's lots of stuff I'm not understanding.

One thing that does stick out from the second article is this

With modern audio electronics, impedance matching degrades audio performance;[1][2] impedance bridging is used instead.

I need to do some more reading I think. 

So what you're saying basically is that impedance matching is not important as long as impedance bridging is used correctly?

Thanks.

Rob

 

[ioaudio]

your interface´s output will be probably 50-100 ohm, thats your source impedance (Z source).
it can drive both 600ohm or 10kohm at the la2´s input, your load impedance (Z load), we call that impedance bridging.

Z source << (much bigger) Z load

10kohm input impedance is common/good practice for studio equipment, yet the pultec network or the original la2 have a lower input impedance.
10kohm:10kohm for the la2 is technically correct, many have done it.

you can have either impedance matching OR bridging.

 

[rob_gould]

OK.  Makes sense... I think

Thank you for the responses

Rob

 

[burdij]

Transformers are impedance reflectors. They do not have a characteristic impedance themselves, if we think of impedance as an "AC resistance". The only impedances they have inherently are those produced by losses such as leakage capacitance and leakage inductance and these are usually very small in well made transformers.

The quote you cited from Wikipeadia is a good example of what is wrong with that information source. The whole idea of impedance matching in audio, dating way back to the telco days when broadcasters were trying to send high quality audio signals from a remote location (Carnegie Hall, for example) to the central broadcasting control room (at Rockefeller Center) was to balance the line and match the line to 600 Ohms. This minimizes losses and pickup of noise.

In the case of driving a device that has a balanced input (the LA2A) from a sound card output, you are going from a source that has a few Ohms impedance to a load that, depending on circuitry within the load device, can vary from 600 Ohms up to infinity. Usually the input impedance of a device, even though it says it is 600 Ohms will be much higher. The reason for this is that, even though the destination may contain a "600:600" line transformer, the output of that transformer may have only a 10K Ohm "real" load across it. By real, I mean a 10K Ohm resistance network. That means that if you connect that network to an Ohmmeter or an impedance analyzer, no matter what frequency you feed it, DC or megahertz, the real part of the resistance will measure 10K Ohms. So what will the impedance be that is "reflected" back to the source? It should read 10K Ohms. Of course, the Ohmmeter will read the characteristic resistance of the winding of the transformer because it is measuring DC. But for any frequency higher than a few Hertz, the impedance will be 10K Ohms.

Now if your LA2A has a terminating impedance of 600 Ohms on the input, For example if it has an actual attenuator as the input level control, the reflected impedance would be 600 Ohms. This will present a problem for the little opamp in the soundcard if it needs to generate enough output voltage to drive the input of the following device to a high level, lets say >+10db. But if the following unit has the ability to add gain, you should be good to go.

While the concept of bridging, adding additional loads to a balanced and terminated transmission line, does involve the idea of driving a higher impedance load with a lower impedance source, there was an actual definition of bridging with regard to a 600 Ohm line. Devices that bridged the line had to have an impedance greater than 1500 Ohms. This was particularly important when VU meters appeared on the scene as they needed to "bridge" the terminated line and give an accurate reading of the power level delivered to the load.

 

[bruno2000]

Would a 600:600 transformer have fewer turns, and a lower DC resistance than a 10k:10k transformer?  How does this play into the math?
Thanks!

 

[Speedskater]

Burdij instead of using "characteristic impedance" for transformers,  I would use "inherent impedance" or "intrinsic impedance".  As "characteristic impedance" is reserved for cables or transmission lines of infinite length.

 

[ioaudio]

Would a 600:600 transformer have fewer turns, and a lower DC resistance than a 10k:10k transformer? 

yes, exactly. the 10k will have more turns & more inductance.

 

[PRR]

> The whole idea of impedance matching in audio..... was to balance the line and match the line to 600 Ohms. This minimizes losses and pickup of noise.

Uh, no?

Impedance and balance are different things.

