back to transformers again

[tempus]

Hey all;

I still haven't got my pedalboard built, or even ordered the parts yet, but I'm getting there. Anyway, I was reading this thread:
http://www.diystompboxes.com/smfforum/index.php?topic=40603.msg292277#msg292277

and I have a couple of questions.

I have a piezo in my electric guitar, and would like to get some ground loop isolation happening. The piezo goes through a buffer 1st, and then to the PA. I'd like to insert a transformer in there to get rid of the resulting ground loop hum. I tried a transformer from a phone, and found that it attenuated the bass very slightly (much less than I expected, in fact). I was thinking about springing for a Jensen or Sowter tranny when I found the above thread and its link to Edcor. For $10 I can get 20 -20 frequency response. I've read RG's posts on this subject, and it's starting to sink in, but I'm still unsure of some things:
1. Is the buffer the reason RGs ABY splitter gives a frequency response down to 60Hz? Is there anything special about RGs buffer (looks pretty straightforward to me but just in case)?
2. I plan to use a JFET buffer mounted in my guitar. How important is the output Z to getting that 60Hz response?
3. Which tranny should I use? From the above post, it looks like a 600:600 should be OK, but I just want to make sure I'm getting this straight. Would this http://www.edcorusa.com/Products/ShowProduct.aspx?ID=301 be suitable? Or will the telephone tranny I've already got give me the same sound?

Thanks

[R.G.]

1. Is the buffer the reason RGs ABY splitter gives a frequency response down to 60Hz?

Yes.

Is there anything special about RGs buffer (looks pretty straightforward to me but just in case)?

Only that it has a very low output impedance from the opamp.

2. I plan to use a JFET buffer mounted in my guitar. How important is the output Z to getting that 60Hz response?

Quite important.

What is happening is that the buffer drives the primary of the transformer. There at the transformer, the signal drives the reflected secondary load and the primary inductance in parallel. The secondary load is high, on the order of a meg or so for most amps attached to the secondary. So what the buffer is really doing is driving the primary inductance. This may be in the 10-100 ohm range at 60Hz, so the amp must have an output impedance which can drive that with low losses. A JFET may not be able to provide enough current. It's worth trying. A MOSFET SRPP mu-amp made from BS170s and using the low-output-impedance mod I show at GEO might be able to do it, but the parts count would be less with an opamp.

3. Which tranny should I use? From the above post, it looks like a 600:600 should be OK,

A 600:600 transformer is not intended for use in this kind of setup. A 600:600 telephone transformer is especially not designed for this kind of use.

Would this http://www.edcorusa.com/Products/ShowProduct.aspx?ID=301 be suitable?

Yes, when driven from a buffer with the ability to drive 600 ohms. I don't know if a single JFET can. Even some opamps would have problems. The NE5532 can do it, as can the LM833. The nice thing about the edcore is that it's already specified at 20-20K response, so it's not being operated outside its range down at 60Hz. All it needs is a buffer.

Or will the telephone tranny I've already got give me the same sound?

I don't think so.

[tempus]

Thanks again for your response RG. I think I'm getting the hang of the transformer thing now. So this transformer http://www.edcorusa.com/Products/ShowProduct.aspx?ID=306would be more suitable since an opamp or a JFET buffer (which would likely have higher output Z) would be able to drive the primary at 10k? On a more basic level, this is basically an Ohm's law thing I think. If I=V/R, then a larger source resistance (i.e. from the buffer output) would reduce the current available to drive the primary, thus limiting the low end frequency response since because of the inductance of the coil more current is needed to push those low frequencies through. Is that right? And even though it's already spec'd at 20-20, it still needs a buffer because the extremely high output Z of a piezo wouldn't be able to drive the 10K impedance of the primary? Now suppose I wanted to send my output to a balanced mic input at the PA. Would this one http://www.edcorusa.com/Products/ShowProduct.aspx?ID=308 be suitable for that task?

Thanks again

[R.G.]

Thanks again for your response RG. I think I'm getting the hang of the transformer thing now. So this transformer http://www.edcorusa.com/Products/ShowProduct.aspx?ID=306would be more suitable since an opamp or a JFET buffer (which would likely have higher output Z) would be able to drive the primary at 10k?

Yes. The nice thing about that one is that it requires no tricks, or bass limitations. Most opamps will be able to drive 10K to full power supply swing with no trouble at all, and no tricks are necessary to extend the bass. So I'd expect full hifi bandwidth out of that one when driven by an opamp on the front end. $6 .36 is a nice price for that if the transformer meets spec.

