audio transfo circuitry

"Pierre Q."

** B is - by far.

The transformer is being use as the maker intended with 600 ohms source and load.

** Circuit A will not have a fixed 22 Kohms input impedance, but one that varies with frequency and is generally much lower. This will not affect results with a pure voltage source, as shown in the schematic.

The higher than intended load impedance will cause a ( possibly large) high frequency peak and consequent ringing.

....... Phil

Both circuits will attenuate the signal by 37dB, but the transformer can be much smaller in the second circuit. If you want just 22k input impedance the you do not need any resistor in there, but connect the transformer directly to the amp and the input directly to the transformer. This way your signal doesn't get attenuated.

.---o----. | | o------. ,-----' .-. Input )|( | |22k )|( | |inside amp .--' '-----. '-'already | | | === '---o----' GND1 GND2 (created by AACircuit v1.28 beta 10/06/04

--
ciao Ban
Apricale, Italy

Actually, they don't. Only the second one has an input impedance of about 22k. The first one has the inductance of the primary essentially in parallel with the 22k secondary load. That is going to approach 600 ohms at 300 Hz.

That means that they are fairly good approximations of an ideal transformer (the load impedance reflects to the primary side according to the turns ratio squared) if the load is about 600 ohms. An ideal transformer transforms power with no loss. In the first case, the transformer would draw a magnetizing current that is insignificant compared to the current drawn by a 600 ohm load impedance transformed back to the primary. But the transformer is not loaded with 600 ohms, but 22k ohms. The magnetizing current will be similarly insignificant compared to this higher load impedance, only at a frequency that is

22600/600 times higher, or about 11,300 Hz. In other words, when the transformer is loaded with some impedance other than 600 ohms, its flat response frequency range shifts.

Better in what sense? Flatter frequency response? Higher input amplitude capability before having its core saturating at the low frequency end? Efficiency is not much of a factor, with that large voltage divider ratio.

The 600 ohm spec is not a requirement, just the load impedance that produces the specified frequency response. With a 600 ohm load, the primary looks close to a 600 ohm impedance over this range of frequency. Below 300 Hz, the transformer primary inductance starts to draw more current than the transformed secondary load does, so the primary impedance falls. If the signal source is a low impedance , as it is with the top version, this isn't so bad for frequency response, except that the primary winding resistive drop increases and you risk core saturation.

Above 3000 Hz, the leakage inductance starts to drop a significant fraction of the signal voltage, so the input impedance rises and so does the output impedance.

I like the second version better, because it passes a smaller signal through the transformer, so the transformer handles only the output power, not the power into the divider. And the transformer is loaded with its specified impedance, so it should have a flat response over the expected frequency range.

If this isolator actually drives a 600 ohm load, you would eliminate the output resistor in the second case, and raise the resistor to 600 ohms in the first case, so the actual load was part of the divider.

In fact, I need a 22K (not critical though) input resistance and also want the signal to be attenuated by roughly 40dB. Reading from other replies, it seems figure B is better in term of frequency response.

Pier

Indeed what I need most is the flatest frequency response that transfo can give so it seems from your reply (and others...) that it does have to be loaded by 600 ohm I/Os. The circuit drives a microphone inout which seems to be several kilohms so I may have to leave the 600 ohm res at the output.

Pier

Pierre, I would use the circuit in Figure B. The effects of winding capacitance will be much less, since the capacitance is shunted by the

600 Ohm resistor across the primary.

I quickly build a prototype of both circuits A and B. The sound with A is so so, may be some ringing and/or weird frequncy response as John stated. With circuit A, sound is ok.

Regards! Pierre

Better than any opinion.

Maybe ;) but opinions are always appreciated. Many times people have come out with a different view or perspective of a design problem that I had never thought of.

Regards!

Pierre