Lowest loss of voltage requires infinite load impedance.

Least voltage pickup requires a zero impedance.

Impedance "matching" is important when audio power is VERY expensive and limited by source impedance.

Early telephone systems had no amplifiers (other than the carbon-mike). The source (microphone) has an impedance. To get the most of this small power out of the mike and into a load, you "match".

Even when vacuum-tubes came, they were not that powerful and VERY expensive. Power out of things generally, and triodes particularly, is limited by internal impedance. Once again, you get the most load-power with "match".

(You might be thinking: tubes have high input impedance. Yeah, so we use step-up transformers for "free voltage gain". What matches? Actually the tube input capacitance at the highest frequency specified, complicated by transformer reactances at high impedance. So while it is not a simple resistive "match", the idea is the same.)

An ideal long transmission line has zero series resistance, zero leakage, zero capacitance, zero inductance. Real lines have some of everything. For low frequency, you "match" the geometric-mean of series and shunt resistance; but anything better than soggy paper has such a broad range between series and shunt resistance that it hardly matters above dozens of ohms. For high frequency you get squeezed between inductance and capacitance, which usually force you toward the 100-1K ohm range.

Very long (relative to a wavelength) lines are terminated to kill reflections. There is "NO" quality-audio situation where reflections matter. The miles of round-trip path-length needed to hear an echo also attenuate the echo.

Without reflections, the "best" way to drive most audio lines is with a low non-Zero source impedance and a very high load impedance. This also leads to simple-splitting: just add loads in parallel like holiday lights. There is a long-broadcast-line convention for 50 ohm source and 1Meg load. In small-studio work we often run 22r-470r source and 10K-50K load.

Since the 1930s, gain has been cheap. And cheaper again in the 1950s and 1970s. "Modern" stuff does NOT need matching, though into the 1960s much stuff clung to match-load convention to replace older gear without confusion. Low-Z sources are standard, though watch the difference between "output impedance" (may be 47r) and "minimum load impedance" (may be 1K).

> I want to have a 1:1 input transformer

The transformer "reflects" whatever it sees, although there is a "design goal" impedance which it works best at.

You want primary Z higher than source Z. Sound-card is likely to be low-Z and able to drive 600r or higher. It may be cleaner with higher Z.

You want secondary Z <= actual load, which is a resistor-pot network around 50K.

600:10K gives 1:4 voltage step-up. Input impedance may be 600 ohms near 20Hz, more like 3K over most of the audio band. Works. 600 ohms is a heavy load for some chips.

600:600 also works, unity voltage.

10K:10K works, and is going to be many-K impedance all across the audio band.

 

[PRR]

> transformers,  I would use "inherent impedance" or "intrinsic impedance".

You are correct that it is not "characteristic", but also not inherent nor intrinsic.

Say "optimum". Or somebody's idea of optimum.

And it gets back to "cost of audio gain/power". If audio power is really cheap, you can force transformer losses. Then you don't have to work near "optimum impedance". But the further off you get, the harder you have to force.

If you work loaded in lower impedance, copper losses become significant.

If you work as-if it was higher impedance, iron losses become significant.

For either mis-use, you could re-design the transformer for better performance/cost ratio. (Of course that is only in a world where custom designs can be pulled from your ear for free; in a real world we take a Standard Part whenever we can.)

Copper-loss does not matter (much) for un-loaded operation.

Loudspeaker Outputs are copper-loss sensitive. I have this here "16 ohm" winding. I want to drive the primary with a lower impedance source and hang 4 ohms on the "16" screw. It works. But the "16" winding probably has over 1 ohm copper resistance. At 16 ohms it is 95% efficient. At 4 ohms it is 80% efficient, 20% lost in the 1 ohm of copper. Even this is not too bad for audio (utility power engineers would be ashamed).

Iron loss on modern iron is small.

