I'm trying to clearly understand how a power supply will become a sagging one vs being stiff.
Would a stiff PS require a high power transformer that constantly provides a lot of current? In contrast, would a PS with a lesser transformer but beefy filter capacitors be able to deliver lot's of current for a few mSec but then, as the capacitors drain during prolonged current draw, start to sag?
In my question, I'm not referring to regulated supplies, just a simple transformer, rectifier, filter cap scenario.
Are my thoughts correct?
Would a stiff PS require a high power transformer that constantly provides a lot of current? In contrast, would a PS with a lesser transformer but beefy filter capacitors be able to deliver lot's of current for a few mSec but then, as the capacitors drain during prolonged current draw, start to sag?
In my question, I'm not referring to regulated supplies, just a simple transformer, rectifier, filter cap scenario.
Are my thoughts correct?
From my understanding..
Amplifying a signal isn't a static affair. Transients in volume and frequency mean sudden demands on power. If the power supply can't provide the current needed the voltages drop.
If you look at a mains power supply - it has AC rectified, that means current is available only at certain points in the AC cycle. The remaining parts of the cycle need capacitance to provide current whilst the power supply can't provide it.
Tube rectifiers also need a voltage drop in the order 100V and they can't provide a lot of current due to their internal resistance. This is why a valve rectifier only has a few uF of capacitance behind it. However that's not a problem to normal tube amps as they don't draw much power but use high voltage instead. The resulting speakers need to be efficient to use the voltage swing rather than draw current.
A solid state rectifier has a drop in the order of a couple of volts and can pass masses of current. Thus you can put more capacitance reserve in to keep the rail voltages high.
Sonically this sag causes changes in the harmonics, and 'smooths' the transient a little compared to the 'harder' solid state.
If you under spec the transformer, then you have a limited amount of current delivered after which the the current will drop and so will the voltage.
Large caps can also act as a low pass filter - a 5 ohm resistor will not cause much of a voltage drop but requires a 6,000uF cap after it to make a 0.5Hz low pass filter to help smooth and reduce noise.
Class A has a a high idle current then has the wave form - take more power (positive part of the sign wave) and then take less power (negative part of the sign wave). However ALL of the devices will be conducting so that power draw is continuous.
Class AB will be overlapping but each device will have some switch off time in the sine wave where the device will not be using power. This results in less power used, thus this could be 50% (classB) to 99% depending the amount of the wave overlapped in the design. Typically 60-75% power saving.. which means if planned you can have a smaller power supply than you'd need at 100%. Smaller caps too. If the signal is small then often you'll be running class A as the signal will be running in the overlap but the amplitude will be small. The downside of AB and B is that the power supply seems more transient behaviour as the devices switch on/off. The output can also see varying impedance changes too.
The problem comes when, as you right have said, the power demand simply outstrips the currently available power.
There are some other aspects of 'sag' that involve the design - parallel caps for example reduce ESR, heating but also increase the area to deliver higher current. If a single capacitor has issues providing enough power fast enough it would have the same effect but parallel caps would mean more current. However that charge would need replenishing.
Amplifying a signal isn't a static affair. Transients in volume and frequency mean sudden demands on power. If the power supply can't provide the current needed the voltages drop.
If you look at a mains power supply - it has AC rectified, that means current is available only at certain points in the AC cycle. The remaining parts of the cycle need capacitance to provide current whilst the power supply can't provide it.
Tube rectifiers also need a voltage drop in the order 100V and they can't provide a lot of current due to their internal resistance. This is why a valve rectifier only has a few uF of capacitance behind it. However that's not a problem to normal tube amps as they don't draw much power but use high voltage instead. The resulting speakers need to be efficient to use the voltage swing rather than draw current.
A solid state rectifier has a drop in the order of a couple of volts and can pass masses of current. Thus you can put more capacitance reserve in to keep the rail voltages high.
Sonically this sag causes changes in the harmonics, and 'smooths' the transient a little compared to the 'harder' solid state.
If you under spec the transformer, then you have a limited amount of current delivered after which the the current will drop and so will the voltage.
Large caps can also act as a low pass filter - a 5 ohm resistor will not cause much of a voltage drop but requires a 6,000uF cap after it to make a 0.5Hz low pass filter to help smooth and reduce noise.
Class A has a a high idle current then has the wave form - take more power (positive part of the sign wave) and then take less power (negative part of the sign wave). However ALL of the devices will be conducting so that power draw is continuous.
Class AB will be overlapping but each device will have some switch off time in the sine wave where the device will not be using power. This results in less power used, thus this could be 50% (classB) to 99% depending the amount of the wave overlapped in the design. Typically 60-75% power saving.. which means if planned you can have a smaller power supply than you'd need at 100%. Smaller caps too. If the signal is small then often you'll be running class A as the signal will be running in the overlap but the amplitude will be small. The downside of AB and B is that the power supply seems more transient behaviour as the devices switch on/off. The output can also see varying impedance changes too.
