Convert Push-Pull Output to SE with Passive Components

[unclejed613]

As such, you might be fine with just using one output with respect to gnd

with some amplifiers, that could trigger built-in protection circuits, because DC current to ground will be interpreted as DC offset. if you want to tap off a signal from one side of the output, capacitive coupling is required. you will only get half the power, but it will be ground referenced. to get full output power, a transformer will work. the transformer can be connected across the bridged outputs, and one end of the secondary is grounded.

 

[unclejed613]

For reasons of experimentation, I was asking here for a way to modify that to a non-reversing current using passive external components. IOW to shift the bipolar signal entirely above ground potential.

ok, well that's at least a little bit clearer... couple the output through a capacitor, and either add DC bias to the signal, or use an added output pair that runs from a single-ended supply. here's an example of what i mean:

it's basically the same output stage as what you have in your original amplifier, and acts as a unity gain buffer, and a level shifter at the same time... depending on the frequency of your signal, you might want to adjust the value of the capacitor. any output stage configuration will work here, the one shown (copypasta from the schematic you posted) is a Sziklai pair output stage, but a darlington output stage will work just as well.

 

[pnielsen]

OK. Picture this. You have two parallel wires. An identical 100Hz sinewave is fed into each, but from opposite ends so their respective currents flow in opposite directions. If that sinewave swings between positive and ground, the "peaks" of opposition (cancelled EMF) occur at 100Hz. On the other hand, if the sinewave swings between plus and minus, the current flow changes direction once each cycle resulting in a doubling of the opposed (cancelled) EMF, or a rate of 200Hz, which is what I do not want.

So to retain the 100Hz component of the original frequency of opposition when using an amp with an output that swings between positive negative, I need to shift the DC level of the output to be wholey between positive and ground.

My initial understanding was that this could be done by tapping one side of the split seconday of an output transformer (with CT to ground), introducing an offset DC voltage to the output signal, or ...?

I was simply asking for suggestions on the best way to do this. What I am trying to do is explained as best I can in the first part of this post. Simply to prevent current reversal in the output of a domestic stereo amplifier, and thus make it unipolar.

 

[unclejed613]

it's still not really clear what you are trying to accomplish. if you are trying to minimize magnetic coupling from the +/-VCC rails (to the input stage for instance, which could generate a small amount of even harmonic distortion), then the best method for reducing magnetic coupling would be to use twisted pairs of wire (each VCC rail and ground) to go from the power supply to the amplifier. this is the method used most often in commercial amplifiers. there is no "current reversal" happening in the output stage, except in the output wire to the speaker itself. the current never reverses in the power supply wires. you will have current in the positive side of the supply during a little more than half the waveform, and current flowing in the negative rail during a little more than half the waveform. hypothetically, there should be very little current at all in the ground wire to the amplifier, and the speaker grounds should actually come from a "star" ground in the power supply.

if you are trying to convert a bipolar (-50V to +50v swing for instance) into a single polarity output (0V at the negative peak, and +100V at the positive peak), then look at the circuit i posted, it will do the level shifting, and maintain the same current capability (it will need to have a positive supply equal to the sum of the bipolar supplies i.e. +100v if the amplifier supplies are +/-50V, and have the same current capability)... just keep in mind that a resistive load is going to dissipate a lot of heat when the amplifier is idling, because 0V at the input of the stereo amp, will idle the output of the level shifter at +50V. having no input (even if you biased the audio input to run the stereo amp at -50V, and bypassed all the protection circuits and offset decoupling) will still give you a +50V idle at the output because of the capacitive coupling going into the level shifter. do not use the level shifter to drive a speaker, it will roast the voice coil of most speakers almost instantaneously, if it doesn't launch the voice coil into orbit first.

 

[pnielsen]

Thanks to all the helpful replies, I think I now have enough options to work with. Tomorrow I will be able to spend time at the bench to test the results .

I will first try the transformer with grounded secondary that unclejed613 described earlier. That seems to be the most straightforward approach. I assume by "ground" is meant the signal input ground.

There is no practical application of what I want to do. It is just an experiment. Sorry if that led to confusion.

 

[ChrisP58]

OK. Picture this. You have two parallel wires. An identical 100Hz sinewave is fed into each, but from opposite ends so their respective currents flow in opposite directions. If that sinewave swings between positive and ground, the "peaks" of opposition (cancelled EMF) occur at 100Hz. On the other hand, if the sinewave swings between plus and minus, the current flow changes direction once each cycle resulting in a doubling of the opposed (cancelled) EMF, or a rate of 200Hz, which is what I do not want.
-
-

Actually, without a load at the end of the wires, no current will flow.
But let's assume that there is a load. In the scenario you give, the magnitude of the magnetic field will double if the sinewaves are in phase, or cancel each other if they are out of phase. But in neither case will the frequency of the magnetic field double.





(*) Note: This argument ignores the speed of light and any other parasitic propagation delays along the length of the wire.

 

[pnielsen]

Yes, that touches upon what I was referring to. Hard to pin down in words since it is generally not spoken of in public.

Again think of two in-phase bipolar currents admitted to opposite ends of two parallel wires. During one half cycle, the current of each is flowing one way. During the second half, it is flowing the other. The points in time (frequency) during which the magnitude of EMF cancellation is at its most intense thus occurs twice each cycle. Once at peak positive voltage swing, and once again at "peak" negative.

Please correct me if I am wrong.

 

[unclejed613]

so what actually happens here with two parallel wires from the power supplies, is on the positive wire, electrons flow from the load (the speaker) through the output transistors, into the power supply during the positive half of the cycle, during the negative half, electrons flow the opposite direction, but in the negative wire. the net magnetic field around the pair of wires is the same frequency as the current through the load. if you are thinking the other power supply wire picks up a signal through magnetic interaction, this is possible, but is "eaten" by the large power supply capacitors if it has any effect at all. the coupling between the supply rails is quite small, and essentially the filter caps absorb it in any case.

if you are trying to reduce the effects of crosstalk between the power supply rails and the input stage of the amplifier, this is a layout problem, not a circuit topology problem. keep the circuits from input stage to voltage amplifier stage (the voltage amplifier is a high impedance node) out of close proximity to the high current portions of the supply rails, keep the power supply wires to the output stage as short as possible. adding some ground plane around sensitive portions of the amp circuits (if you happen to be designing a board), and possibly routing the supply wires through a steel partition between the amp and power supply can be useful in minimizing magnetic coupling of power supply currents to the input stage and VAS.

if you want more specifics, find Douglas Self's Audio Power Amplifier Design Handbook. he has at least a whole chapter on proper layout of power supply and power wiring. i recommend you get the latest edition of it, but if you are on a tight budget, you can find an older copy of it here: https://archive.org/download/fe_Aud...sistor_Amplifiers_Design_Handbook_D._Self.pdf


it's the third edition, and the latest edition is the sixth edition, with a lot of newer information (50% larger than the 5th edition).

the third edition still has a lot of good information, especially about what your question seems to be. i once worked for a company that made excellent amplifiers, and as part of the final test procedure, we used a BIC pen body with notches cut in it to move the power supply wires around, then hot glued them in place. the procedure was done with a distortion analyzer, and the goal was to null out the magnetic coupling between the power supply wires and the input stage. the procedure worked, but i'm willing to guess that changing the routing of the wires might have made that time consuming procedure unnecessary. the procedure was done at 20khz (the magnetic effect is more pronounced at 20khz than any lower frequency, and has a 6db/octave slope)