On a more basic level, this is basically an Ohm's law thing I think. If I=V/R, then a larger source resistance (i.e. from the buffer output) would reduce the current available to drive the primary, thus limiting the low end frequency response since because of the inductance of the coil more current is needed to push those low frequencies through. Is that right?

Yes. Now you ARE beginning to understand transformers.

And even though it's already spec'd at 20-20, it still needs a buffer because the extremely high output Z of a piezo wouldn't be able to drive the 10K impedance of the primary?

On target again.

Now suppose I wanted to send my output to a balanced mic input at the PA. Would this one http://www.edcorusa.com/Products/ShowProduct.aspx?ID=308 be suitable for that task?

Yes. Note that it will also step down the signal voltage with the impedance. This one will step down by SQRT(15k/600) = SQRT(25) = 5 so the output signal will be 1/5 the input signal voltage level. This is a loss of 20*log₁₀(5) = 14db, which will have to be made up at the PA.

[tempus]

Thanks again RG.

This one will step down by SQRT(15k/600) = SQRT(25) = 5 so the output signal will be 1/5 the input signal voltage level. This is a loss of 20*log10(5) = 14db, which will have to be made up at the PA.

Wouldn't it need to be attenuated even more if it was going into a mic input? And that would be a balanced signal too right? Also, if I were to use the 10k/10k tranny, and just use that as a line out to the PA, what wold be the output impedance going to the PA input? I'm thinking if it's 10K that it might be too high for a line input (which is usually 47K I think).

Thanks again

[R.G.]

Wouldn't it need to be attenuated even more if it was going into a mic input?

Maybe/probably. Most PA input s have a wide range of gains available on mic inputs. Or they did when I last messed with that.

And that would be a balanced signal too right?

Yes. I realize now that I didn't look for whether the secondary of the 600 was center tapped. I should have.

Also, if I were to use the 10k/10k tranny, and just use that as a line out to the PA, what wold be the output impedance going to the PA input? I'm thinking if it's 10K that it might be too high for a line input (which is usually 47K I think).

This brings up a subtle point - transformers don't have impedances, they have ratios. The impedance numbers only tell you the impedances where the frequency specifications hold true.

For a 10K:10K transformer, if you load the secondary with a 10K resistor, the primary presents a 10K load in parallel with the primary inductance (and some other parasitics) to the driving source. If you change that secondary 10K load resistor to a 1K, the transformer reflects that into the primary as a 1K load. A 10K:10K transformer is more properly thought of as a 1:1 transformer which meets its low frequency spec when loaded with 10K on the secondary and driven from much less than that.

In the case of a 10K:10K driving a PA input. the source impedance driving that input would be the driving impedance; if this is an opamp, it's the feedback-lowered impedance of the opamp, plus stray stuff like the resistances of the primary and secondary winding wires, stuff like that. If the transformer output is, say 20 ohms, and the wiring resistances are 100 ohms each for primary and secondary. The opamp 20 plus the wire 100 are reflected at 1:1 to the secondary, becoming 120, plus the secondary wire resistance of 100 ohms, total of 220 ohms source impedance at the output to the PA. The opamp driving the transformer sees the 47K of the PA input plus 100 ohms of resistance on the secondary (47100 ohms) reflected to the primary at 1:1 (still 47100) plus the 100 of the primary, or 47200 ohms in parallel with the primary inductance for a load.

Low frequency response is almost entirely dominated by the driving source impedance's ability to drive the primary inductance and the reflected secondary load; usually this reduces to the ability to drive the primary inductance. The high frequency response gets complicated by winding capacitances, leakage inductances, loading, etc. Most signal transformers are much better at high frequencies than the specs show, at least in terms of usable signal. The flatness of the response gets bad due to resonances and such that are affected by matched loads damping the transformer.  I would put a small cap and a 10K resistor on the output of the transformer in question to ensure that any ringing gets damped.

[tempus]

OK thanks again RG.

I would put a small cap and a 10K resistor on the output of the transformer in question to ensure that any ringing gets damped.

Is the value of the resistor here chosen based on the transformer's impedance (i.e., 10K)? If I were to use a 10K:150 ohm transformer would the value of the resistor change?

Thanks

[R.G.]

Is the value of the resistor here chosen based on the transformer's impedance (i.e., 10K)? If I were to use a 10K:150 ohm transformer would the value of the resistor change?

