transformer for isolation - what's going on here?

[tempus]

Hey all;

I'm thinking about incorporating a ground loop isolation circuit in my new pedalboard. I have a piezo and magnetic pickups in my guitar, and since one will go to my amp and the other to the PA, there will definitely be a ground loop issue. So I pulled a transformer out of an old phone yesterday. One side of it measured 250 ohms, the other had 2 taps measuring 200 and 400. I ran my piezo signal through it, and aside from losing a little on the bottom end (this is listening through a PA) it sounded pretty much the same as without the transformer (aside from a total absence of ground loop hum).

I don't understand this. First off, isn't the input Z of the transformer only 200 ohms? I ran the signal through a buffer first, but even still, I would have thought that loading would be an issue. And what about the high frequency content? Surely this lowly telephone transformer can't have a range beyond a few kHz. But I listened specifically for this, and the high end was fine. What gives?

Thanks

[Paul Marossy]

I don't know about the specs on your transformer, but I can tell you that telephone transformers are not designed for frequencies below a certain range. They are typically designed for the 300/400Hz to 3.4Khz range. The bass frequency response is limited because of the limitations in telephone system components: transformers and capacitors can be smaller if they don't have to deal with lowest frequencies. Other reason to drop out the lowest frequencies is to keep the possibly strong mains frequency (50 or 60 Hz and it's harmonics) humming away from the audio signal you will hear.

So, assuming that you have one designed for 400Hz-3.4Khz, when you play anything below an open A it's going to lose it's strength. You might be able to do a workaround by using an EQ after the transformer, and a buffer before it also helps.

Here is something I designed a few years ago that uses a Mouser 42TM018 transformer with a buffer circuit. It seemed to work pretty well with a piezo equipped guitar.




EDIT: You might also want to check out this post - http://www.diystompboxes.com/smfforum/index.php?topic=40603.0

[R.G.]

I'm thinking about incorporating a ground loop isolation circuit in my new pedalboard. I have a piezo and magnetic pickups in my guitar, and since one will go to my amp and the other to the PA, there will definitely be a ground loop issue. So I pulled a transformer out of an old phone yesterday. One side of it measured 250 ohms, the other had 2 taps measuring 200 and 400. I ran my piezo signal through it, and aside from losing a little on the bottom end (this is listening through a PA) it sounded pretty much the same as without the transformer (aside from a total absence of ground loop hum).

I don't understand this. First off, isn't the input Z of the transformer only 200 ohms? I ran the signal through a buffer first, but even still, I would have thought that loading would be an issue. And what about the high frequency content? Surely this lowly telephone transformer can't have a range beyond a few kHz. But I listened specifically for this, and the high end was fine. What gives?

- the resistances of the primary and secondary windings have very little to do with the transformer's voltage/current/impedance ratio; the resistance only tells you the resistance. In particular, it tells you nothing about the input impedance of the transformer, because the input impedance consists of the primary inductance in parallel with the reflected secondary load.
- telephone transformers are designed around 600 ohm impedances; That is, they expect the phone line side to be a 600 ohm impedance driving them.
- have you read the article on hum-free splitters at geofex.com?

Your description matching what I would expect - drive a 600 ohm transformer with a pickup into a PA input and you'll lose bass because of the loading of the primary inductance. You'll probably lose a lot of treble from a standard pickup as well.

[tempus]

drive a 600 ohm transformer with a pickup into a PA input and you'll lose bass because of the loading of the primary inductance

That's the thing that's puzzling me though. There is a barely noticeable loss of bass - so little that I had to A/B it a bunch of times, and I'm still not sure there's actually any difference.

RG you suggest a transformer from Mouser for your ABY splitter. Is it likely that what I pulled out of the phone is similar to what you were suggesting? I know it's hard to guess, but do you think it's worth trying one from Mouser, or is there not likely to be any difference?

Thanks

[Paul Marossy]

You might not be noticing much loss in the bass frequencies because of the low voltages/currents involved. See this quote from the post I linked above:

Why are transformer specs given in ohms AND ratio?  I've seen 600:600 and 10K:10K used almost interchangably in isolation applications, ostensibly because they're both 1:1.  What's the rule of thumb in deciding which is the more important figure, ohms or ratio, or (God forbid) both?

Good question.

Transformers don't have ohms. Transformers have ratios and limits. When you see ohms, there are unstated specifications about the transformer you are expected to know.

Some of these are related to the limits. A transformer may be specified as 10K:10K and (to pluck a number out of the air) 100mW. That is, the transformer is being represented to you as delivering one tenth of a watt into a 10K load resistor attached to the secondary. You're expected to know or figure out the relevant parts of the following:

100mW into 10K is computed by ohm's law. Power = I * V, and we know that V = I *R, so we can pervert those equations into P = V^2/R (that being how I write "V squared divided by R). So we can also figure that V = SQRT(P*R) = SQRT(0.1*10,000) or 31.6V. In this case, the volts are RMS, so that 100mW transformer will transform a bunch of volts. There's probably some fudge factor there, too.

There's not much current in that. 31.6V into 10K ohms is only 3.16ma.

A transformer specified at 600:600 ohms 100mW is going to be able to do an output of V = 7.74V maximum and 12.9ma by the same calculations.

So far so good. But back to the limits. The transformer is not only limited in power delivery, it's limited in frequency as well. Most audio transformers are assumed to be voice/telephone grade unless you pay extra, and the implication of that is that they'll do 300Hz to 3kHz. So back at our 10K transformer, they will only have put enough windings on the primary to keep the core from saturating with 31.6V at 300Hz, and enough to keep the inductance over 10K at 300Hz, or a minimum of about 5.3H. They may have put more in there, but that's all you could reasonably count on in the absence of a datasheet. The leakage could be quite bad, but most small transformers have much better treble response than they have to.  The 600:600 transformer can have lower inductance because it has much lower volt-seconds impressed on its primary.

Note that the 100mW 10K to 10K and the 600 to 600 are probably the same sized core. All that changes is the number of turns and the wire gauge. The power handling is a property of the core and window area, not the impedance level (in a broad generalities way).

So when they tell you 10K:10K 100mW audio, they're really saying "31.6V to 31.6V unloaded, up to 3.16ma secondary current, at frequencies between 300Hz and 3kHz". Only they tell you the square of the turns ratio (that is, the impedance transformation ratio) times the expected load impedance they designed for.

Did that help?