![[-]](https://theultimatetone.com/images/collapse.png "[-]")
06-05-2025, 01:01 PM
Hi Guys
Most modern solid-state power amps use symmetric supply voltages, commonly referred to as "split rails". This just means there is, say, +/-50Vdc supporting the circuit. In absolute terms each rail is the same voltage, only the polarity is different.
The supply voltage has to be sufficient to accommodate the peak signal amplitude into the speaker load at full load. So, the +/-50V will accommodate a 40Vpk signal into 8-ohms, corresponding to 200Wpk or 100Wrms. There is 10V extra, which accommodates losses over the transistors and current-sense resistors (emitter resistors). If the supply stays solid at 50V at full load, there should a few more volts possible into the load as we do not want to rate the power of the amp right at its limit lest it may not reach that value under low-mains or other conditions. On any case, the supply rails need to be in proportion to the power output desired.
Most PA circuits live within the same supply voltage as the output stage requires. However, there are some designs where the front-end circuitry requires a higher voltage, and these higher rails can be created in various ways. The two broad methods are supply stacking and having independent sources.
Supply stacking is exactly as it sounds: a second supply is stacked on top of the main supply. Where a typical PSU has a center-tapped winding, a full bridge over the whole winding and filter caps and bleeder resistors from each of the bridge outputs to the CT, the stacked supply adds another small supply on top of each rail. This smaller supply is often a small OPT with dual secondaries, where each secondary has a full bridge across it, followed by a cap and bleeder resistor. These small supplies are identical and one has its negative end tied to the positive main rail, while the second small supply has its positive end tied to the negative main rail. We end up with V+HI, V+, CT (0), V-, V-HI. Note that the V+/-HI are relatively low-current and would not be suitable in a class-G or class-H amplifier.
To size the small PT, we look at the current draw of the front-end circuitry. The typical differential input stage draws constant current. The transimpedance stage may have a constant current source for one side, but the other side may draw more current under load. Predrivers or driver stages draw modest currents but can sink and source much higher currents, so these should be assessed in simulation or in a real circuit. How much current the small supply must support depends on the circuit load, and we most often have drivers and predrivers supported by the main supply, alleviating some of the variable loading on the small supply.
We also consider the nominal voltage that the front-end circuit operates from. Because it sits atop the main supply, which will sag under load, we have to account for that sag plus any sag that may occur due to low-mains to both PTs. This generally means that the small supply voltage might be twice the difference we need between the idling main rail and the idling front-end rail. For example, a mosfet amp might have +/-65V main rails with +/-75V for the front-end. We intuitively view this as needing 10V boost supplies. If that is all we add, then as the 65V sags, so too will the 75V, and this may not be good for the stability of the quiescent point for the output stage, or possibly for the overall circuit stability.
If we double the boosted supply to 20V and add a ground-referenced regulator to provide 75V, then we accommodate up to 10V of sag for the front-end. The reality is that we need a bit more sacrificial voltage here.
As a reference,the inherent regulation of typical toroidal PTs is quite good, varying with the VA rating:
80VA 10%
120VA 9%
160VA 7.7%
225VA 7.2%
300VA 7.1%
500VA 5.1%
625VA 4.1%
Some front-end circuits incorporate active current sources while others use resistive ones. For the latter, a regulated supply maintains bias stability, but means the amp might be unstable while powering up. As mentioned, the regulator must be ground-referenced and it is simplest to use a discreet circuit for this. For active CS circuits, we may still wish to regulate the voltage, or at least have an active hum filter for each boosted rail.
The PT for the boosted supplies may be quite small depending on the circuit, maybe 10-15VA, but could be up to 50VA for a large amp or when significant drive currents must be supported by the boosted rail. Each boost supply is derived from one secondary, so whatever the current and voltage product is for one boost supply, we must double it for the PT VA rating.
The small PT VA goes up proportionately if we use it to create fully independent high rails that are ground-referenced. The circuit is identical to the main supply, just a smaller PT with higher voltages.
The boost supply PT can be a small toroid, semi-toroid or EI type. EMI from this PT should be low as the load is balanced over the secondaries and fairly constant.
