Transformer Frequency Response

I've looked on the site, and have researched elsewhere but I still cant find anything that really explains what the frequency response for a transformer is.

From what wikipedia states...

Operation of a transformer at its designed voltage but at a higher frequency than intended will lead to reduced magnetizing current; at lower frequency, the magnetizing current will increase. Operation of a transformer at other than its design frequency may require assessment of voltages, losses, and cooling to establish if safe operation is practical. For example, transformers may need to be equipped with "volts per hertz" over-excitation relays to protect the transformer from overvoltage at higher than rated frequency.

Click to expand...

To me this seems more aimed at the utility frequency (50-60hz in the US)
I guess my question more relates to how this relates to frequency response in audio. 

When selecting a transformer for say a balanced line input, does the transformers frequency response have anything to do with the way it will handle the incoming frequencies?

Sorry for the nubblyness and hopefully I put this in the right forum this time...  :-[

reason why im asking:
Im looking for something VERY small (smaller than a carnhill) to turn a balanced signal unbalanced.

Jensen transformer used to publish some good data on their transformers.

http://www.jensen-transformers.com/apps_wp.html

try here

JR

sr1200 said:

From what wikipedia states...

Operation of a transformer at its designed voltage but at a higher frequency than intended will lead to reduced magnetizing current; at lower frequency, the magnetizing current will increase. Operation of a transformer at other than its design frequency may require assessment of voltages, losses, and cooling to establish if safe operation is practical. For example, transformers may need to be equipped with "volts per hertz" over-excitation relays to protect the transformer from overvoltage at higher than rated frequency.

Click to expand...

To me this seems more aimed at the utility frequency (50-60hz in the US)

Click to expand...

That is correct. They are not talking abut audio transformers.

sr1200 said:

When selecting a transformer for say a balanced line input, does the transformers frequency response have anything to do with the way it will handle the incoming frequencies?

Click to expand...

Think of the frequency response as an eq curve. If your transformer (or whatever) is down 3db at 100hz, then anything you send through it will be 3 decibels quieter at 100hz (plus or minus the overall gain/reduction of the transformer). For audio, you are looking for as close as possible to a flat frequency response between 20hz and 20khz. You will see the frequency response listed in different ways, but any transformer you are going to want to buy will have it listed. Take a look at the Jensen specs I've attached: There is a column that says '20Hz/20kHz re:1kHz' this the frequency response with reference to 1khz. For the JT-11P-1, it says -0.04dB/-0.05dB - that means it is down 0.04dB at 20hz and 0.05dB at 20khz referenced from 1khz. That is very flat - So it wont affect the eq of your signal very much, which is the goal. Probably more often, you will see frequency response listed like this: "+/- 1.5db 30-15000hz" - it's the same idea.

sr1200 said:

reason why im asking:
Im looking for something VERY small (smaller than a carnhill) to turn a balanced signal unbalanced.

Click to expand...

There are plenty of transformers smaller than a Carnhill that will be plenty flat, but there is a limit to how small you can go before the specs will drop off. Jensen, Cinemag, Altran, Ed Anderson, Sowter all make transformers that will fit the bill.

Best, Ben

In broad strokes the heavy lifting for transformers is passing low frequency signals cleanly. For mains power transformers a 50 Hz transformer, needs  6/5ths more iron and copper than a 60 hz transformer.

For passing audio, a 20 Hz transformer needs to be 2x the size of a 40 Hz transformer.

JR

Thanks for all the replies!

I think im starting to get what the "theory" is behind these things.

I looked into all the manufacturers listed and wow this could get REALLY expensive.  Gonna see if theres another way around the mountain not using TX's.  Looking to make an 8 or 16 input device which would put me at $320 - $640+ on the low end of things (and still not as small as id like) just for my input stage...

JohnRoberts said:

For passing audio, a 20 Hz transformer needs to be 2x the size of a 40 Hz transformer.

Click to expand...

... or have x1.41 of Pri turns.

Best, M

sr1200 said:

reason why im asking:
Im looking for something VERY small (smaller than a carnhill) to turn a balanced signal unbalanced.

