So I'm considering building a 2 case integrated amplifier, where one case will house the input selection,attenuator, buffer (dcb1) and power amplifier (f5) and the second case will have the toroidal transformers and digital circuitry (lcduino) that controls the input selection and attenuator.
Regarding the f5 power supply I am considering 3 options:
1- Have the rectifiers and CRC board near the toroidal and have Clean DC feed via umbilical cord.
2- have the rectifiers and CrC board in the amplifier case, feeding ac via de umbilical cord.
3- have the rectifiers and 2 capacitors near the toroidal and the main CRC filtering board near the amplifier boards. This looks like the best solution as the umbilical cord carries rectified dc current and the CRC board close to the amp boards guarantees low impedance and resistance (correct?). or should the main filtering section be near the trafo and put the smaller caps near the amp boards.
What do you think of these options?
Regarding the f5 power supply I am considering 3 options:
1- Have the rectifiers and CRC board near the toroidal and have Clean DC feed via umbilical cord.
2- have the rectifiers and CrC board in the amplifier case, feeding ac via de umbilical cord.
3- have the rectifiers and 2 capacitors near the toroidal and the main CRC filtering board near the amplifier boards. This looks like the best solution as the umbilical cord carries rectified dc current and the CRC board close to the amp boards guarantees low impedance and resistance (correct?). or should the main filtering section be near the trafo and put the smaller caps near the amp boards.
What do you think of these options?
I would probably choose something like option 3.
Think in terms of current, more than voltage. The signal that makes the sound at the speakers is current that comes directly from the reservoir and decoupling capacitors (and occasionally from the rectifiers). These are large dynamic current waveforms with the exact shape of the music waveform. See the image at: http://www.diyaudio.com/forums/power-supplies/216409-power-supply-resevoir-size-38.html#post3117390
Whenever there is a change in the current's amplitude (which could typically be very, very often), it will try to change the capacitors' voltage. Also, the inductance (and the resistance) of the conductors (and the capacitors) will make possibly-large voltage disturbances on the power rails, since V = L di/dt. But most amplifiers do better when the rail voltage stays almost constant. So it's much better to have the shortest-possible distance between the capacitors and the power output devices. That's why there should also be decoupling capacitors right at the output devices. That's also why using many smaller capacitors in parallel is better than one large one, and why using power and ground planes is better than using a wire; low inductance and low resistance.
If you get as many capacitors as possible in parallel, as close to the power output devices as possible, then you wouldn't have to worry too much about how far away the main reservoir capacitors might be.
But there is still the problem of star grounding. There is no "clean DC" on an umbilical. Remember the current in the umbilical will be as dynamic and large as the music signal to the speakers. Those currents, both outgoing and incoming, WILL induce significant time-varying voltages across the umbilical's wires, giving rail-voltage fluctuations and ground bounce, at the amplifier end of the wires. Also, you would definitely need to use separate ground-return and ground reference wires, to be able to even hope that the ground reference would not contain a significant time-varying voltage component. And you would have to tightly twist the power and power ground wires together, and maybe also shield them, so they wouldn't radiate too much.
So again, keeping everything as short and close together as possible seems like the best way.
Think in terms of current, more than voltage. The signal that makes the sound at the speakers is current that comes directly from the reservoir and decoupling capacitors (and occasionally from the rectifiers). These are large dynamic current waveforms with the exact shape of the music waveform. See the image at: http://www.diyaudio.com/forums/power-supplies/216409-power-supply-resevoir-size-38.html#post3117390
Whenever there is a change in the current's amplitude (which could typically be very, very often), it will try to change the capacitors' voltage. Also, the inductance (and the resistance) of the conductors (and the capacitors) will make possibly-large voltage disturbances on the power rails, since V = L di/dt. But most amplifiers do better when the rail voltage stays almost constant. So it's much better to have the shortest-possible distance between the capacitors and the power output devices. That's why there should also be decoupling capacitors right at the output devices. That's also why using many smaller capacitors in parallel is better than one large one, and why using power and ground planes is better than using a wire; low inductance and low resistance.
If you get as many capacitors as possible in parallel, as close to the power output devices as possible, then you wouldn't have to worry too much about how far away the main reservoir capacitors might be.
But there is still the problem of star grounding. There is no "clean DC" on an umbilical. Remember the current in the umbilical will be as dynamic and large as the music signal to the speakers. Those currents, both outgoing and incoming, WILL induce significant time-varying voltages across the umbilical's wires, giving rail-voltage fluctuations and ground bounce, at the amplifier end of the wires. Also, you would definitely need to use separate ground-return and ground reference wires, to be able to even hope that the ground reference would not contain a significant time-varying voltage component. And you would have to tightly twist the power and power ground wires together, and maybe also shield them, so they wouldn't radiate too much.
So again, keeping everything as short and close together as possible seems like the best way.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.