This is a trick for smoothing the high end response. The low end is happy with mismatches because the dominant impedance is the primary inductance. At the high end, the primary inductance forms so little of the load that it can be ignored. But the parasitics start being significant. In this case, the leakage inductances and the self-capacitances of the windings make a resonant circuit that produces bumps and notches in the response. These cannot be gotten rid of short of rewinding the transformer differently, but they can be damped so they don't get out of hand. In this case, it is usually a good guess to provide the specified secondary impedance load for the damping resistance. Here is one of the few places where maximum power transfer does matter in audio - you want the maximum power transfer out of the resonances.

So yes, if you were using a 10K:150 to drive a load of more than perhaps 600 ohms, use a 150R resistor and change the cap to make the resistor be the load in the upper audio region. For a cheapo specified transformer (i.e. 300-3K passband) I'd want to put that in the range of a few KHz. For a full-audio specified transformer, I'd start with the rolloff being above audio and only bring it into the audio band based on testing for flatness.

[tempus]

So would a suitable value for the cap be .0047uF? I used the equation f_c=1/2*pi*RC. I'm not sure that this is the correct equation though, or if the inductance of the transformer should be taken into account.

Thanks

[R.G.]

So would a suitable value for the cap be .0047uF? I used the equation f_c=1/2*pi*RC. I'm not sure that this is the correct equation though, or if the inductance of the transformer should be taken into account.

The inductance of the transformer is long since passed, at least the primary inductance. It's a high inductance, and by the upper end of the audio spectrum, it's so high an impedance as to be neglected. The leakage inductances matter, but you're trying to use this trick to be under the frequency where they start hurting you. In this middle region, what matters to calculating the rolloff of the R-C load is the R, the C, and the driving impedance as transformed to the secondary. If you're driving the primary of a 10K:150 transformer with a 10K source impedance, that becomes a 150R impedance on the secondary. However, the leakage inductance is rising, and that appears in series with the transformer no matter where you look at it. So the idea is to add a 150 ohms resistive load above some frequency, and you can simply calculate the capacitor with the 150 ohm resistor. If you use 4.7nF and 150 ohms, the turnover frequency will be F = 1/(2*pi*R*C) = 226kHz. I would move that down by at least 5:1 or about 22nF to put the rollover at about 45kHz. This ensures the effect is outside audio. Putting it down at 22kHz with a 47nF cap would be OK in most instances.

This is a simplification, useful for quick calculation. The actual values of the leakage inductance and winding self capacitance get into it if you're after calculating everything in closed form. But since those must generally be measured, not calculated, doing it the simplified way is generally good enough, at least until you measure and find out whether there's a problem or not.

[doitle]

I thought this might be a good place to ask my own transformer question. I have a transformer I pulled out of an old Cobra Clock Radio Telephone unit. I measured the three... sections of it and it seems to go:
Primary: 120 Ohms.
Secondary 1: 12 Ohms
Secondary 2: 50 Ohms

So does this mean my transformer would transform 120V AC RMS to 12V AC RMS and 50V AC RMS? If so this might be a neat accompanying transformer for a higher headroom valvecaster...

[R.G.]

I thought this might be a good place to ask my own transformer question. I have a transformer I pulled out of an old Cobra Clock Radio Telephone unit. I measured the three... sections of it and it seems to go:
Primary: 120 Ohms.
Secondary 1: 12 Ohms
Secondary 2: 50 Ohms

So does this mean my transformer would transform 120V AC RMS to 12V AC RMS and 50V AC RMS? If so this might be a neat accompanying transformer for a higher headroom valvecaster...

Unfortunately, no. The wire resistance (which is most likely what you measured) has almost nothing to do with the voltages or impedances involved in the transformer. Wire resistance is a parasitic, to be minimized or designed down to where it can be ignored, in transformer design.

A lot depends on what the transformer used to do, whether it was a power transformer or an audio transformer.

[doitle]

Drat. It was a power transformer. Looking at the PCB where I disconnected it I can see a bridge rectifier and I'm guessing the transformer did NOT put out 50V as I originally thought. The power caps are only 16V. a 1000uF and a 220uF.

[tempus]

OK, finally getting around to setting this thing up. Is this the proper way to wire the transformer for a balanced output to a mic input? I'm unsure about the ringing cancellation circuit and where the grounds are supposed to be connected to avoid ground loops.



Thanks