Click to expand...

http://edcorusa.com/products/146-wsm10k-150.aspx

I believe this was the budget transformer NYDave recommended for his reamp schematic.  It's a step down transformer, but it will take a balanced in and output an unbalanced signal.  It's also cheap, and smaller than some transformers although certainly not small.  I've seen some transformers that are 5mmx5mm.  Too bad they weren't for audio...

Transformer frequency response 101:

First you need to know the impedance of your driving source. Let's say is is a microphone with a source impedance of 150 ohms.

The low frequency response limit of the transformer is determined the primary winding inductance of the transformer at the lowest frequency of interest. For a 1dB drop at the lowest frequency you want the impedance of the transformer to be twice the source resistance at that frequency. So for a 150 ohm source and 1dB down at 20Hz we need:

2*pi*20*L  (transformer impedance at 20Hz) = 300 (twice 150) so L = 300/(40*pi) = 2.4 Henries

From the data about the transformer core material you can then work out how many turns of wire the primary will need. From this you can work out the primary resistance (which is one thing that determines losses) and the winding capacitance (which is one thing that determines HF response). In the above example both of these will be quite small.

Let's say we want a 1:10 ratio to drive a tube mic pre grid then we are looking 10 times as many turns for the secondary. The resistive losses will be higher (as the wire is longer) but we can feed into a relatively high impedance so they will be small. The winding capacitance will be a lot higher and since we are feeding into a high impedance this will lead to a tailing off of the HF response.

So in simple terms, LF response is determined principally by primary inductance, and HF response by winding capacitance (leakage inductance also plays a role in HF losses too but its value depends a lot of the physical arrangement of the winding in the transformer so there's no easy way to guess what they will be).

Clearly for a good LF response you need more inductance which usually means more turns and a bigger transformer. For low turns ratios, winding capacitance is not usually a problem but for higher ratios it will limit HF response.

HTH

Cheers

Ian

Millz, thanks man that def. fits the budget more!  Might have to increase the size of the case to accommodate, but at that price, i probably shouldn't complain much.

AS AN EXCERCIZE:
for 10Kohm (line) Looking at the example millz gave...

Ti= 20K(2 * impedance)  L=20K/(40*pi) = 159.24H? (now this is only for 1db down at 20Hz correct?) stat sheet for that unit is less than 1db down so im guessing the numbers would be a little different with rounding.

The stat sheet on the above TX also states a ratio of 8.2:1 so i would guess that the 159.24 is divided by the 8.2 to give around 19.x H which is what the stat sheet says. (19H) (are you supposed to divide by the ratio or did I just get "lucky" math there?)

I think my head is going to essplode! LOL

::EDIT:: and as far as that being a step down, i can just step up on the output stage no?

sr1200 said:

Millz, thanks man that def. fits the budget more!  Might have to increase the size of the case to accommodate, but at that price, i probably shouldn't complain much.

AS AN EXCERCIZE:
for 10Kohm (line) Looking at the example millz gave...

Ti= 20K(2 * impedance)  L=20K/(40*pi) = 159.24H? (now this is only for 1db down at 20Hz correct?) stat sheet for that unit is less than 1db down so im guessing the numbers would be a little different with rounding.

The stat sheet on the above TX also states a ratio of 8.2:1 so i would guess that the 159.24 is divided by the 8.2 to give around 19.x H which is what the stat sheet says. (19H) (are you supposed to divide by the ratio or did I just get "lucky" math there?)

I think my head is going to essplode! LOL

::EDIT:: and as far as that being a step down, i can just step up on the output stage no?

Click to expand...

Glad I could help.

They have other 1:1 transformers that are basically the same price and size.  You'll have to look over the data sheets to make sure they also allow you to go from balanced to unbalanced but they're the same series as the last link I posted, just 1:1 ratios.

http://edcorusa.com/category/13-wsmseries.aspx?turnsratio=11

how long are your wire runs?
why do you need transformer coupling?
a load of budget transformers for short runs in a less than hostile environment (not next to a MW radio transmitter) may be more compromise than solution

reason why im asking:
Im looking for something VERY small (smaller than a carnhill) to turn a balanced signal unbalanced.

Click to expand...

Look at zero-field transformers, e.g. NTP's zft modules or the input of NTP179-160. Useable with very-small line input transformers, even self-made ferrite core stuff should work.