Every transformer has at least three reactances which limit frequency response. In hi-Z work (over 10K) you should do math. For 4r-10K nominal windings, loaded, using lower impedance shifts the optimum frequency range down: more bass extension, less treble extension. Un-loaded, you get the deeper bass and still get the treble. So once again the "best" impedance "match" is a very low-Z source and a high-Z load.

 

[Entropia Cub]

EDIT : By the way the interface is an Echo Audiofire 12, and no output impedance value is quoted on the spec sheet...

Output impedance of the Echo Audiofire 12 is 300R (with a TRS jack, balanced), 150R (with a TS jack, unbalanced).

 

[burdij]

Sorry, you are correct, "characteristic" applies to transmission line. I did, however, use "inherent" later in the same paragraph. The point of the statement was to say that the transformer has no impedance value caused by some internal physical process within the device, other than those caused by the small loses due to the device not being a perfect theoretical construct. I often see people assuming that if the manufacturer labels a transformer with specific impedances, those are the only ones that can be used.

I also should have mentioned that a loss of voltage level will occur in a matched transmission line due to resistance of the circuit. But in a circuit where all the losses are minimized, these voltage losses will ideally occur along the vector of the real part of the impedance.

> The whole idea of impedance matching in audio..... was to balance the line and match the line to 600 Ohms. This minimizes losses and pickup of noise.

Uh, no?

Impedance and balance are different things.

Lowest loss of voltage requires infinite load impedance.

Least voltage pickup requires a zero impedance.

And I guess I don't understand the significance of these statements with respect to the discussion of driving a 600 Ohm +4db load with a sound card output at -10db. I think Rob is looking for advice on a real world problem.

 

[[silent:arts]]

PRR, thanks for the great explanation. really valuable as always.

...I think Rob is looking for advice on a real world problem.

just put any 1:1 transformer in, you don't have to order any custom made for this.
you could look at the input transformers I measured for the D-AOC, similar application.

 

[CJ]

i would say to someone who fell off the hay wagon and broke something,

yes, you want to reduce the step up ratio to make the control work better.
but what does this mean to both
a) the sound card
and b) the la2a?

a)the sound card - look at the output impedance rating of the card and multiply by 10 for your pri impedance of the matching transformer.

b)la2a- make sure that this impedance, which will be the same  on the secondary due to the 1:1 ratio, will not bother the la2a.

what would bother the la2a?
well, you could screw up two things:

1)you could not have enough power to drive the 1 meg grid impedance, this would mean having a transformer with a much higher impedance than 1 meg, say 10 meg?
have you ever heard of a 10 meg transformer?
ok, then.

that leaves

2) T4 module. if you click on the link in my sig line, you will see a lot of resistance paths going to ground.
LDR cells, 100K pot, 68k input divider, blah blah blah.

well, instead of doing a bunch of math, just make sure if you use a 600:600 transformer that the much lower dcr does not mess with the controls.
it probably won't, but since all transformers are differently rated by the winders ( i have seen some real dogs out there) , just do an A/B test with the 600 ohm sec vs the 10k sec.

if the 600 ohms does not short out all the compression action, by all means, use it, as the frequency response will be much flatter than a 10K winding.

 

 

[rob_gould]

Hi All,

Thanks for the responses - some fantastic info in here.  Not had a chance to digest it yet as I've been away for a few days, but I'll look at it tonight. 

Cheers,

Rob

 

[s2udio]

some fantastic info in here.  

+1

This therad has cleared a lot of "grey" areas of this subject for me........good stuff !

Should be added to the meta ............

 

[CJ]

i should mention that maybe not the dcr so much as the inductance of the 600 ohm winding.
if the 600 ohms is really 100 ohms due to whatever, this is what may cause drop off of signal or compression control variances.

some people read the dials, most just adjust by ear since the cells change from day to day.

just set it for "frank" and leave it alone.

or set it for "bing; if you are using a ribbon.

hey, has anybody tried the ua610 into the ua la2a?