Question about isolation transformers in R.G.'s Hum-Free ABY splitter box

[kraal]

Hi,

In his article "Hum-Free(er) A/B/Y Splitter Box with Transformer Isolation" [1], R.G. uses 42TM018 transformers.
After having had a look at the datasheet for this transformer I see that it has the following specifications.

• frequency range: 300 Hz to 3.4 kHz
• Maximum output:  200mW

My questions are the following:

• Can somebody explain why this frequency range is sufficient ? (so far I was thinking that the usual range for a guitar is 80-1200 Hz (without harmonics, and up to 6k and above with them)
• Can somebody explain how the required "maximum output" has to be calculated ?

Thank you in advance for helping me understand :-)

[1] http://www.geofex.com/FX_images/TransformerSplitter.pdf

[Eb7+9]

maybe it was designed to hard-pan a couple of Pignose amps ...
in a strictly DC-only environment

[JerS]

They work fine in spite of the rating. I have used them for isolation applications on guitar and there is no noticeable drop in low end.

[PRR]

Transformer frequency response depends on how they are driven and loaded. "Matched" telephone applications may have limited power. R.G. dedicates a relatively beefy chip to this job. For typical parasitic impedances the bass may go 5 to 20 times lower than the "matched" case. So 60Hz or better.

Also the trannies are really only sold to do a job. In telco use that job is 300-3.5kHz. But it would actually be hard to wind a core that small to work that bad. In fact the treble runs well past 15KHz, typically 30KHz.

[MikeA]

+1 on the driving and loading answers above.  I used these, driven by op amps at unity gain into a 1M load, and they were flat beyond 10 kHz on the high side.  I didn't make a note about the low side but they were fine for a guitar signal (<80 Hz.)  MikeA

[kraal]

Hello,

Thank you all for your replies and sorry for the late reply (I was sick during the last few days).

The frequency range part of my question is now clear(er), however I still don't get why a transformer with a 200mW rating is sufficient.
In another thread people reported having used successfully a transformer rated at 150mW. But apart from "it works" I found no answer explaining why it works (with figures)
In other words how small can the transformer be?

Michel

[R.G.]

It's a question of the iron core. A given quantity of iron can only hold so much magnetic field energy before it saturates and stops supporting more transformer action.

There's a lot of background on this issue that I'm skipping to get to this but the key parameter is the volt-time integral, literally the amount of voltage times seconds that your signal puts across the primary. That's what "loads up" the field into the iron. And this makes the low frequency response issues dominate how small a transformer can be. In this instance you're looking for "good enough" response down and past the 82hz of a stock-tuned guitar, or the ... um... 78Hz of a drop-D tuning. For bass, you tend to need to get an octave lower.

The low frequency response sets the "time" part of the volt-time integral. The time is the time for one half-cycle. The voltage is what you can then play with to get a smaller core. For smaller voltage signals, you can use a smaller core to get the same frequency response.

It's a many-faceted issue. 200mW is sufficient because we don't usually in low signal audio look for maximum power transfer. We want maximum voltage transfer. The power a 1V guitar signal can push into a 1M ohm input resistor is 1 MICRO-watt, and that's fine if we get the frequency response we want. A one milliwatt transformer would do the job if we could get it to do the frequency response, and have power left over. Power's not an issue. The smallest practical transformers have plenty of power rating.

The impedance ratio and stated frequency response get into in subtler ways. If a transformer is rated at a low frequency of 300hz and an impedance ratio of, say, 10k to 10K, that tells us a whole lot about its insides. A transformer moves voltage and current across a magnetic field coupler and voltage isolation barrier. We pay for that isolation and conversion by having to feed the core with magnetic field. It's subtle, but that can be condensed into having to drive the primary inductance with the incoming signal voltage.

The primary inductance of a transformer is what sets the low frequency response. A transformer with a 300hz low end and a 10K impedance limit has a primary inductance that eats half of the incoming current at that frequency and impedance. The watts rating tells you how big the incoming voltage can be (that volt-time integral thing) and not saturate the iron, so you can actually calculate how much current the primary inductance has to be fed to pay for the transformer operation. That's one point of the article on geofex. If you drive a transformer primary that's rated at 300Hz and say, 1K ohms, but you drive it with a burly driver that can keep pouring in more current as the frequency gets lower than 300hz, the signal voltage at the secondary can be prevented from falling off to a much lower frequency. In my example, using an opamp to drive the primary, I extended the "300hz" rated trannie down to 60hz, a 5:1 improvement. This eventually runs into the volt-time integral problem, but it's a useful dodge where all you're interested in is the voltage transformation, which fortunately we are.