Jakob E.

ruffrecords said:

Clearly for a good LF response you need more inductance which usually means more turns and a bigger transformer. For low turns ratios, winding capacitance is not usually a problem but for higher ratios it will limit HF response.

Click to expand...

Hello Ian,

Just a little clarification, more turns also can be achieved with thinner wire. But of course, again we are getting into increased DCR and copper losses/loading issues.

But then again, bigger transformer is not always necessarily good thing and as always, depends on the design goals and compromises. Leaving aside cost issues, on one hand we have to fight DCR, inductance, and saturation, on another, in audio transformers there is such thing as "to wake up the core" and the bigger the core, the worst low signal resolution.

While LF response is by far a decisive factor, the high end is little easier in a sense it is rather a cost issue of some winding techniques and topologies to extend response.

So in short, all of those are bordering and balancing very fine line and none of audio transformers manufacturers will tell you the whole story until you show some $$$ to get a perfectly matched to your particular application transformer. Isn't it wonderful!!!

Best, M

sr1200 said:

Millz, thanks man that def. fits the budget more!  Might have to increase the size of the case to accommodate, but at that price, i probably shouldn't complain much.

AS AN EXCERCIZE:
for 10Kohm (line) Looking at the example millz gave...

Ti= 20K(2 * impedance)  L=20K/(40*pi) = 159.24H? (now this is only for 1db down at 20Hz correct?) stat sheet for that unit is less than 1db down so im guessing the numbers would be a little different with rounding.

The stat sheet on the above TX also states a ratio of 8.2:1 so i would guess that the 159.24 is divided by the 8.2 to give around 19.x H which is what the stat sheet says. (19H) (are you supposed to divide by the ratio or did I just get "lucky" math there?)

Click to expand...

Remember I said the first thing you need to know is the impedance of the source driving your transformer? A 10K (bridging) transformer is meant to 'look like' 10K to whatever source is driving it - but the driving source impedance is expected to be much less than 10K. Usually 10K bridging transformers were meant bridge a 600 ohm line, the source impedance of which would be much less than 600 ohms.

Looking at the example you cited, the primary winding inductance is 19H (that's the 10K winding) and the response is within a dB at 20Hz. So:

2*pi*20*19 (primary impedance at 20Hz) = 2 * R (minimum source impedance)

so R = 2*pi*20*19/2 = 1193 ohms. Therefore anything with a source impedance below 1K will give no more than a 1dB drop at 20Hz.

Since the primary inductance is 19H and the turns ratio is 8.2:1 and inductance is proportional to the number of turns squared, the secondary inductance should be 19/(8.2*8.2) = 0.28H. If you wanted to use this as a 150 ohm mic input then:

2*pi*20*0.28 = 2 * R (maximum source impedance) so R = 18 ohms so I reckon this will not be so good this way round.

Cheers

Ian

Marik said:

So in short, all of those are bordering and balancing very fine line and none of audio transformers manufacturers will tell you the whole story until you show some $$$ to get a perfectly matched to your particular application transformer. Isn't it wonderful!!!

Click to expand...

I agree with all your points. I just did not want to get into too much detail in a 101 post. Transformer design is something of a black art.

Regarding the $$$, I like working with Sowter (apart from the fact they are local to me) because they will do you a custom transformer for the same price as a regular one. They are not the cheapest but they make very good transformers.

Cheers

Ian

Just to chime the same bell as Ian:
A transformer does not have an intrinsic frequency response.
A xfmr winding that is quoted as 150r down to 20Hz can be used as a 600r at 80Hz.
HF response is a little more debatable, because the combination of leakage inductance and stray capacitance tends to impart a more definte limit to it. But the source impedance and leakage inductance compensation (Zobel network) allow a certain degree of alignment.

Hey,

Thanks for all the input and information everyone.  (this is kinda what i expected college to be like but wasnt... ya know a bunch of informed people debating on issues that the dolts in the class could pick up on until they themselves became educated...)

Anywho.. the runs are very very small...  The plan was to make a "semi-colored" line mixer with panning pot and bypass button 8-12-16 channel (depending on cost)  My idea is kind of based around a neve preamp design with input/output transformers but little else (a few resistors to go to the left/right rails) a balanced insert after the output tranny.