transformer impedance question

[Nikolay]

Hi to all. I have a small signal transformer, but at this time this transformer is unknown for me.

I start with turn ratio measurement: The transformer have 1:5 turns ratio. This is 1:25 impedance ratio.
This transformer can be either 600:15k and 200:5k and so on.
The inductance at 1kHz with open secondary is 38.2H for the primary and 1.43H for secondary. At this time there is no matter which is primary and which secondary.

As we know the transformer impedance is variable and depend by what the transformer see from the other side, but what's the best input and output impedance?

I don't know how to measue the real input (or output) impedance. If I know this, I can use this transformer for the proper usage.
I have 2ch scope, LCR with fixed frequencies: 100Hz, 1kHz and 10kHz, multimeter, signal generator and of course calculator I don;t have THD analyser, but I can use spectralab software for some cursory measurement.

I'd like to ask you what's the accurate procedure to measure this transformer to identify it.

Thanks, Niki.

EDIT: I forget to tell you that the DC resistance for the primary is 1.18K and for the secondary 54 ohm

 

[emrr]

This is all practical observation, and no math.

99% of the time 54 ohm DCR corresponds to 600 ohms.    That can be considered a reasonable initial benchmark for fast evaluation.  There
are exceptions, but rare. 

Core size will dictate level handing abilities, and possibly define it as input or output usage. 
Size should give an idea, if open frame.  If potted, a larger size can be deceptive, and hide the actual core size. 

Inductance will dictate lowest available frequency, and will in turn suggest acceptable real world impedances.  It will perform better at lower impedances, and worse at higher, which should give a clue.  If too low in impedance, you may see resonance boosts at extreme lows and highs, and this will tell you that it is operating out of intended range. 

Does your LCR give AC resistance measurements?  If so, you can take some comparative measurements with varying secondary loads, and that may inform. 

 

[Nikolay]

Thanks emrr for the answer. Everything you write I think is based over your effort. I don't have this effort with the transformers. I can't see what's inside the can. I try to open it without cutting, but I can't It's hard closed unfortunately.
Here is the picture of the transformer.


It's small, looks like a little bit bigger than OEP transformers. I can't see what's the lamination and how they are placed

Yes, my LCR can measure AC resistance. When I connect 600ohm at the low impedance wiring, I measure 16-17kohm impedance at the high impedance wiring depend by frequency.

If you are right, this mean that the transformer is 600ohm:15kohm.

Is there any other way (not indirect suggestion) using measuring the transformer to get real values? I'd like more to teach myself, not only to get some values of this transformer.

I read this article:
http://www.twin-x.com/groupdiy/albums/userpics/Transformer_Articles_1.pdf
But I can't read the formulas correctly due to low quality of the scan, but I get some strange values with the calculator.

On the page 3, there is a formula: Z = sqrt ( R^2 (this is the DC resistance) + Xn^2 (net reactance))

But if net reactance is equal to inductive reactance, I see that the wiring's impedance depend by frequency. In real world this is normal, but with these formulas I get really non typical values, and I was confused

 

[emrr]

Definitely an input based on size and shielding alone. 

All evidence points towards 600:15K.

Impedance versus frequency will definitely twist your mind.  As with speakers, it's a nominal figure, rather than absolute. 

I think it best to try it in some simple experimental circuits, and see where you like it.  Try it in front of a 5532/4 chip, etc.  If you have a simple tube preamp, connect it to the input grid and try it there.  Lots of possibilities. 

 

[PRR]

An audio transformer does not have a "real impedance".

It reflects impedances, with several added un-wanted impedances.

We basically steer between the unwanted impedances.

Series DCR is "bad", so we like it to be 5% of design audio impedance. This suggests 1080:26600. However "great bass extension" may suggest DCR as high as 10% of design audio impedance. 540:11800.

So 600 ohms is a good bet.

Now look at inductance versus estimate impedance. 1.4H at 60Hz is 540 ohms. So above 60Hz it may be 540 ohms or higher; below 60Hz the impedance falls from under 500 ohms at 60Hz to 180 ohms at 20Hz.

So if we believe your LCR meter (they can be confused by iron cores), this is more likely a very good telephone transformer (-0.5dB at 300Hz, -10dB at 20Hz) than a music transformer).

The next step is to use a true 600 ohm source and drive it at very low level, sweep down from 1KHz. If it is really 1.4H, it will be sagging 3dB at 67Hz. However it may be several times better.

Working at 150 ohm impedance, response will extend 4 times deeper, past 20Hz. However 150 ohm circuit with 54 ohm of copper resistance suggests significant loss of signal. That would be bad in an output transformer or in a mike transformer. In a line input transformer, some loss may be very acceptable.

Now sweep up above 1KHz, both with "estimated" load and with no-load. The response will fall at some high frequency. It may peak first. It will be different for different secondary loadings.

You should vary source and load impedances, looking for some "best fit" to your idea of "the audio band".

You can never know what band, and thus what impedances, the original designer aimed at.

 

[Wonderlandaudio]

An audio transformer does not have a "real impedance".

It reflects impedances

Lots of people still dont understand this concept. Transformers "transform". Thanks PRR for being so clear

 

[hobiesound]

(dcr and a little extra) x 10 , is usually the recommended impedance. At least that's what i use for unknown transformers.

Grtz

thomas

 

[hitchhiker]

Great thread !  It would be very interesting on this subject to read some explanations specific to tube and sand state class A and then 5534 or 990 etc for line in and out.   

 

[ruffrecords]

Hi to all. I have a small signal transformer, but at this time this transformer is unknown for me.