None of this is affected by the high end spec on the transformer. That is set purely by the wire positions in winding the primary and secondary coils and their physical layout. As Paul said, it's hard to wind a small transformer that won't get over 10kHz. My tested sample was flat to 20kHz and peaked, then rolled off at 22kHz. You mostly get the audio bandwidth for free >>>IF and ONLY IF<<< all you're looking for is voltage transformation. Telecom transformers (these are) are rated for specified impedances and they do in general have to transfer some amount of power (historically if nothing else) into nominally 600 ohm telephone lines. That's a different world.

[kraal]

It's a question of the iron core. A given quantity of iron can only hold so much magnetic field energy before it saturates and stops supporting more transformer action.

There's a lot of background on this issue that I'm skipping to get to this but the key parameter is the volt-time integral, literally the amount of voltage times seconds that your signal puts across the primary. That's what "loads up" the field into the iron. And this makes the low frequency response issues dominate how small a transformer can be. In this instance you're looking for "good enough" response down and past the 82hz of a stock-tuned guitar, or the ... um... 78Hz of a drop-D tuning. For bass, you tend to need to get an octave lower.

The low frequency response sets the "time" part of the volt-time integral. The time is the time for one half-cycle. The voltage is what you can then play with to get a smaller core. For smaller voltage signals, you can use a smaller core to get the same frequency response.

It's a many-faceted issue. 200mW is sufficient because we don't usually in low signal audio look for maximum power transfer. We want maximum voltage transfer. The power a 1V guitar signal can push into a 1M ohm input resistor is 1 MICRO-watt, and that's fine if we get the frequency response we want. A one milliwatt transformer would do the job if we could get it to do the frequency response, and have power left over. Power's not an issue. The smallest practical transformers have plenty of power rating.

The impedance ratio and stated frequency response get into in subtler ways. If a transformer is rated at a low frequency of 300hz and an impedance ratio of, say, 10k to 10K, that tells us a whole lot about its insides. A transformer moves voltage and current across a magnetic field coupler and voltage isolation barrier. We pay for that isolation and conversion by having to feed the core with magnetic field. It's subtle, but that can be condensed into having to drive the primary inductance with the incoming signal voltage.

The primary inductance of a transformer is what sets the low frequency response. A transformer with a 300hz low end and a 10K impedance limit has a primary inductance that eats half of the incoming current at that frequency and impedance. The watts rating tells you how big the incoming voltage can be (that volt-time integral thing) and not saturate the iron, so you can actually calculate how much current the primary inductance has to be fed to pay for the transformer operation. That's one point of the article on geofex. If you drive a transformer primary that's rated at 300Hz and say, 1K ohms, but you drive it with a burly driver that can keep pouring in more current as the frequency gets lower than 300hz, the signal voltage at the secondary can be prevented from falling off to a much lower frequency. In my example, using an opamp to drive the primary, I extended the "300hz" rated trannie down to 60hz, a 5:1 improvement. This eventually runs into the volt-time integral problem, but it's a useful dodge where all you're interested in is the voltage transformation, which fortunately we are.

None of this is affected by the high end spec on the transformer. That is set purely by the wire positions in winding the primary and secondary coils and their physical layout. As Paul said, it's hard to wind a small transformer that won't get over 10kHz. My tested sample was flat to 20kHz and peaked, then rolled off at 22kHz. You mostly get the audio bandwidth for free >>>IF and ONLY IF<<< all you're looking for is voltage transformation. Telecom transformers (these are) are rated for specified impedances and they do in general have to transfer some amount of power (historically if nothing else) into nominally 600 ohm telephone lines. That's a different world.

Thank you a lot for this detailed and very interesting answer R.G. ! I'l re-read it a few times to make sure I understand correctly every part of it. I'll also have a look at the "volt-time integral".
Maybe one more question: I could not find the article on geofx where the circuit is explained, I only fount the "image" of the schematics, and a pdf explaining how the circuit was modified. Can somebody point me to this article ?

Best regards,