measuring transformer frequency

[mrphotodude]

I'm looking for tips, tools or links to learn how to measure the frequency response of transformers for audio output.
Test equip is not an issue as i have access to a huge lot of high end stuff, i just don't know how to use alot of it or where to start.
-ken

 

[abbey road d enfer]

I'm looking for tips, tools or links to learn how to measure the frequency response of transformers for audio output.
Test equip is not an issue as i have access to a huge lot of high end stuff, i just don't know how to use alot of it or where to start.
-ken

Measuring the response of a xfmr is just the same as any other transmission element: inject a signal at the input and graph the output.
So basically you need an oscillator and an AC voltmeter.
BUT: xfmrs are very sensitive to the impedance of the source that drives them and somewhat sensitive to the load they are connected to.
Roughly, the higher the source impedance, the worse the LF response, the higher the load capacitance, the worse the HF response.

 

[JohnRoberts]

Transformers also have a LF signal handling constraint related to size, i.e. they will distort sooner the lower the audio frequency being passed.  This is not a frequency response spec as much as  frequency handling spec.

JR

 

[mrphotodude]

do spectrum analyzers generally put out a frequency and graph items or do i need to take  a sig gen and let it sweep between say 20hz-20khz and plot it?

 

[abbey road d enfer]

Most spectrum analysers (either filter-based or FFT-based) do not have proper selectivity/resolution for accurate measurement of frequency limits.
You need a good ole oscillator, a meter, pen and paper if you don't have an AP system. Assuming you do audio BW assessment, extend your measurements down to 10Hz and up to 60kHz.

 

[mrphotodude]

the tests i have been trying are composed of running a sig gen through the primary then the secondary into a vtvm set on 0 dbm then changing the sig from around 10hz to around 30khz in steps to see where the freq rises and falls. it seems most of the transformers i have go around 25hz to 18-22 khz within +/- 2dbm. a few only reach to 15khz before dropping 5dbm plus.
i was trying to find other ways to do this but i guess i had it down.
what other things should i look for to determine if the transformer is good to use? other then impedance i mean, i can calculate that.
i have boxes of iron and am just trying to determine what to try in what project.
Does it really come down to bench it and listen?

 

[CJ]

i like to use a dual trace scope with digital voltage readout,

this way you can look at the input and output at the same time,

this will let you keep the input voltage constant,

show you when your generator craps out at the low end,

and will also show you distortion and phase shift,

newer scopes will have digital readouts for phase angles also,

then put a 1ook resistor in series with a poly cap, around .1 to 1 uf,

hook it across the sec, put the scope leads on the resistor,

put the scope into x-y mode and drive the heck out of it with a square wave at 100 hz and lower to see the BH curve which will tell you what kind of core is in there, alloys, sq or round anneal,

 

[abbey road d enfer]

the tests i have been trying are composed of running a sig gen through the primary then the secondary into a vtvm set on 0 dbm

This is the 21st century. dBm is to be avoided. dBu is correct, it's a voltage (dBm is a power into 600r, inherited from the telephone system). 0dBu is too high for testing mic input xfmrs. You should run them at -20 dBu for FR assessment. Then you may want to increase the level and note how it distorts and how it truncates the LF response.

then changing the sig from around 10hz to around 30khz in steps to see where the freq rises and falls. it seems most of the transformers i have go around 25hz to 18-22 khz within +/- 2dbm

Once again, this is not the correct unit. A FR tolerance is expressed in plain dB's , not dBm or dBu.

a few only reach to 15khz before dropping 5dbm plus.

You may have noted that there is often a peak before the drop. If not, that's probably because the output impedance of your generator is too low. Mic iput xfmr's want to see 150-200r.

i was trying to find other ways to do this but i guess i had it down.

I wouldn't say so. You have to refine your procedure.

what other things should i look for to determine if the transformer is good to use? other then impedance i mean, i can calculate that.

I'm not sure you can. Nothing in your procedure allows proper assessment of the impedance the xfmr is optimized for.

Does it really come down to bench it and listen?

That is the thing you will do after proper objective evaluation; you may end up with a bunch of suitable candidates, that you will have to evaluate in situ, i.e. putting them at work and assessing the performance of the working unit.

 

[lassoharp]

Does it really come down to bench it and listen?

This will help get to the heart of the matter as to whether you like it's overall sound fairly quickly.


Your bridging pad for the test set up will give a fairly constant source Z.  A mic may stray from it's quoted nominal source Z.  It would probably be useful to see how closely the sine wave measurements correlate with your in circuit sonic impressions regarding top, bottom, and midrange response. 

 

[CJ]

you can just use volts  for the transformer response graph,

that way you don't have to worry about dbu, dbm, db, dbx, dby,dbz, ddt, fbi or the cia.

if somebody from the anal retentive dept need the 3 db corner freqs, just have them convert the graph themselves.

 

[abbey road d enfer]

you can just use volts  for the transformer response graph,

Except that volts are too much for a mic input primary; a fraction of volt is preferrable - about 1/10th. 

if somebody from the anal retentive dept need the 3 db corner freqs, just have them convert the graph themselves.

The LF -3dB corner frequency is paramount in assessing the nominal impedance the xfmr is supposed to be. One only needs to know that -3dB is 0.7 times.

 

[audiomixer]

you can just use volts  for the transformer response graph,

that way you don't have to worry about dbu, dbm, db, dbx, dby,dbz, ddt, fbi or the cia.

if somebody from the anal retentive dept need the 3 db corner freqs, just have them convert the graph themselves.

I don't really agree as I think that getting the concept of decibels is a good thing to have in out audio world and knowing the difference between -20 dBu and +20 dBu when feeding a mic / line transformer can make a real difference.
defining the frequency response plus 0.32V minus 0.226V re 0.774 volts is a little lame, no?
we all use the term dB often in right and in wrong context, but to get practice with dB is a good thing. now there are a number of (free) tools for that, and (free) software that can help, specially once calibrated to a reference voltage.... other incule online db converters and calculators (I tend to use these to once in a while). I like sengpiel: http://www.sengpielaudio.com/calculator-db-volt.htm

now on the subject of transformers I would start with Arta www.artalabs.hr and a good audio interface, best with calibrated input level. calibrated could mean that when u feed it with a known voltage, like a sine 1kHz at 0dBu through a Neutrik Minirator your software displays the correct level.... drive the transformer with adequate output impedance (comparable to your expected source impedance, add series resistor to the primary side). measure two channels - transformer input and transformer output. the input should not really change with low impedance sources (line, most microphones, your audio interface), but hey, you can still discard the curve later.... Arta can also measure distortion. Note: changing the preamp gain kills your calibration. calibrate at the max expected level and give it 6 dB (FS....) headroom. measuring the audio path (output connected to input) can help to check for (stupid) mistakes.

I hope this helps,

- Michael

 

[CJ]

if you want to compare transformers, then the units don't matter, as long as you keep them the same,

db is a pain in the ass to plot, you have to get the volts off the scope readout then use the calculator,

plus, db will flatten out the self res curve so you will not see how dramatic it can be with certain transformers,

call me lazy but after you sweep 500 tranformers, you will be using volts, believe me,

of course you have to watch the insertion level,

some transformers have the max level on the label, som not,

just get a good signal with no distortion,

most mic inputs will take 1 volt p-p, the core is usually way bigger than needed,

beyer transformers might be the exception, they use those micro cores,

but a utc a-10 will take a lot of input level before it distorts,

same with triad and peerless,

 

[CJ]

forgot about an easy way to plot it to db,

just use log graph paper, doh on me! 

but even that can be a pain ,