The problem comes when, as you right have said, the power demand simply outstrips the currently available power.
There are some other aspects of 'sag' that involve the design - parallel caps for example reduce ESR, heating but also increase the area to deliver higher current. If a single capacitor has issues providing enough power fast enough it would have the same effect but parallel caps would mean more current. However that charge would need replenishing.
Yes, I forgot to mention I was referring to a solid state class AB power amplifier supply.
I understand class A amplifiers are a bit different as they draw all their current at all times, right?
The fact that the capacitors are essential in a transformer, solid state rectifier, filter capacitor power supply is true both in a sagging and stiff power supply.
As I see it, the capacitors are a temporary reservoir, there to smoothen the rectified voltage. A larger capacitor would be able to reduce ripple and also handle transients better. But it would be drained in a transient or two, unless the transformer is able to deliver the current needed to replenish the charge.
So, a stiff supply would have a transformer sized to cover the peaks and not just the average current consumption with adequate capacitors while a sagging one would employ a smaller transformer, with capacitors not really making much difference?
I understand class A amplifiers are a bit different as they draw all their current at all times, right?
The fact that the capacitors are essential in a transformer, solid state rectifier, filter capacitor power supply is true both in a sagging and stiff power supply.
As I see it, the capacitors are a temporary reservoir, there to smoothen the rectified voltage. A larger capacitor would be able to reduce ripple and also handle transients better. But it would be drained in a transient or two, unless the transformer is able to deliver the current needed to replenish the charge.
So, a stiff supply would have a transformer sized to cover the peaks and not just the average current consumption with adequate capacitors while a sagging one would employ a smaller transformer, with capacitors not really making much difference?
Correct. A reserve cap tends to smooth the power working to smooth the ripple. As the current load increases and/or the frequency decreases between the pulses the more capacitance you will need to maintain the same ripple voltage peak-to-peak.
Caps can also be used to filter noise but the one after the rectifier typically is focused on ripple. Ie a for a full wave (ie both parts of the cycle) then that would be 100Hz for a 50Hz supply. For a half wave (ie one part of the cycle) then that would be 50Hz for a 50Hz, where there's a full 1/2 missing that the cap needs to cope with.
In the end a capacitor is only temporary storage. In fact the capacitors leak current so you loose some power. A way of looking at it is:
Enough power per cycle to cover transients and maximum volume = stiff and it would not sag.
If you drop the transformer power but increase capacitance, there will be a point where there will be a short fall of current and it will sag.
Personally I prefer enough power, with more than enough capacitance. Given that all your amp is doing is modulating the power source. However at some point "enough power" results in an exponential cost, complex designs and a considerable amount of heat.
Transformers have an output impedance, usually expressed as the number of amps the secondary is capable of supplying steady state. So when loaded, the tranny with the higher current capability is going to sag less - and that's going to happen, obviously, even without the rectifier / capacitor part.
Add that section and, because of the funny non-sinusoidal current waveform the transformer sees, I imagine its sag is even worse than when the secondary is directly loaded by a resistor for a given VA. So when loaded with a full wave bridge / capacitor, you'll need even bigger VA capability from the tranny for the same power load, to prevent sag.
I seem to remember something called power factor regarding capacitive loaded bridge rectifiers and that being 0.7. I'll guess then, that for a certain power load on the DC side, you'll need a tranny with ~1.5X that for secondary VA rating, to prevent sag.
I'm sure I'll be corrected shortly if my thinking isnt right -
Add that section and, because of the funny non-sinusoidal current waveform the transformer sees, I imagine its sag is even worse than when the secondary is directly loaded by a resistor for a given VA. So when loaded with a full wave bridge / capacitor, you'll need even bigger VA capability from the tranny for the same power load, to prevent sag.
I seem to remember something called power factor regarding capacitive loaded bridge rectifiers and that being 0.7. I'll guess then, that for a certain power load on the DC side, you'll need a tranny with ~1.5X that for secondary VA rating, to prevent sag.
I'm sure I'll be corrected shortly if my thinking isnt right -
I did some experiments, some time ago, with a typical class AB amp with an undersized transformer and minimal supply capacity. Nothing too scientific, just a series of tests over a week or so. I came away with the subjective feeling that extra capacitors could not make up for an inadequate transformer. It was also apparent that the listening experience was much preferred with a larger transformer.
Why was this? If the psrr of the amp is good and it's not clipping you should notice no difference apart from slightly cooler output devices and a slightly warmer transformer in the case of the smaller supply.
A 'stiff' supply normally means one with low impedance. However when caps are the means of stiffening the supply it needs a very large capacitance to get stiffness at the lowest frequencies. Take a typical 10,000uF res cap, it has a reactance of ~0.8ohm at 20Hz - a supply with an impedance this high can scarcely be considered stiff.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.