I start with turn ratio measurement: The transformer have 1:5 turns ratio. This is 1:25 impedance ratio.
This transformer can be either 600:15k and 200:5k and so on.
The inductance at 1kHz with open secondary is 38.2H for the primary and 1.43H for secondary. At this time there is no matter which is primary and which secondary.

As we know the transformer impedance is variable and depend by what the transformer see from the other side, but what's the best input and output impedance?

As a rule for audio work you want the reactance of the winding inductance to be twice the source impedance at the lowest frequency of interest (where the response will be 3dB down). At 20Hz the reactance of 1.43H is nearly 180 ohms so with a 150 ohm source you will be about 3dB down at 20Hz. So you could use this as a 200:5K but  as a 600:15K it will have a poor LF response.

Cheers

Ian

 

[lassoharp]

As a rule for audio work you want the reactance of the winding inductance to be twice the source impedance at the lowest frequency of interest (where the response will be 3dB down). At 20Hz the reactance of 1.43H is nearly 180 ohms so with a 150 ohm source you will be about 3dB down at 20Hz. So you could use this as a 200:5K but  as a 600:15K it will have a poor LF response.

Does anyone know how the rule of thumb that Ian stated translates to cases of SE small to med power pentodes (high source impedance) into typical loads of say 3 to 5K.  I ask because for good bass response it would seem to require a very high inductance (like ~200H) on the primary, and would also have to maintain that inductance with a large amount of unbalanced DC.  Is this correct?

 

[mjk]

As a rule for audio work you want the reactance of the winding inductance to be twice the source impedance at the lowest frequency of interest (where the response will be 3dB down). At 20Hz the reactance of 1.43H is nearly 180 ohms so with a 150 ohm source you will be about 3dB down at 20Hz. So you could use this as a 200:5K but  as a 600:15K it will have a poor LF response.

Does anyone know how the rule of thumb that Ian stated translates to cases of SE small to med power pentodes (high source impedance) into typical loads of say 3 to 5K.  I ask because for good bass response it would seem to require a very high inductance (like ~200H) on the primary, and would also have to maintain that inductance with a large amount of unbalanced DC.  Is this correct?

Since the winding inductance is reactive, it's a 90 degree vector normal to the signal, meaning the voltage ratio will be -3dB when the X(l) and source resistance are equal.

Your estimate of required inductance is in the ballpark, assuming pentode stage effective impedance = plate resistance in parallel with the reflected load. Good luck achieving this with a SE OPT. 35-40H is more like it. Fortunately your pentode has high transconductance. Meaning that a small change in grid voltage can produce a large change in plate current. So a little feedback can make the pentode act like a low impedance source, which is what you need to do to get flat voltage response down into the bass range with a SE pentode amp.

Cheers,

Michael

 

[lassoharp]

Good luck achieving this with a SE OPT. 35-40H is more like it

Exactly. 

I knew that NFB would factor in here and it seems it would work ok on a 10X reduction scale for some pentodes.  One other particular circuit that came to mind was the Langevin 111/WE120 where it looks like you'd have to be lowering effective 6SJ7 plate resistance from 1M or greater to a very low -say 3.5K (assuming 40H) source impedance to get around a 30Hz corner.  I'm likely missing another factor somewhere in trying to break this down - maybe not true pentode operation in this instance?


http://www.audiosharing.com/archive/western/we_amp/pdf/No.120-A.pdf



 

 

[mjk]

Good luck achieving this with a SE OPT. 35-40H is more like it

Exactly. 

I knew that NFB would factor in here and it seems it would work ok on a 10X reduction scale for some pentodes.  One other particular circuit that came to mind was the Langevin 111/WE120 where it looks like you'd have to be lowering effective 6SJ7 plate resistance from 1M or greater to a very low -say 3.5K (assuming 40H) source impedance to get around a 30Hz corner.  I'm likely missing another factor somewhere in trying to break this down - maybe not true pentode operation in this instance?


http://www.audiosharing.com/archive/western/we_amp/pdf/No.120-A.pdf

Oops, I automatically assumed a power amp. With a 600 ohm OPT I think one can tolerate more DCR and thus get higher primary inductance of ca. 100H for reasonable standing current of ca. 20 mA. You would still need feedback with a pentode.

The WE120 is a classic feedback scheme from output plate to driver cathode. This is an excellent circuit with a simple feedback path to illustrate the effective plate resistance due to feedback. The effective plate resistance would not be calculated as a ratio of the pentode plate resistance, but rather based on transconductance and feedback ratio.

One way is to calculate how much plate current changes if the plate voltage is forced to change by one volt and use Ohm's law to estimate the effective plate resistance.

The total feedback ratio (for the WE120 midband it's derived from (1K/101K) * (triode stage voltage gain -1)) determines how much the grid voltage changes when you change the plate voltage, and the transconductance determines by how much plate current changes in response. One would need to plot the 1st stage load line accounting for the 1K unbypassed Rk etc. to determine the 1st stage gain.

Also for driving 600 ohm load one would want something like <100 ohm output resistance, so calculate the needed Rp effective based on that and the OPT turns ratio^2.

Cheers,

Michael

 

[lassoharp]

The effective plate resistance would not be calculated as a ratio of the pentode plate resistance, but rather based on transconductance and feedback ratio.

That makes sense.  Thanks Michael for the very good breakdown - the handling of the partial unbypassed cathode on V1 had been a slight mystery too.

 

[mjk]

Thanks, though I find I probably oversimplified a few things. It occurs to me that the triode's cathode resistance (Rp||Rl/(stage gain-1)) is in parallel with the 1K for the feedback divider. Also for the gm figure I would get it off the output tube load line rather than use the datasheet number. I may have overlooked something else I would catch in an actual design... :