Have fun
Most modern solid-state power amps use symmetric supply voltages, commonly referred to as "split rails". This just means there is, say, +/-50Vdc supporting the circuit. In absolute terms each rail is the same voltage, only the polarity is different.
The supply voltage has to be sufficient to accommodate the peak signal amplitude into the speaker load at full load. So, the +/-50V will accommodate a 40Vpk signal into 8-ohms, corresponding to 200Wpk or 100Wrms. There is 10V extra, which accommodates losses over the transistors and current-sense resistors (emitter resistors). If the supply stays solid at 50V at full load, there should a few more volts possible into the load as we do not want to rate the power of the amp right at its limit lest it may not reach that value under low-mains or other conditions. On any case, the supply rails need to be in proportion to the power output desired.
Most PA circuits live within the same supply voltage as the output stage requires. However, there are some designs where the front-end circuitry requires a higher voltage, and these higher rails can be created in various ways. The two broad methods are supply stacking and having independent sources.
Supply stacking is exactly as it sounds: a second supply is stacked on top of the main supply. Where a typical PSU has a center-tapped winding, a full bridge over the whole winding and filter caps and bleeder resistors from each of the bridge outputs to the CT, the stacked supply adds another small supply on top of each rail. This smaller supply is often a small OPT with dual secondaries, where each secondary has a full bridge across it, followed by a cap and bleeder resistor. These small supplies are identical and one has its negative end tied to the positive main rail, while the second small supply has its positive end tied to the negative main rail. We end up with V+HI, V+, CT (0), V-, V-HI. Note that the V+/-HI are relatively low-current and would not be suitable in a class-G or class-H amplifier.
To size the small PT, we look at the current draw of the front-end circuitry. The typical differential input stage draws constant current. The transimpedance stage may have a constant current source for one side, but the other side may draw more current under load. Predrivers or driver stages draw modest currents but can sink and source much higher currents, so these should be assessed in simulation or in a real circuit. How much current the small supply must support depends on the circuit load, and we most often have drivers and predrivers supported by the main supply, alleviating some of the variable loading on the small supply.
We also consider the nominal voltage that the front-end circuit operates from. Because it sits atop the main supply, which will sag under load, we have to account for that sag plus any sag that may occur due to low-mains to both PTs. This generally means that the small supply voltage might be twice the difference we need between the idling main rail and the idling front-end rail. For example, a mosfet amp might have +/-65V main rails with +/-75V for the front-end. We intuitively view this as needing 10V boost supplies. If that is all we add, then as the 65V sags, so too will the 75V, and this may not be good for the stability of the quiescent point for the output stage, or possibly for the overall circuit stability.
If we double the boosted supply to 20V and add a ground-referenced regulator to provide 75V, then we accommodate up to 10V of sag for the front-end. The reality is that we need a bit more sacrificial voltage here.
As a reference,the inherent regulation of typical toroidal PTs is quite good, varying with the VA rating:
80VA 10%
120VA 9%
160VA 7.7%
225VA 7.2%
300VA 7.1%
500VA 5.1%
625VA 4.1%
Some front-end circuits incorporate active current sources while others use resistive ones. For the latter, a regulated supply maintains bias stability, but means the amp might be unstable while powering up. As mentioned, the regulator must be ground-referenced and it is simplest to use a discreet circuit for this. For active CS circuits, we may still wish to regulate the voltage, or at least have an active hum filter for each boosted rail.
The PT for the boosted supplies may be quite small depending on the circuit, maybe 10-15VA, but could be up to 50VA for a large amp or when significant drive currents must be supported by the boosted rail. Each boost supply is derived from one secondary, so whatever the current and voltage product is for one boost supply, we must double it for the PT VA rating.
The small PT VA goes up proportionately if we use it to create fully independent high rails that are ground-referenced. The circuit is identical to the main supply, just a smaller PT with higher voltages.
The boost supply PT can be a small toroid, semi-toroid or EI type. EMI from this PT should be low as the load is balanced over the secondaries and fairly constant.
Have fun
